Source: https://akholi.com/outsourcing-2026-agentic-ai-tier-1-banks/study-data/part-2-c170/
Author: Phil Hatch
Publisher: Akholi

Rate how reliably the tools’ built-in guardrails and safety controls held during production use

Study Data · Part 2 · Indicator Deep-Dive Analysis

# Rate how reliably the tools' built-in guardrails and safety controls held during production use

Indicator C170 · By Phil Hatch · Akholi · October 9, 2026

← Study Data← Previous indicatorNext indicator →

## On this page

27 sectionsIndicator SpecificationSummary of FindingsDistribution of RatingsCoverage and Origin of Written ResponsesDistribution of Written SentimentCorrespondence Between Ratings and Written SentimentRating Profile of Respondents Providing Written ResponsesDescriptive Statistics by SegmentTests of Difference Between SegmentsWritten Sentiment by SegmentAssociation with Cancellation RateAssociation with Customer SatisfactionAssociation with the Engagement Indicator MeanAssociation with Intent to LeaveAssociation of Written Sentiment with the Study MeasuresComparison of Associations Across the Four MeasuresClustering and Effective Sample SizeEquivalence Tests of Null ResultsRobustness ChecksMeasure Independence and Part-Whole CorrectionComparison with the Study NormComparison of Written Sentiment with Other Part 2 IndicatorsCo-Occurring Indicators and ThemesInterpretation of the DataLimitationsDefinitionsSources and Provenance

## Indicator Specification

Table 1. Indicator specificationTable

| Field | Value |
| --- | --- |
| Indicator code | C170 |
| Question as asked | Rate how reliably the tools' built-in guardrails and safety controls held during production use. |
| Failure category | AI technology |
| Place in the survey | Challenge Deep Dive > 5.4 AI > Security and Control |
| Indicator class | Respondent level |
| Indicator type | Rated, POS |
| Response scale | 1 to 10, higher is the better outcome |
| Answering base | 787 of 850 respondents, across 150 of 150 engagements |
| Non-answers | I do not know 0, I prefer to not answer 0, blank 63 |
| Written responses addressing the indicator | 17 responses from 17 respondents on 14 engagements (15 primary, 2 secondary) |

## Summary of Findings

Table 2. Key results: association of the indicator with the four study measuresTable

| Measure | Form | n | r | 95% CI | p |
| --- | --- | --- | --- | --- | --- |
| Cancellation rate | Descriptive contrast | 150 engagements | −0.66 | −0.74 to −0.56 | < 0.001 |
| Customer satisfaction | Clean estimate | 94 engagements | 0.65 | 0.52 to 0.76 | < 0.001 |
| Engagement indicator mean | Leave-one-out | 150 engagements | 0.86 | 0.82 to 0.90 | < 0.001 |
| Intent to leave | Respondent level, clustered | 742 respondents | −0.66 | −0.71 to −0.60 | < 0.001 |

The mean rating was 4.7 and the median 5.0 (n = 787 respondents). Ratings of 3 or below, the distressed range, made up 32.02% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.66, 95% CI −0.74 to −0.56, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.65, 95% CI 0.52 to 0.76, p < 0.001, n = 94 engagements), a higher engagement indicator mean computed without the indicator (r = 0.86, 95% CI 0.82 to 0.90, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.66, cluster bootstrap 95% CI −0.71 to −0.60, p < 0.001 with standard errors clustered by engagement, n = 742 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 51.61, Holm p < 0.001, omega squared 0.424) and vendor type (Welch F 4.64, Holm p = 0.003, omega squared 0.024).

Of the 17 written responses, 10 were positive and 7 negative, the reverse of the balance in most Part 2 indicators. Most positive responses said the guardrails held and placed the respondent's concern elsewhere. The negative responses described guardrails that let through data exposure, prohibited wording or actions on accounts that should have had a second review. No written outcome test reached significance.

## Distribution of Ratings

Table 3. Distribution of ratings, respondent levelTable

| Answer | Respondents | Share of answered (base 787) | Cumulative share | Engagements with at least one | Share of 150 engagements | Distressed |
| --- | --- | --- | --- | --- | --- | --- |
| 1 | 62 | 7.88% | 7.88% | 46 | 30.67% | Yes |
| 2 | 69 | 8.77% | 16.65% | 47 | 31.33% | Yes |
| 3 | 121 | 15.37% | 32.02% | 81 | 54.00% | Yes |
| 4 | 130 | 16.52% | 48.54% | 82 | 54.67% | No |
| 5 | 126 | 16.01% | 64.55% | 84 | 56.00% | No |
| 6 | 115 | 14.61% | 79.16% | 75 | 50.00% | No |
| 7 | 83 | 10.55% | 89.71% | 51 | 34.00% | No |
| 8 | 54 | 6.86% | 96.57% | 30 | 20.00% | No |
| 9 | 22 | 2.80% | 99.36% | 16 | 10.67% | No |
| 10 | 5 | 0.64% | 100.00% | 4 | 2.67% | No |
| I do not know | 0 | Not in base | Not in base | 0 | 0.00% | Not applicable |
| I prefer to not answer | 0 | Not in base | Not in base | 0 | 0.00% | Not applicable |
| Answered, total | 787 | 100.00% | | 150 | 100.00% | |

Table 4. Distribution of ratings, engagement levelTable

| Engagement level summary | Engagements | Mean of engagement means | SD | Min | Median | Max | Engagements with a distressed mean | Share of engagements distressed |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Engagement means, goodness view | 150 | 4.6 | 1.8 | 1.5 | 4.3 | 8.8 | 32 | 21.33% |

Ratings of 3 or below, the distressed range, accounted for 252 of the 787 answers (32.02%). The engagement mean fell in the distressed range in 32 of 150 engagements.

## Coverage and Origin of Written Responses

Figure 1. Written responses addressing the indicator, by question answered and written score (n = 17 responses)ExhibitNote: Each bar is one survey question. C177 is the written-answer question of the indicator's survey section, so its responses are primary. The responses to every other question are secondary. Segments wide enough to hold a label carry their count. C177 asked: “Add any comments on the agentic AI tools that the ratings in this section do not capture.” C030 asked: “What are the top 3 things that you love most about your job?”Table 5. Written responses by originTable

| Origin | Responses | Respondents | Engagements | Negative | Strongest negative (−10) | Positive | Mean written score |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Primary | 15 | 15 | 13 | 7 of 15 (46.67%) | 2 of 15 (13.33%) | 8 of 15 (53.33%) | 0.7 |
| Secondary | 2 | 2 | 1 | 0 of 2 | 0 of 2 | 2 of 2 | ‡ |
| All responses | 17 | 17 | 14 | 7 of 17 (41.18%) | 2 of 17 (11.76%) | 10 of 17 (58.82%) | 1.2 |

Note: ‡ Base under 10. No statistic is reported.

The primary responses numbered 15, all written in answer to C177, “Add any comments on the agentic AI tools that the ratings in this section do not capture.” The remaining 2 responses, 11.76% of the total, were secondary and came from 1 other questions. Respondents who wrote both a primary and a secondary response numbered 0. Negative responses made up 46.67% of primary responses (7 of 15) and 0 of the 2 secondary responses.

Eight responses to C177 said the guardrails or safety controls held, one over a whole quarter. One respondent described a sandbox that caught an agent reaching for a settlement system, and another said the controls held except twice. These respondents placed their concerns in memory, consistency between runs, long multi-step jobs and the absence of an audit trail. C030 asked respondents what they loved most about their job, and its two responses listed guardrails that hold among the conditions they valued.

Six responses described a guardrail that let something through. An employee's home address went into a public ticket note, and the respondent said the guardrails caught nothing and the team caught it. Real investor account numbers went into a test file although the guardrail was meant to stop it. On hardship cases the guardrails let through wording the team was not allowed to use with a debtor. One agent pulled statements from a second custodian account because an instruction was loose, and another would action a customer cash account without a second review. One respondent said the safety controls had never been built for the team's kind of tickets, and another said they sometimes blocked clean work.

## Distribution of Written Sentiment

Figure 2. Distribution of written sentiment (n = 17 responses)ExhibitNote: Column height is the share of the 17 responses. Labels give the count and the share.Table 6. Summary statistics of written sentimentTable

| Statistic | Value | Basis |
| --- | --- | --- |
| Mean written score | 1.2 | † n = 17 |
| Median written score | 5.0 | n = 17 responses |
| Negative responses | 7 of 17 (41.18%) | † n = 17 |
| Strongest negative responses (−10) | 2 of 17 (11.76%) | Statements of totality, constancy or serious harm |
| Positive responses | 10 of 17 (58.82%) | +5 or +10 |
| Net balance | 17.65 percentage points | Share positive minus share negative, base 17 responses |
| Respondent-level mean | 1.2 | Mean of each respondent's own mean written score, n = 17 respondents |

Note: † Base under 30. No interval is reported.

Of the 17 responses, 7 (41.18%, 95% CI 16.67% to 68.75%) were negative and 2 (11.76%) carried the strongest negative score of −10. Positive responses numbered 10 (58.82%), and no response was neutral. The mean written score was 1.2 (95% CI −2.5 to 4.7, n = 17 responses), and the median was 5.0.

## Correspondence Between Ratings and Written Sentiment

Figure 3. Written score against the respondent's rating of the indicator (n = 17 responses with a rating)ExhibitNote: Each marker is one combination of written score and rating, and its area is proportional to the number of responses at that combination. The written score runs from −10 at the left to +10 at the right. The rating rises from the worst outcome at the bottom to the best at the top, and ratings of 3 or below are the distressed range. Of the 17 responses, 0 had no rating and are not shown.

Of the 17 responses, 17 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.57, 95% CI 0.14 to 0.84, p = 0.018, n = 17 responses). The written responses corroborated the rating. Of the 7 negative responses with a rating, 4 came from respondents whose rating was in the distressed range.

## Rating Profile of Respondents Providing Written Responses

Figure 4. Distribution of ratings, respondents who wrote about the indicator and respondents who did notExhibitNote: The columns of each group sum to the whole of that group: 17 respondents who wrote about the indicator and gave a rating, and 770 who gave a rating without writing about it.

Respondents who wrote about the indicator and gave a rating numbered 17, with a mean rating of 5.4. The 770 respondents who rated the indicator without writing about it had a mean of 4.6. The difference, 0.8 (95% CI −0.3 to 1.9), included zero. Ratings in the distressed range made up 23.53% of the ratings from respondents who wrote (4 of 17) and 32.21% of the ratings from respondents who did not (248 of 770).

## Descriptive Statistics by Segment

Table 7a. Descriptive statistics by segment, part 1 of 2Table

| Segment | Base | Answered | Answer rate | Engagements | Mean | 95% CI | SD |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 850 | 787 | 92.59% of 850 | 150 | 4.7 | 4.5 to 4.8 | 2.1 |
| Outsourcing model | | | | | | | |
| CXM | 232 | 215 | 92.67% of 232 | 42 | 4.8 | 4.5 to 5.1 | 2.2 |
| ITO | 244 | 227 | 93.03% of 244 | 40 | 4.8 | 4.6 to 5.1 | 1.9 |
| BPO | 209 | 189 | 90.43% of 209 | 37 | 4.3 | 4.0 to 4.6 | 2.1 |
| KPO | 165 | 156 | 94.55% of 165 | 31 | 4.7 | 4.4 to 5.1 | 2.2 |
| Institution type | | | | | | | |
| Asset management | 132 | 126 | 95.45% of 132 | 24 | 4.4 | 4.1 to 4.8 | 2.2 |
| Wealth management and private banking | 140 | 127 | 90.71% of 140 | 23 | 4.7 | 4.3 to 5.1 | 2.3 |
| Investment banking (advisory and capital-markets origination) | 105 | 99 | 94.29% of 105 | 19 | 4.5 | 4.1 to 4.9 | 1.9 |
| Insurance (life, property and casualty, health) | 78 | 71 | 91.03% of 78 | 14 | 4.8 | 4.3 to 5.4 | 2.3 |
| Commercial and business banking | 82 | 76 | 92.68% of 82 | 13 | 5.1 | 4.6 to 5.5 | 1.9 |
| Payments and merchant acquiring (networks, acquiring, processing) | 71 | 65 | 91.55% of 71 | 13 | 4.5 | 4.0 to 5.0 | 2.0 |
| Credit card and payments issuing | 73 | 69 | 94.52% of 73 | 12 | 4.2 | 3.8 to 4.7 | 1.8 |
| Retail and consumer banking | 73 | 66 | 90.41% of 73 | 12 | 4.6 | 4.0 to 5.1 | 2.2 |
| Markets and trading (sales and trading, global markets) | 57 | 53 | 92.98% of 57 | 11 | 5.5 | 4.9 to 6.2 | 2.4 |
| Private equity and private capital | 39 | 35 | 89.74% of 39 | 9 | 4.5 | 3.9 to 5.2 | 1.8 |
| Process family | | | | | | | |
| Corporate and back-office functions | 248 | 232 | 93.55% of 248 | 47 | 4.5 | 4.3 to 4.8 | 2.2 |
| Technology engineering and operations | 204 | 190 | 93.14% of 204 | 35 | 5.2 | 4.9 to 5.4 | 1.9 |
| Client servicing and communications | 94 | 86 | 91.49% of 94 | 16 | 4.5 | 3.9 to 5.0 | 2.4 |
| Trade, settlement and reconciliation | 80 | 71 | 88.75% of 80 | 15 | 4.3 | 3.8 to 4.8 | 2.1 |
| Payments, cards and disputes | 73 | 66 | 90.41% of 73 | 12 | 4.7 | 4.4 to 5.1 | 1.6 |
| Credit, underwriting and suitability | 57 | 55 | 96.49% of 57 | 9 | 4.9 | 4.4 to 5.5 | 2.0 |
| Regulatory, audit, tax and legal | 51 | 46 | 90.20% of 51 | 9 | 4.4 | 3.7 to 5.1 | 2.3 |
| Financial crime and client onboarding | 43 | 41 | 95.35% of 43 | 7 | 3.8 | 3.2 to 4.4 | 1.9 |
| Class of work | | | | | | | |
| LOB Operational | 451 | 419 | 92.90% of 451 | 76 | 4.5 | 4.3 to 4.7 | 2.1 |
| Corporate Functional | 399 | 368 | 92.23% of 399 | 74 | 4.8 | 4.6 to 5.0 | 2.1 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 175 | 160 | 91.43% of 175 | 31 | 5.5 | 5.2 to 5.8 | 1.9 |
| Active, Transition/Cut-Over | 250 | 238 | 95.20% of 250 | 43 | 5.6 | 5.4 to 5.9 | 1.9 |
| Active, Production | 103 | 97 | 94.17% of 103 | 20 | 5.3 | 4.9 to 5.7 | 2.1 |
| Canceled | 322 | 292 | 90.68% of 322 | 56 | 3.2 | 3.0 to 3.3 | 1.5 |
| Respondent role | | | | | | | |
| Individual Contributor | 667 | 606 | 90.85% of 667 | 150 | 4.6 | 4.4 to 4.7 | 2.2 |
| Engagement Manager | 150 | 150 | 100.00% of 150 | 150 | 4.9 | 4.5 to 5.2 | 2.0 |
| Portfolio Director | 31 | 29 | 93.55% of 31 | 29 | 5.4 | † | 1.8 |
| Customer Management Executive | 2 | 2 | 2 of 2 | 2 | ‡ | | |
| Vendor type | | | | | | | |
| IT services major | 316 | 292 | 92.41% of 316 | 59 | 4.9 | 4.6 to 5.1 | 2.1 |
| BPO specialist | 198 | 187 | 94.44% of 198 | 37 | 4.2 | 3.9 to 4.5 | 2.1 |
| Consulting firm | 117 | 109 | 93.16% of 117 | 23 | 5.3 | 4.9 to 5.7 | 2.2 |
| Boutique or specialist provider | 87 | 79 | 90.80% of 87 | 17 | 4.4 | 4.0 to 4.9 | 2.0 |
| Pure-play agentic AI firm | 75 | 67 | 89.33% of 75 | 14 | 4.3 | 3.8 to 4.8 | 2.1 |
| Not reported, recovered by engagement | 57 | 53 | 92.98% of 57 | 45 | 4.4 | 3.9 to 4.9 | 1.8 |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported.Table 7b. Descriptive statistics by segment, part 2 of 2Table

| Segment | Min | Q1 | Median | Q3 | Max | Distressed n | Distressed % |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 252 | 32.02% |
| Outsourcing model | | | | | | | |
| CXM | 1.0 | 3.0 | 5.0 | 7.0 | 10.0 | 70 | 32.56% |
| ITO | 1.0 | 3.0 | 5.0 | 6.0 | 9.0 | 58 | 25.55% |
| BPO | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 71 | 37.57% |
| KPO | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 53 | 33.97% |
| Institution type | | | | | | | |
| Asset management | 1.0 | 3.0 | 4.0 | 6.0 | 9.0 | 45 | 35.71% |
| Wealth management and private banking | 1.0 | 3.0 | 4.0 | 7.0 | 10.0 | 44 | 34.65% |
| Investment banking (advisory and capital-markets origination) | 1.0 | 3.0 | 5.0 | 6.0 | 9.0 | 29 | 29.29% |
| Insurance (life, property and casualty, health) | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 24 | 33.80% |
| Commercial and business banking | 1.0 | 3.8 | 5.0 | 6.0 | 9.0 | 19 | 25.00% |
| Payments and merchant acquiring (networks, acquiring, processing) | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 19 | 29.23% |
| Credit card and payments issuing | 1.0 | 3.0 | 4.0 | 5.0 | 9.0 | 24 | 34.78% |
| Retail and consumer banking | 1.0 | 3.0 | 5.0 | 6.0 | 9.0 | 23 | 34.85% |
| Markets and trading (sales and trading, global markets) | 1.0 | 4.0 | 6.0 | 7.0 | 10.0 | 13 | 24.53% |
| Private equity and private capital | 1.0 | 3.0 | 4.0 | 6.0 | 8.0 | 12 | 34.29% |
| Process family | | | | | | | |
| Corporate and back-office functions | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 85 | 36.64% |
| Technology engineering and operations | 1.0 | 4.0 | 5.0 | 6.8 | 10.0 | 36 | 18.95% |
| Client servicing and communications | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 35 | 40.70% |
| Trade, settlement and reconciliation | 1.0 | 3.0 | 4.0 | 6.0 | 9.0 | 28 | 39.44% |
| Payments, cards and disputes | 1.0 | 4.0 | 5.0 | 6.0 | 8.0 | 13 | 19.70% |
| Credit, underwriting and suitability | 1.0 | 3.0 | 5.0 | 6.0 | 9.0 | 15 | 27.27% |
| Regulatory, audit, tax and legal | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 21 | 45.65% |
| Financial crime and client onboarding | 1.0 | 3.0 | 4.0 | 5.0 | 8.0 | 19 | 46.34% |
| Class of work | | | | | | | |
| LOB Operational | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 148 | 35.32% |
| Corporate Functional | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 104 | 28.26% |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 1.0 | 4.0 | 5.0 | 7.0 | 9.0 | 23 | 14.38% |
| Active, Transition/Cut-Over | 1.0 | 4.0 | 6.0 | 7.0 | 10.0 | 33 | 13.87% |
| Active, Production | 1.0 | 4.0 | 5.0 | 7.0 | 10.0 | 21 | 21.65% |
| Canceled | 1.0 | 2.0 | 3.0 | 4.0 | 8.0 | 175 | 59.93% |
| Respondent role | | | | | | | |
| Individual Contributor | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 207 | 34.16% |
| Engagement Manager | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 41 | 27.33% |
| Portfolio Director | 1.0 | 4.0 | 6.0 | 7.0 | 8.0 | 4 | 13.79% |
| Customer Management Executive | ‡ n = 2 | | | | | | |
| Vendor type | | | | | | | |
| IT services major | 1.0 | 3.0 | 5.0 | 6.0 | 10.0 | 83 | 28.42% |
| BPO specialist | 1.0 | 3.0 | 4.0 | 6.0 | 10.0 | 73 | 39.04% |
| Consulting firm | 1.0 | 4.0 | 5.0 | 7.0 | 9.0 | 27 | 24.77% |
| Boutique or specialist provider | 1.0 | 3.0 | 4.0 | 6.0 | 8.0 | 27 | 34.18% |
| Pure-play agentic AI firm | 1.0 | 3.0 | 4.0 | 6.0 | 9.0 | 24 | 35.82% |
| Not reported, recovered by engagement | 1.0 | 3.0 | 4.0 | 6.0 | 8.0 | 18 | 33.96% |

Note: ‡ Base under 10. No statistic is reported.

Every figure in the segment tables is at respondent level. Segments with fewer than 30 respondents carry no interval. Segments with fewer than 10 report counts only.

## Tests of Difference Between Segments

Table 8a. Tests of difference between segments, part 1 of 2Table

| Segment family | Welch F | df | p | Holm p | Omega squared | 90% CI |
| --- | --- | --- | --- | --- | --- | --- |
| Outsourcing model | 0.72 | 3, 78.5 | 0.542 | 1.000 | −0.006 | 0.000 to 0.042 |
| Institution type | 0.51 | 8, 49.4 | 0.843 | 1.000 | −0.024 | 0.000 to 0.033 |
| Process family | 0.94 | 4, 40.7 | 0.449 | 1.000 | −0.003 | 0.000 to 0.064 |
| Class of work | 1.14 | 1, 147.1 | 0.288 | 1.000 | 0.001 | 0.000 to 0.047 |
| Engagement status and lifecycle stage | 51.61 | 3, 53.4 | < 0.001 | < 0.001 | 0.424 | 0.330 to 0.510 |
| Respondent role | 4.11 | 2, 73.4 | 0.020 | 0.102 | 0.006 | 0.000 to 0.020 |
| Vendor type | 4.64 | 5, 231.6 | < 0.001 | 0.003 | 0.024 | 0.009 to 0.047 |

Table 8b. Tests of difference between segments, part 2 of 2Table

| Segment family | Kruskal-Wallis H | p | Distressed chi-square | df | p | Cramer's V | Levels excluded |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Outsourcing model | 2.92 | 0.403 | 5.49 | 3 | 0.139 | 0.19 | None |
| Institution type | 4.83 | 0.776 | 7.74 | 8 | 0.470, simulated | 0.23 | Private equity and private capital, n = 9 |
| Process family | 5.55 | 0.235 | 7.44 | 4 | 0.112, simulated | 0.24 | Credit, underwriting and suitability, n = 9, Regulatory, audit, tax and legal, n = 9, Financial crime and client onboarding, n = 7 |
| Class of work | 0.96 | 0.327 | 1.23 | 1 | 0.267 | 0.09 | None |
| Engagement status and lifecycle stage | 72.19 | < 0.001 | 50.14 | 3 | < 0.001, simulated | 0.58 | None |
| Respondent role | 7.27 | 0.026 | 7.20 | 2 | 0.027 | 0.10 | Customer Management Executive, n = 2 |
| Vendor type | 23.71 | < 0.001 | 9.30 | 5 | 0.098 | 0.11 | None |

Segment families recorded at engagement level were tested on engagement means. Respondent role and vendor type were tested on respondents. Where an expected count fell below 5, the test of the distressed share used Fisher's exact test for a two-by-two table, marked exact, or a Monte Carlo test, marked simulated. After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 51.61, Holm p < 0.001, omega squared 0.424). Active engagements in Transition/Cut-Over rated highest (mean 5.6, n = 238 respondents) and canceled engagements lowest (mean 3.2, n = 292 respondents). Ratings also differed by vendor type (Welch F 4.64, Holm p = 0.003, omega squared 0.024). Consulting firms rated highest (mean 5.3, n = 109 respondents) and BPO specialists lowest (mean 4.2, n = 187 respondents). Outsourcing model, institution type, process family, class of work and respondent role showed no difference.

## Written Sentiment by Segment

Figure 5. Written sentiment by engagement status and lifecycle stage, share of each group's responsesExhibitNote: Each bar is the whole of its group's responses, divided by written score from −10 at the left to +10 at the right. Segments large enough to hold a label carry their share.Table 9. Written sentiment by segmentTable

| Segment | Responses | Respondents | Engagements | Mean written score | Negative | Strongest negative (−10) |
| --- | --- | --- | --- | --- | --- | --- |
| Engagement status and lifecycle stage | | | | | | |
| Active, Build-Out | 4 | 4 | 3 | ‡ | 1 of 4 | 0 of 4 |
| Active, Transition/Cut-Over | 9 | 9 | 7 | ‡ | 5 of 9 | 1 of 9 |
| Active, Production | 2 | 2 | 2 | ‡ | 1 of 2 | 1 of 2 |
| Canceled | 2 | 2 | 2 | ‡ | 0 of 2 | 0 of 2 |
| Respondent role | | | | | | |
| Individual Contributor | 11 | 11 | 11 | −0.5 | 6 of 11 (54.55%) | 2 of 11 (18.18%) |
| Engagement Manager | 5 | 5 | 5 | ‡ | 1 of 5 | 0 of 5 |
| Portfolio Director | 1 | 1 | 1 | ‡ | 0 of 1 | 0 of 1 |
| Customer Management Executive | 0 | 0 | 0 | No responses | | |

Note: ‡ Base under 10. No statistic is reported.

Written sentiment did not differ by engagement status and lifecycle stage across the 14 engagements with a written response (Kruskal-Wallis H 2.85, df 3, p = 0.415). Fewer than two groups held 10 or more responses, so the groups are reported as counts only: active engagements in Build-Out (4), active engagements in Transition/Cut-Over (9), active engagements in Production (2) and canceled engagements (2).

No test across respondent roles was run, because fewer than two roles had 30 or more written responses. Individual contributors wrote 11, engagement managers wrote 5 and portfolio directors wrote 1 responses, reported as counts only.

## Association with Cancellation Rate

Cancellation rate is the share of engagements canceled by the study date, measured at engagement level (150 engagements, 56 canceled). Canceled engagements were rated after cancellation. The association is therefore a descriptive contrast and not a forecast of cancellation.

Table 10a. Association with cancellation rate by segment, part 1 of 2Table

| Segment | Engagements | Canceled | Cancellation rate | Point-biserial r | 95% CI | p | Holm p |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 150 | 56 | 37.33% of 150 | −0.66 | −0.74 to −0.56 | < 0.001 | < 0.001 |
| Outsourcing model | | | | | | | |
| CXM | 42 | 15 | 35.71% of 42 | −0.62 | −0.78 to −0.39 | < 0.001 | < 0.001 |
| ITO | 40 | 15 | 37.50% of 40 | −0.68 | −0.82 to −0.46 | < 0.001 | < 0.001 |
| BPO | 37 | 14 | 37.84% of 37 | −0.68 | −0.82 to −0.46 | < 0.001 | < 0.001 |
| KPO | 31 | 12 | 38.71% of 31 | −0.69 | −0.84 to −0.45 | < 0.001 | < 0.001 |
| Institution type | | | | | | | |
| Asset management | 24 | 13 | 54.17% of 24 | −0.83 | † | < 0.001 | < 0.001 |
| Wealth management and private banking | 23 | 8 | 34.78% of 23 | −0.65 | † | < 0.001 | 0.009 |
| Investment banking (advisory and capital-markets origination) | 19 | 9 | 47.37% of 19 | −0.60 | † | 0.007 | 0.041 |
| Insurance (life, property and casualty, health) | 14 | 5 | 35.71% of 14 | −0.69 | † | 0.007 | 0.041 |
| Commercial and business banking | 13 | 4 | 30.77% of 13 | −0.64 | † | 0.018 | 0.070 |
| Payments and merchant acquiring (networks, acquiring, processing) | 13 | 3 | 23.08% of 13 | −0.57 | † | 0.041 | 0.122 |
| Credit card and payments issuing | 12 | 5 | 41.67% of 12 | −0.45 | † | 0.139 | 0.257 |
| Retail and consumer banking | 12 | 4 | 33.33% of 12 | −0.76 | † | 0.004 | 0.030 |
| Markets and trading (sales and trading, global markets) | 11 | 3 | 27.27% of 11 | −0.83 | † | 0.002 | 0.016 |
| Private equity and private capital | 9 | 2 | 2 of 9 | ‡ | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 47 | 19 | 40.43% of 47 | −0.66 | −0.79 to −0.45 | < 0.001 | < 0.001 |
| Technology engineering and operations | 35 | 10 | 28.57% of 35 | −0.66 | −0.82 to −0.43 | < 0.001 | < 0.001 |
| Client servicing and communications | 16 | 9 | 56.25% of 16 | −0.69 | † | 0.003 | 0.024 |
| Trade, settlement and reconciliation | 15 | 8 | 53.33% of 15 | −0.72 | † | 0.002 | 0.021 |
| Payments, cards and disputes | 12 | 2 | 16.67% of 12 | −0.46 | † | 0.129 | 0.257 |
| Credit, underwriting and suitability | 9 | 1 | 1 of 9 | ‡ | | | |
| Regulatory, audit, tax and legal | 9 | 4 | 4 of 9 | ‡ | | | |
| Financial crime and client onboarding | 7 | 3 | 3 of 7 | ‡ | | | |
| Class of work | | | | | | | |
| LOB Operational | 76 | 30 | 39.47% of 76 | −0.68 | −0.79 to −0.54 | < 0.001 | < 0.001 |
| Corporate Functional | 74 | 26 | 35.14% of 74 | −0.64 | −0.75 to −0.48 | < 0.001 | < 0.001 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 31 | 0 | 0.00% of 31 | a | | | |
| Active, Transition/Cut-Over | 43 | 0 | 0.00% of 43 | a | | | |
| Active, Production | 20 | 0 | 0.00% of 20 | a | | | |
| Canceled | 56 | 56 | 100.00% of 56 | a | | | |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported. a Not tested: no variation in cancellation.Table 10b. Association with cancellation rate by segment, part 2 of 2Table

| Segment | Odds ratio per point | 95% CI | AUC | Rate, distressed engagements | Rate, other engagements | Fisher p | TOST p |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 0.18 | 0.10 to 0.32 | 0.91 | 90.63% of 32 | 22.88% of 118 | < 0.001 | 1.000 |
| Outsourcing model | | | | | | | |
| CXM | 0.27 | 0.12 to 0.64 | 0.88 | 5 of 7 | 28.57% of 35 | 0.077 | 0.995 |
| ITO | 0.08 | 0.01 to 0.49 | 0.92 | 5 of 5 | 28.57% of 35 | 0.005 | 0.999 |
| BPO | 0.06 | 0.01 to 0.52 | 0.96 | 91.67% of 12 | 12.00% of 25 | < 0.001 | 0.999 |
| KPO | 0.14 | 0.03 to 0.57 | 0.94 | 8 of 8 | 17.39% of 23 | < 0.001 | 0.998 |
| Institution type | | | | | | | |
| Asset management | 0.03 | † n = 24 | 1.00 | 8 of 8 | 31.25% of 16 | 0.002 | 1.000 |
| Wealth management and private banking | 0.27 | † n = 23 | 0.90 | 5 of 6 | 17.65% of 17 | 0.009 | 0.982 |
| Investment banking (advisory and capital-markets origination) | 0.17 | † n = 19 | 0.86 | 1 of 1 | 44.44% of 18 | 0.474 | 0.935 |
| Insurance (life, property and casualty, health) | 0.18 | † n = 14 | 0.93 | 3 of 3 | 18.18% of 11 | 0.027 | 0.960 |
| Commercial and business banking | 0.13 | † n = 13 | 0.89 | 1 of 1 | 25.00% of 12 | 0.308 | 0.925 |
| Payments and merchant acquiring (networks, acquiring, processing) | 0.02 | † n = 13 | 0.92 | 2 of 3 | 10.00% of 10 | 0.108 | 0.861 |
| Credit card and payments issuing | 0.37 | † n = 12 | 0.80 | 2 of 3 | 3 of 9 | 0.523 | 0.704 |
| Retail and consumer banking | 0.19 | † n = 12 | 1.00 | 4 of 4 | 0 of 8 | 0.002 | 0.980 |
| Markets and trading (sales and trading, global markets) | 0.31 | † n = 11 | 1.00 | 2 of 2 | 1 of 9 | 0.055 | 0.993 |
| Private equity and private capital | ‡ n = 9 | | | | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 0.21 | 0.08 to 0.56 | 0.90 | 91.67% of 12 | 22.86% of 35 | < 0.001 | 0.999 |
| Technology engineering and operations | 0.03 | 0.00 to 0.63 | 0.96 | 3 of 3 | 21.88% of 32 | 0.018 | 0.997 |
| Client servicing and communications | 0.32 | † n = 16 | 0.86 | 4 of 5 | 45.45% of 11 | 0.308 | 0.974 |
| Trade, settlement and reconciliation | 0.20 | † n = 15 | 0.90 | 4 of 5 | 40.00% of 10 | 0.282 | 0.982 |
| Payments, cards and disputes | 0.17 | † n = 12 | 0.85 | 1 of 1 | 9.09% of 11 | 0.167 | 0.718 |
| Credit, underwriting and suitability | ‡ n = 9 | | | | | | |
| Regulatory, audit, tax and legal | ‡ n = 9 | | | | | | |
| Financial crime and client onboarding | ‡ n = 7 | | | | | | |
| Class of work | | | | | | | |
| LOB Operational | 0.16 | 0.07 to 0.36 | 0.93 | 94.74% of 19 | 21.05% of 57 | < 0.001 | 1.000 |
| Corporate Functional | 0.20 | 0.09 to 0.45 | 0.89 | 84.62% of 13 | 24.59% of 61 | < 0.001 | 1.000 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | a | | | | | | |
| Active, Transition/Cut-Over | a | | | | | | |
| Active, Production | a | | | | | | |
| Canceled | a | | | | | | |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported. a Not tested: no variation in cancellation.

Across the study set, higher ratings came with less cancellation (r = −0.66, 95% CI −0.74 to −0.56, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in sixteen of the 20 segment levels tested.

## Association with Customer Satisfaction

Customer satisfaction is the percentage recorded against each engagement. The primary estimate uses the 94 active engagements. A descriptive contrast uses all 150 engagements.

Table 11a. Association with customer satisfaction by segment, part 1 of 2Table

| Segment | Active engagements | Clean r | 95% CI | p | Holm p | Spearman rho | p |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 94 | 0.65 | 0.52 to 0.76 | < 0.001 | < 0.001 | 0.67 | < 0.001 |
| Outsourcing model | | | | | | | |
| CXM | 27 | 0.74 | † | < 0.001 | < 0.001 | 0.76 | < 0.001 |
| ITO | 25 | 0.36 | † | 0.073 | 0.220 | 0.34 | 0.095 |
| BPO | 23 | 0.81 | † | < 0.001 | < 0.001 | 0.85 | < 0.001 |
| KPO | 19 | 0.79 | † | < 0.001 | < 0.001 | 0.78 | < 0.001 |
| Institution type | | | | | | | |
| Asset management | 11 | 0.37 | † | 0.258 | 0.258 | 0.33 | 0.324 |
| Wealth management and private banking | 15 | 0.73 | † | 0.002 | 0.011 | 0.72 | 0.002 |
| Investment banking (advisory and capital-markets origination) | 10 | 0.75 | † | 0.013 | 0.053 | 0.63 | 0.053 |
| Insurance (life, property and casualty, health) | 9 | ‡ | | | | | |
| Commercial and business banking | 9 | ‡ | | | | | |
| Payments and merchant acquiring (networks, acquiring, processing) | 10 | 0.88 | † | < 0.001 | 0.006 | 0.82 | 0.004 |
| Credit card and payments issuing | 7 | ‡ | | | | | |
| Retail and consumer banking | 8 | ‡ | | | | | |
| Markets and trading (sales and trading, global markets) | 8 | ‡ | | | | | |
| Private equity and private capital | 7 | ‡ | | | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 28 | 0.68 | † | < 0.001 | < 0.001 | 0.72 | < 0.001 |
| Technology engineering and operations | 25 | 0.36 | † | 0.074 | 0.220 | 0.31 | 0.126 |
| Client servicing and communications | 7 | ‡ | | | | | |
| Trade, settlement and reconciliation | 7 | ‡ | | | | | |
| Payments, cards and disputes | 10 | 0.88 | † | < 0.001 | 0.006 | 0.86 | 0.001 |
| Credit, underwriting and suitability | 8 | ‡ | | | | | |
| Regulatory, audit, tax and legal | 5 | ‡ | | | | | |
| Financial crime and client onboarding | 4 | ‡ | | | | | |
| Class of work | | | | | | | |
| LOB Operational | 46 | 0.80 | 0.66 to 0.88 | < 0.001 | < 0.001 | 0.81 | < 0.001 |
| Corporate Functional | 48 | 0.52 | 0.27 to 0.70 | < 0.001 | 0.002 | 0.51 | < 0.001 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 31 | 0.58 | 0.28 to 0.77 | < 0.001 | 0.006 | 0.63 | < 0.001 |
| Active, Transition/Cut-Over | 43 | 0.78 | 0.62 to 0.87 | < 0.001 | < 0.001 | 0.77 | < 0.001 |
| Active, Production | 20 | 0.58 | † | 0.007 | 0.034 | 0.57 | 0.009 |
| Canceled | 0 | a | | | | | |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported. a Not tested: canceled engagements fall outside the active population.Table 11b. Association with customer satisfaction by segment, part 2 of 2Table

| Segment | Slope, points per rating point | 95% CI | R squared | TOST p | All engagements | Descriptive r | p |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 4.83 | 3.67 to 5.98 | 0.43 | 1.000 | 150 | 0.77 | < 0.001 |
| Outsourcing model | | | | | | | |
| CXM | 5.37 | † n = 27 | 0.55 | 0.999 | 42 | 0.78 | < 0.001 |
| ITO | 2.80 | † n = 25 | 0.13 | 0.633 | 40 | 0.69 | < 0.001 |
| BPO | 5.35 | † n = 23 | 0.65 | 1.000 | 37 | 0.81 | < 0.001 |
| KPO | 6.25 | † n = 19 | 0.62 | 0.999 | 31 | 0.84 | < 0.001 |
| Institution type | | | | | | | |
| Asset management | 2.33 | † n = 11 | 0.14 | 0.592 | 24 | 0.84 | < 0.001 |
| Wealth management and private banking | 4.03 | † n = 15 | 0.54 | 0.985 | 23 | 0.76 | < 0.001 |
| Investment banking (advisory and capital-markets origination) | 5.99 | † n = 10 | 0.56 | 0.958 | 19 | 0.75 | < 0.001 |
| Insurance (life, property and casualty, health) | ‡ n = 9 | | | | | | |
| Commercial and business banking | ‡ n = 9 | | | | | | |
| Payments and merchant acquiring (networks, acquiring, processing) | 7.73 | † n = 10 | 0.77 | 0.997 | 13 | 0.82 | < 0.001 |
| Credit card and payments issuing | ‡ n = 7 | | | | | | |
| Retail and consumer banking | ‡ n = 8 | | | | | | |
| Markets and trading (sales and trading, global markets) | ‡ n = 8 | | | | | | |
| Private equity and private capital | ‡ n = 7 | | | | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 4.68 | † n = 28 | 0.46 | 0.995 | 47 | 0.78 | < 0.001 |
| Technology engineering and operations | 2.94 | † n = 25 | 0.13 | 0.631 | 35 | 0.70 | < 0.001 |
| Client servicing and communications | ‡ n = 7 | | | | | | |
| Trade, settlement and reconciliation | ‡ n = 7 | | | | | | |
| Payments, cards and disputes | 8.92 | † n = 10 | 0.78 | 0.998 | 12 | 0.74 | 0.006 |
| Credit, underwriting and suitability | ‡ n = 8 | | | | | | |
| Regulatory, audit, tax and legal | ‡ n = 5 | | | | | | |
| Financial crime and client onboarding | ‡ n = 4 | | | | | | |
| Class of work | | | | | | | |
| LOB Operational | 6.95 | 5.35 to 8.54 | 0.64 | 1.000 | 76 | 0.81 | < 0.001 |
| Corporate Functional | 3.20 | 1.63 to 4.77 | 0.27 | 0.961 | 74 | 0.72 | < 0.001 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 3.65 | 1.69 to 5.60 | 0.33 | 0.968 | 31 | 0.58 | < 0.001 |
| Active, Transition/Cut-Over | 6.69 | 4.98 to 8.40 | 0.60 | 1.000 | 43 | 0.78 | < 0.001 |
| Active, Production | 3.75 | † | 0.34 | 0.931 | 20 | 0.58 | 0.007 |
| Canceled | a | | | | | | |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported. a Not tested: canceled engagements fall outside the active population.

Across the active engagements, higher ratings came with higher customer satisfaction (r = 0.65, 95% CI 0.52 to 0.76, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in twelve of the 16 segment levels tested.

## Association with the Engagement Indicator Mean

The engagement indicator mean is the mean of an engagement's ratings across the rated indicators, measured at engagement level (150 engagements). The indicator is one of the rated indicators. Each comparison therefore uses the mean recomputed without it.

Table 12a. Association with the engagement indicator mean by segment, part 1 of 2Table

| Segment | Engagements | Leave-one-out r | 95% CI | p | Holm p | Spearman rho | p |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 150 | 0.86 | 0.82 to 0.90 | < 0.001 | < 0.001 | 0.86 | < 0.001 |
| Outsourcing model | | | | | | | |
| CXM | 42 | 0.91 | 0.83 to 0.95 | < 0.001 | < 0.001 | 0.90 | < 0.001 |
| ITO | 40 | 0.73 | 0.55 to 0.85 | < 0.001 | < 0.001 | 0.76 | < 0.001 |
| BPO | 37 | 0.89 | 0.80 to 0.94 | < 0.001 | < 0.001 | 0.89 | < 0.001 |
| KPO | 31 | 0.89 | 0.79 to 0.95 | < 0.001 | < 0.001 | 0.87 | < 0.001 |
| Institution type | | | | | | | |
| Asset management | 24 | 0.88 | † | < 0.001 | < 0.001 | 0.92 | < 0.001 |
| Wealth management and private banking | 23 | 0.91 | † | < 0.001 | < 0.001 | 0.89 | < 0.001 |
| Investment banking (advisory and capital-markets origination) | 19 | 0.81 | † | < 0.001 | < 0.001 | 0.78 | < 0.001 |
| Insurance (life, property and casualty, health) | 14 | 0.90 | † | < 0.001 | < 0.001 | 0.95 | < 0.001 |
| Commercial and business banking | 13 | 0.92 | † | < 0.001 | < 0.001 | 0.95 | < 0.001 |
| Payments and merchant acquiring (networks, acquiring, processing) | 13 | 0.93 | † | < 0.001 | < 0.001 | 0.88 | < 0.001 |
| Credit card and payments issuing | 12 | 0.79 | † | 0.002 | 0.002 | 0.67 | 0.017 |
| Retail and consumer banking | 12 | 0.86 | † | < 0.001 | 0.001 | 0.83 | < 0.001 |
| Markets and trading (sales and trading, global markets) | 11 | 0.86 | † | < 0.001 | 0.002 | 0.80 | 0.003 |
| Private equity and private capital | 9 | ‡ | | | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 47 | 0.89 | 0.82 to 0.94 | < 0.001 | < 0.001 | 0.85 | < 0.001 |
| Technology engineering and operations | 35 | 0.72 | 0.51 to 0.85 | < 0.001 | < 0.001 | 0.73 | < 0.001 |
| Client servicing and communications | 16 | 0.93 | † | < 0.001 | < 0.001 | 0.89 | < 0.001 |
| Trade, settlement and reconciliation | 15 | 0.92 | † | < 0.001 | < 0.001 | 0.91 | < 0.001 |
| Payments, cards and disputes | 12 | 0.82 | † | < 0.001 | 0.002 | 0.73 | 0.008 |
| Credit, underwriting and suitability | 9 | ‡ | | | | | |
| Regulatory, audit, tax and legal | 9 | ‡ | | | | | |
| Financial crime and client onboarding | 7 | ‡ | | | | | |
| Class of work | | | | | | | |
| LOB Operational | 76 | 0.90 | 0.85 to 0.94 | < 0.001 | < 0.001 | 0.90 | < 0.001 |
| Corporate Functional | 74 | 0.83 | 0.74 to 0.89 | < 0.001 | < 0.001 | 0.82 | < 0.001 |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 31 | 0.73 | 0.51 to 0.86 | < 0.001 | < 0.001 | 0.75 | < 0.001 |
| Active, Transition/Cut-Over | 43 | 0.80 | 0.66 to 0.89 | < 0.001 | < 0.001 | 0.79 | < 0.001 |
| Active, Production | 20 | 0.83 | † | < 0.001 | < 0.001 | 0.80 | < 0.001 |
| Canceled | 56 | 0.48 | 0.24 to 0.66 | < 0.001 | 0.001 | 0.43 | < 0.001 |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported.Table 12b. Association with the engagement indicator mean by segment, part 2 of 2Table

| Segment | r with indicator included | Change in r | Slope | 95% CI | R squared | TOST p | Equivalent |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | 0.87 | 0.00 | 0.72 | 0.65 to 0.79 | 0.74 | 1.000 | No |
| Outsourcing model | | | | | | | |
| CXM | 0.91 | 0.00 | 0.78 | 0.67 to 0.90 | 0.82 | 1.000 | No |
| ITO | 0.74 | −0.01 | 0.60 | 0.42 to 0.78 | 0.54 | 1.000 | No |
| BPO | 0.90 | 0.00 | 0.67 | 0.56 to 0.79 | 0.80 | 1.000 | No |
| KPO | 0.90 | 0.00 | 0.78 | 0.63 to 0.93 | 0.80 | 1.000 | No |
| Institution type | | | | | | | |
| Asset management | 0.89 | 0.00 | 0.70 | † n = 24 | 0.78 | 1.000 | No |
| Wealth management and private banking | 0.91 | 0.00 | 0.68 | † n = 23 | 0.83 | 1.000 | No |
| Investment banking (advisory and capital-markets origination) | 0.82 | 0.00 | 0.64 | † n = 19 | 0.66 | 0.999 | No |
| Insurance (life, property and casualty, health) | 0.90 | 0.00 | 0.74 | † n = 14 | 0.81 | 1.000 | No |
| Commercial and business banking | 0.92 | 0.00 | 0.86 | † n = 13 | 0.85 | 1.000 | No |
| Payments and merchant acquiring (networks, acquiring, processing) | 0.93 | 0.00 | 0.84 | † n = 13 | 0.86 | 1.000 | No |
| Credit card and payments issuing | 0.79 | 0.00 | 0.71 | † n = 12 | 0.62 | 0.988 | No |
| Retail and consumer banking | 0.86 | 0.00 | 0.75 | † n = 12 | 0.74 | 0.998 | No |
| Markets and trading (sales and trading, global markets) | 0.86 | 0.00 | 0.58 | † n = 11 | 0.74 | 0.997 | No |
| Private equity and private capital | ‡ n = 9 | | | | | | |
| Process family | | | | | | | |
| Corporate and back-office functions | 0.90 | 0.00 | 0.74 | 0.63 to 0.85 | 0.80 | 1.000 | No |
| Technology engineering and operations | 0.73 | −0.01 | 0.61 | 0.40 to 0.81 | 0.52 | 1.000 | No |
| Client servicing and communications | 0.94 | 0.00 | 0.75 | † n = 16 | 0.87 | 1.000 | No |
| Trade, settlement and reconciliation | 0.92 | 0.00 | 0.69 | † n = 15 | 0.84 | 1.000 | No |
| Payments, cards and disputes | 0.83 | 0.00 | 0.90 | † n = 12 | 0.68 | 0.995 | No |
| Credit, underwriting and suitability | ‡ n = 9 | | | | | | |
| Regulatory, audit, tax and legal | ‡ n = 9 | | | | | | |
| Financial crime and client onboarding | ‡ n = 7 | | | | | | |
| Class of work | | | | | | | |
| LOB Operational | 0.90 | 0.00 | 0.78 | 0.70 to 0.87 | 0.81 | 1.000 | No |
| Corporate Functional | 0.83 | 0.00 | 0.66 | 0.56 to 0.77 | 0.69 | 1.000 | No |
| Engagement status and lifecycle stage | | | | | | | |
| Active, Build-Out | 0.74 | −0.01 | 0.51 | 0.33 to 0.69 | 0.53 | 0.999 | No |
| Active, Transition/Cut-Over | 0.81 | 0.00 | 0.66 | 0.50 to 0.81 | 0.64 | 1.000 | No |
| Active, Production | 0.83 | 0.00 | 0.63 | † n = 20 | 0.69 | 1.000 | No |
| Canceled | 0.48 | −0.01 | 0.38 | 0.19 to 0.57 | 0.23 | 0.936 | No |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported.

Higher ratings came with a higher engagement indicator mean (r = 0.86, 95% CI 0.82 to 0.90, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in twenty-four of the 24 segment levels tested.

## Association with Intent to Leave

Intent to leave is the answer to C019, “How likely are you to leave your current job in the next 12 months?”, rated 1 to 10. It is measured at respondent level on up to 798 respondents, with the engagement-level form reported alongside.

Table 13a. Association with intent to leave by segment, part 1 of 2Table

| Segment | Respondents | r | 95% CI | Cluster bootstrap 95% CI | p | Holm p |
| --- | --- | --- | --- | --- | --- | --- |
| Study set | 742 | −0.66 | −0.69 to −0.61 | −0.71 to −0.60 | < 0.001 | < 0.001 |
| Outsourcing model | | | | | | |
| CXM | 199 | −0.69 | −0.76 to −0.61 | −0.78 to −0.58 | < 0.001 | < 0.001 |
| ITO | 218 | −0.54 | −0.63 to −0.44 | −0.62 to −0.44 | < 0.001 | < 0.001 |
| BPO | 180 | −0.72 | −0.79 to −0.65 | −0.79 to −0.64 | < 0.001 | < 0.001 |
| KPO | 145 | −0.70 | −0.77 to −0.61 | −0.82 to −0.54 | < 0.001 | < 0.001 |
| Institution type | | | | | | |
| Asset management | 117 | −0.63 | −0.73 to −0.51 | −0.76 to −0.49 | < 0.001 | < 0.001 |
| Wealth management and private banking | 121 | −0.68 | −0.77 to −0.58 | −0.80 to −0.53 | < 0.001 | < 0.001 |
| Investment banking (advisory and capital-markets origination) | 92 | −0.52 | −0.66 to −0.36 | −0.67 to −0.28 | < 0.001 | < 0.001 |
| Insurance (life, property and casualty, health) | 66 | −0.77 | −0.86 to −0.66 | −0.87 to −0.61 | < 0.001 | < 0.001 |
| Commercial and business banking | 71 | −0.64 | −0.76 to −0.48 | −0.80 to −0.43 | < 0.001 | < 0.001 |
| Payments and merchant acquiring (networks, acquiring, processing) | 63 | −0.69 | −0.80 to −0.54 | −0.85 to −0.44 | < 0.001 | < 0.001 |
| Credit card and payments issuing | 66 | −0.59 | −0.73 to −0.41 | −0.78 to −0.30 | < 0.001 | < 0.001 |
| Retail and consumer banking | 60 | −0.78 | −0.86 to −0.65 | −0.86 to −0.62 | < 0.001 | < 0.001 |
| Markets and trading (sales and trading, global markets) | 52 | −0.76 | −0.86 to −0.61 | −0.85 to −0.57 | < 0.001 | < 0.001 |
| Private equity and private capital | 34 | −0.68 | −0.83 to −0.45 | § n = 9 | < 0.001 | < 0.001 |
| Process family | | | | | | |
| Corporate and back-office functions | 212 | −0.73 | −0.79 to −0.66 | −0.80 to −0.63 | < 0.001 | < 0.001 |
| Technology engineering and operations | 182 | −0.53 | −0.62 to −0.41 | −0.63 to −0.40 | < 0.001 | < 0.001 |
| Client servicing and communications | 81 | −0.63 | −0.74 to −0.47 | −0.79 to −0.23 | < 0.001 | < 0.001 |
| Trade, settlement and reconciliation | 69 | −0.66 | −0.77 to −0.50 | −0.76 to −0.45 | < 0.001 | < 0.001 |
| Payments, cards and disputes | 63 | −0.65 | −0.78 to −0.48 | −0.80 to −0.46 | < 0.001 | < 0.001 |
| Credit, underwriting and suitability | 51 | −0.84 | −0.90 to −0.73 | § n = 9 | < 0.001 | < 0.001 |
| Regulatory, audit, tax and legal | 44 | −0.82 | −0.90 to −0.70 | § n = 9 | < 0.001 | < 0.001 |
| Financial crime and client onboarding | 40 | −0.56 | −0.74 to −0.30 | § n = 7 | < 0.001 | < 0.001 |
| Class of work | | | | | | |
| LOB Operational | 395 | −0.65 | −0.71 to −0.59 | −0.73 to −0.56 | < 0.001 | < 0.001 |
| Corporate Functional | 347 | −0.66 | −0.72 to −0.60 | −0.73 to −0.58 | < 0.001 | < 0.001 |
| Engagement status and lifecycle stage | | | | | | |
| Active, Build-Out | 152 | −0.47 | −0.58 to −0.34 | −0.65 to −0.25 | < 0.001 | < 0.001 |
| Active, Transition/Cut-Over | 226 | −0.57 | −0.65 to −0.48 | −0.70 to −0.40 | < 0.001 | < 0.001 |
| Active, Production | 89 | −0.57 | −0.69 to −0.41 | −0.74 to −0.35 | < 0.001 | < 0.001 |
| Canceled | 275 | −0.09 | −0.21 to 0.02 | −0.24 to 0.05 | 0.121 | 0.121 |
| Respondent role | | | | | | |
| Individual Contributor | 562 | −0.66 | −0.70 to −0.61 | −0.71 to −0.60 | < 0.001 | < 0.001 |
| Engagement Manager | 150 | −0.65 | −0.73 to −0.54 | −0.73 to −0.55 | < 0.001 | < 0.001 |
| Portfolio Director | 29 | −0.63 | † | † | < 0.001 | < 0.001 |
| Customer Management Executive | 1 | ‡ | | | | |
| Vendor type | | | | | | |
| IT services major | 279 | −0.65 | −0.71 to −0.57 | −0.72 to −0.56 | < 0.001 | < 0.001 |
| BPO specialist | 174 | −0.72 | −0.79 to −0.64 | −0.81 to −0.60 | < 0.001 | < 0.001 |
| Consulting firm | 102 | −0.59 | −0.70 to −0.45 | −0.74 to −0.39 | < 0.001 | < 0.001 |
| Boutique or specialist provider | 73 | −0.74 | −0.83 to −0.61 | −0.84 to −0.55 | < 0.001 | < 0.001 |
| Pure-play agentic AI firm | 65 | −0.64 | −0.77 to −0.47 | −0.76 to −0.47 | < 0.001 | < 0.001 |
| Not reported, recovered by engagement | 49 | −0.55 | −0.72 to −0.31 | −0.73 to −0.32 | < 0.001 | < 0.001 |

Note: § Too few engagements for a cluster bootstrap interval. † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported.Table 13b. Association with intent to leave by segment, part 2 of 2Table

| Segment | Spearman rho | p | Share at 7 or above, distressed | Share at 7 or above, other | Engagements | Engagement level r | 95% CI | TOST p |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Study set | −0.66 | < 0.001 | 88.46% of 234 | 37.40% of 508 | 150 | −0.75 | −0.82 to −0.68 | 1.000 |
| Outsourcing model | | | | | | | | |
| CXM | −0.70 | < 0.001 | 79.03% of 62 | 27.74% of 137 | 42 | −0.78 | −0.88 to −0.63 | 1.000 |
| ITO | −0.56 | < 0.001 | 96.49% of 57 | 47.20% of 161 | 40 | −0.61 | −0.78 to −0.37 | 0.993 |
| BPO | −0.74 | < 0.001 | 87.88% of 66 | 34.21% of 114 | 37 | −0.83 | −0.91 to −0.69 | 1.000 |
| KPO | −0.70 | < 0.001 | 91.84% of 49 | 38.54% of 96 | 31 | −0.82 | −0.91 to −0.65 | 1.000 |
| Institution type | | | | | | | | |
| Asset management | −0.66 | < 0.001 | 100.00% of 42 | 45.33% of 75 | 24 | −0.71 | † | 0.996 |
| Wealth management and private banking | −0.68 | < 0.001 | 80.95% of 42 | 30.38% of 79 | 23 | −0.79 | † | 1.000 |
| Investment banking (advisory and capital-markets origination) | −0.52 | < 0.001 | 92.31% of 26 | 54.55% of 66 | 19 | −0.66 | † | 0.975 |
| Insurance (life, property and casualty, health) | −0.78 | < 0.001 | 100.00% of 21 | 31.11% of 45 | 14 | −0.88 | † | 1.000 |
| Commercial and business banking | −0.63 | < 0.001 | 93.75% of 16 | 32.73% of 55 | 13 | −0.76 | † | 0.986 |
| Payments and merchant acquiring (networks, acquiring, processing) | −0.71 | < 0.001 | 84.21% of 19 | 34.09% of 44 | 13 | −0.81 | † | 0.995 |
| Credit card and payments issuing | −0.57 | < 0.001 | 82.61% of 23 | 39.53% of 43 | 12 | −0.74 | † | 0.974 |
| Retail and consumer banking | −0.79 | < 0.001 | 75.00% of 20 | 25.00% of 40 | 12 | −0.90 | † | 1.000 |
| Markets and trading (sales and trading, global markets) | −0.73 | < 0.001 | 92.31% of 13 | 30.77% of 39 | 11 | −0.86 | † | 0.997 |
| Private equity and private capital | −0.66 | < 0.001 | 75.00% of 12 | 45.45% of 22 | 9 | ‡ | ‡ | |
| Process family | | | | | | | | |
| Corporate and back-office functions | −0.72 | < 0.001 | 88.00% of 75 | 37.96% of 137 | 47 | −0.83 | −0.90 to −0.72 | 1.000 |
| Technology engineering and operations | −0.56 | < 0.001 | 94.29% of 35 | 38.10% of 147 | 35 | −0.58 | −0.77 to −0.31 | 0.978 |
| Client servicing and communications | −0.55 | < 0.001 | 83.87% of 31 | 50.00% of 50 | 16 | −0.75 | † | 0.992 |
| Trade, settlement and reconciliation | −0.59 | < 0.001 | 92.86% of 28 | 51.22% of 41 | 15 | −0.77 | † | 0.993 |
| Payments, cards and disputes | −0.67 | < 0.001 | 58.33% of 12 | 23.53% of 51 | 12 | −0.81 | † | 0.993 |
| Credit, underwriting and suitability | −0.83 | < 0.001 | 92.86% of 14 | 21.62% of 37 | 9 | ‡ | ‡ | |
| Regulatory, audit, tax and legal | −0.78 | < 0.001 | 100.00% of 20 | 29.17% of 24 | 9 | ‡ | ‡ | |
| Financial crime and client onboarding | −0.56 | < 0.001 | 84.21% of 19 | 42.86% of 21 | 7 | ‡ | ‡ | |
| Class of work | | | | | | | | |
| LOB Operational | −0.64 | < 0.001 | 87.05% of 139 | 36.72% of 256 | 76 | −0.77 | −0.85 to −0.66 | 1.000 |
| Corporate Functional | −0.68 | < 0.001 | 90.53% of 95 | 38.10% of 252 | 74 | −0.74 | −0.83 to −0.62 | 1.000 |
| Engagement status and lifecycle stage | | | | | | | | |
| Active, Build-Out | −0.44 | < 0.001 | 57.89% of 19 | 21.05% of 133 | 31 | −0.48 | −0.71 to −0.15 | 0.868 |
| Active, Transition/Cut-Over | −0.54 | < 0.001 | 71.88% of 32 | 20.62% of 194 | 43 | −0.61 | −0.77 to −0.38 | 0.994 |
| Active, Production | −0.59 | < 0.001 | 50.00% of 20 | 14.49% of 69 | 20 | −0.60 | † | 0.941 |
| Canceled | −0.08 | 0.174 | 100.00% of 163 | 100.00% of 112 | 56 | −0.03 | −0.29 to 0.23 | 0.022 |
| Respondent role | | | | | | | | |
| Individual Contributor | −0.66 | < 0.001 | 88.89% of 189 | 38.07% of 373 | 150 | −0.76 | −0.82 to −0.68 | 1.000 |
| Engagement Manager | −0.65 | < 0.001 | 87.80% of 41 | 35.78% of 109 | 150 | −0.65 | −0.73 to −0.54 | 1.000 |
| Portfolio Director | −0.63 | < 0.001 | ‡ n = 4 | 36.00% of 25 | 29 | −0.63 | † | 0.985 |
| Customer Management Executive | ‡ n = 1 | | | | | | | |
| Vendor type | | | | | | | | |
| IT services major | −0.66 | < 0.001 | 88.75% of 80 | 38.69% of 199 | 59 | −0.74 | −0.84 to −0.60 | 1.000 |
| BPO specialist | −0.70 | < 0.001 | 90.77% of 65 | 38.53% of 109 | 37 | −0.80 | −0.89 to −0.64 | 1.000 |
| Consulting firm | −0.58 | < 0.001 | 88.46% of 26 | 35.53% of 76 | 23 | −0.71 | † | 0.995 |
| Boutique or specialist provider | −0.73 | < 0.001 | 83.33% of 24 | 40.82% of 49 | 17 | −0.87 | † | 1.000 |
| Pure-play agentic AI firm | −0.64 | < 0.001 | 91.30% of 23 | 21.43% of 42 | 14 | −0.76 | † | 0.988 |
| Not reported, recovered by engagement | −0.55 | < 0.001 | 81.25% of 16 | 45.45% of 33 | 42 | −0.56 | −0.74 to −0.30 | 0.976 |

Note: † Base under 30. No interval is reported. ‡ Base under 10. No statistic is reported.

Respondents who gave higher ratings reported lower intent to leave (r = −0.66, 95% CI −0.69 to −0.61, Holm p < 0.001, n = 742 respondents). The association held after Holm correction in thirty-six of the 37 segment levels tested.

## Association of Written Sentiment with the Study Measures

Table 14. Written sentiment and the study measuresTable

| Comparison | With a negative response | Without a negative response | Test | Statistic | n | p |
| --- | --- | --- | --- | --- | --- | --- |
| Cancellation rate, engagements | 0 of 6 | 56 of 144 (38.89%) | Fisher exact | | 150 | 0.084 |
| Customer satisfaction, active engagements | n = 6 | 63.93%, n = 88 | ‡ | | 94 | |
| Engagement mean written score and cancellation, engagements | | | Point-biserial r | 0.39 | 14 | 0.168 |
| Respondent mean written score and intent to leave, respondents | | | Spearman rho | 0.04 | 16 | 0.890 |

Note: ‡ Base under 10. No statistic is reported.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 144 other engagements, 136 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 6 engagements with a negative response, 0 were canceled, against 56 of the 144 other engagements (38.89%). The difference did not reach significance (Fisher exact p = 0.084). The comparison of customer satisfaction was not computed, because one of the two groups held too few active engagements.

Two correlations test written sentiment directly. Across the 14 engagements with a written response, the engagement mean written score was not associated with cancellation (point-biserial r 0.39, p = 0.168). Across 16 respondents, the respondent mean written score was not associated with intent to leave (Spearman rho 0.04, p = 0.890). Of the three comparisons, none reached significance at 0.05.

## Comparison of Associations Across the Four Measures

Table 15. Associations across the four measuresTable

| Measure | Form | Unit | n | r | 95% CI | p | Equivalent to zero | Rank among the four |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Cancellation rate | Descriptive contrast | Engagement | 150 | −0.66 | −0.74 to −0.56 | < 0.001 | No | 2 |
| Customer satisfaction | Clean estimate | Engagement | 94 | 0.65 | 0.52 to 0.76 | < 0.001 | No | 4 |
| Engagement indicator mean | Leave-one-out | Engagement | 150 | 0.86 | 0.82 to 0.90 | < 0.001 | No | 1 |
| Intent to leave | Respondent level, clustered | Respondent | 742 | −0.66 | −0.71 to −0.60 | < 0.001 | No | 3 |

Table 16. Pairwise comparison of associationsTable

| Pair compared | Difference in r | Steiger z | p | Holm p | Reading |
| --- | --- | --- | --- | --- | --- |
| Cancellation rate and customer satisfaction | −0.11 | −5.39 | < 0.001 | < 0.001 | Differed |
| Cancellation rate and engagement indicator mean | −0.20 | −6.03 | < 0.001 | < 0.001 | Differed |
| Cancellation rate and intent to leave | −0.10 | −2.78 | 0.005 | 0.011 | Differed |
| Customer satisfaction and engagement indicator mean | −0.10 | −4.17 | < 0.001 | < 0.001 | Differed |
| Customer satisfaction and intent to leave | 0.01 | 0.45 | 0.651 | 0.651 | Did not differ |
| Engagement indicator mean and intent to leave | 0.11 | 4.37 | < 0.001 | < 0.001 | Differed |

The four associations were compared in pairs on their engagement-level forms, each oriented so that a higher value is the better outcome, across the 150 engagements with all four measures. After Holm correction, five of the six pairs differed. The association with the engagement indicator mean was the strongest and the association with customer satisfaction the weakest. The associations with customer satisfaction and intent to leave did not differ (Holm p = 0.651).

## Clustering and Effective Sample Size

Table 17. Clustering and effective sample sizeTable

| Statistic | Value | Reading |
| --- | --- | --- |
| Clusters (engagements) | 150 | Engagements with one or more answers. |
| Mean respondents per engagement | 5.2 | Answers per engagement, among engagements with one or more answers. |
| ICC(1) | 0.61 | Share of the variance that lies between engagements. |
| Design effect | 3.57 | One plus the product of the ICC and the mean cluster size less one. |
| Nominal respondent n | 787 | Respondents who answered. |
| Effective n | 220 | Nominal n divided by the design effect. |
| Cluster bootstrap applied | Yes | Applied to every respondent-level interval where the ICC is 0.05 or above. |

Answers clustered within engagements (ICC 0.61). Clustering reduced the 787 respondent answers to the equivalent of about 220 independent answers. Respondent-level intervals therefore used the cluster bootstrap. Respondent-level tests of intent to leave used standard errors that allow for clustering by engagement.

## Equivalence Tests of Null Results

Table 18. Equivalence tests of null resultsTable

| Measure | n | r | Equivalence bound | TOST p, lower | TOST p, upper | Equivalent to zero | Achieved power | MDE at power 0.80 | Detection floor for this n |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Cancellation rate | 150 | −0.66 | \|r\| < 0.30 | 1.000 | < 0.001 | No | 0.96 | 0.23 | 0.23 |
| Customer satisfaction | 94 | 0.65 | \|r\| < 0.30 | < 0.001 | 1.000 | No | 0.84 | 0.29 | 0.23 |
| Engagement indicator mean | 150 | 0.86 | \|r\| < 0.30 | < 0.001 | 1.000 | No | 0.96 | 0.23 | 0.23 |
| Intent to leave | 742, effective 207 | −0.66 | \|r\| < 0.30 | 1.000 | < 0.001 | No | 0.99 | 0.19 | 0.23 |

No association was equivalent to zero. Each of the four differed from zero. The detection floor shown is the study floor at 150 engagements. Intent to leave was tested on its effective sample size.

## Robustness Checks

Table 19. Robustness checksTable

| Check | Measure | Primary r | Check r | 95% CI | Change | Conclusion holds |
| --- | --- | --- | --- | --- | --- | --- |
| Rank correlation in place of Pearson | Cancellation rate | −0.66 | −0.69 | a | −0.03 | Yes |
| Rank correlation in place of Pearson | Customer satisfaction | 0.65 | 0.67 | a | 0.01 | Yes |
| Rank correlation in place of Pearson | Engagement indicator mean | 0.86 | 0.86 | a | 0.00 | Yes |
| Rank correlation in place of Pearson | Intent to leave | −0.66 | −0.66 | a | 0.00 | Yes |
| Non-answers excluded, then imputed at the engagement mean | Intent to leave | −0.66 | −0.66 | a | 0.00 | Yes |
| Active engagements only | Engagement indicator mean | 0.86 | 0.79 | 0.69 to 0.85 | −0.08 | Yes |
| Active engagements only | Intent to leave | −0.66 | −0.53 | −0.59 to −0.46 | 0.13 | Yes |
| Most influential engagement removed (largest Cook's distance) | Cancellation rate | −0.66 | −0.67 | −0.75 to −0.58 | −0.02 | Yes |
| Most influential engagement removed (largest Cook's distance) | Customer satisfaction | 0.65 | 0.67 | 0.54 to 0.77 | 0.02 | Yes |
| Most influential engagement removed (largest Cook's distance) | Engagement indicator mean | 0.86 | 0.87 | 0.82 to 0.90 | 0.01 | Yes |
| Most influential engagement removed (largest Cook's distance) | Intent to leave | −0.66 | −0.66 | −0.70 to −0.62 | 0.00 | Yes |
| Engagements weighted by respondent count | Cancellation rate | −0.66 | −0.66 | a | 0.00 | Yes |
| Engagements weighted by respondent count | Customer satisfaction | 0.65 | 0.66 | a | 0.01 | Yes |
| Engagements weighted by respondent count | Engagement indicator mean | 0.86 | 0.86 | a | 0.00 | Yes |
| Bootstrap of engagements | Cancellation rate | −0.66 | −0.66 | −0.73 to −0.58 | 0.00 | Yes |
| Bootstrap of engagements | Customer satisfaction | 0.65 | 0.65 | 0.54 to 0.75 | 0.00 | Yes |
| Bootstrap of engagements | Engagement indicator mean | 0.86 | 0.86 | 0.82 to 0.90 | 0.00 | Yes |
| Bootstrap of engagements | Intent to leave | −0.66 | −0.65 | −0.71 to −0.60 | 0.00 | Yes |
| Thin-base segments removed | Cancellation rate | −0.66 | −0.66 | −0.77 to −0.52 | −0.01 | Yes |
| Thin-base segments removed | Customer satisfaction | 0.65 | 0.56 | 0.34 to 0.72 | −0.09 | Yes |
| Thin-base segments removed | Engagement indicator mean | 0.86 | 0.84 | 0.76 to 0.89 | −0.02 | Yes |
| Thin-base segments removed | Intent to leave | −0.66 | −0.66 | −0.71 to −0.60 | 0.00 | Yes |

Note: a No interval for this check.

Each change was computed before rounding and may differ by 0.01 from the difference between the two printed values.

All 22 checks held. A check holds when its test falls on the same side of 0.05 as the primary test. Where the primary test falls below 0.05, the check must also keep the sign of the primary correlation. The thin-base check removes the engagements in any level of outsourcing model or process family with fewer than 30 engagements. Institution type is not used for that check, because every one of its levels has fewer than 30 engagements.

## Measure Independence and Part-Whole Correction

Each of the four measures was computed from data the indicator does not contain. Each result is therefore a finding and not an arithmetic consequence. The indicator is one of the rated indicators, and the engagement indicator mean was recomputed without it. The correction changed r by 0.00, from 0.87 to 0.86.

## Comparison with the Study Norm

Table 20. Comparison with the study normTable

| Measure | The indicator, oriented r | Study median, oriented r | Deviation | Fisher z | p | Holm p |
| --- | --- | --- | --- | --- | --- | --- |
| Cancellation rate | 0.66 | 0.60 | 0.06 | 1.22 | 0.221 | 0.866 |
| Customer satisfaction | 0.77 | 0.72 | 0.05 | 1.24 | 0.216 | 0.866 |
| Engagement indicator mean | 0.86 | 0.84 | 0.03 | 1.17 | 0.243 | 0.866 |
| Intent to leave | 0.75 | 0.71 | 0.04 | 1.13 | 0.258 | 0.866 |
| Spread of deviations | | | 0.03 | | | |
| Verdict | Typical | | | | | |

Each deviation was computed before rounding and may differ by 0.01 from the difference between the two printed values.

The comparison covered 150 engagements. The indicator followed the study's typical pattern on all four measures, and no deviation held after Holm correction.

## Comparison of Written Sentiment with Other Part 2 Indicators

Figure 6. Share of negative written responses, Part 2 indicators with 30 or more written responses (n = 71 indicators)ExhibitNote: Each column is one indicator, ordered from the highest share of negative responses. The indicator is the amber column. The dashed line marks the median share, 80.82%.

Among the 71 Part 2 indicators with 30 or more written responses, the indicator's share of negative responses, 41.18%, ranked 71st from the highest, against a median of 80.82%. Its mean written score, 1.2, ranked 71st from the most negative, against a median of −4.7.

## Co-Occurring Indicators and Themes

Figure 7. Other indicators and themes in the same written responses (n = 17 responses)ExhibitNote: Themes carry no score. Labels give the count and the share of the 17 responses. The figure shows each indicator and each theme among the five most frequent of its kind, with every tie at the cut. C168: “Rate how consistently the agents took only intended and authorized actions when operating connected tools and systems.” C157: “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.” C160: “Rate the end-to-end reliability of the tools on multi-step tasks, from first step to finished output.” C175: “Rate the stability of the tools' behavior over the engagement when nothing changed on the engagement side.” C164: “Rate how reliably the tools retained instructions and constraints given earlier in long tasks and sessions.” C169: “Rate how safely the tools handled sensitive data, with no exposure through outputs, logs, or connected systems.” Emotional states in the same responses, as counts: .

The indicator most often present in the same responses was C168, “Rate how consistently the agents took only intended and authorized actions when operating connected tools and systems.”, in 5 of the 17 responses (29.41%). C157, “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.” (4 responses), C160, “Rate the end-to-end reliability of the tools on multi-step tasks, from first step to finished output.” (4 responses), C175, “Rate the stability of the tools' behavior over the engagement when nothing changed on the engagement side.” (4 responses) and C164, “Rate how reliably the tools retained instructions and constraints given earlier in long tasks and sessions.” (3 responses) followed.

The most frequent theme was “Occurs every cycle or most days”, in 1 responses (5.88%). The second was “Effort is invisible in the numbers”, in 1 responses.

## Interpretation of the Data

About a third of all ratings of how reliably the built-in guardrails held fell in the distressed range (252 of 787 answers, 32.02%). The rating was associated with all four study measures, and every association held under each of the 22 robustness checks. Canceled engagements rated far below the three active stages (mean 3.2, n = 292 respondents).

The written responses were few and leaned positive, with a negative share of 41.18% against a Part 2 median of 80.82% among indicators with 30 or more written responses. The ratings and the written evidence therefore point in different directions, and the ratings, on the full base, carry the weight. Where respondents did describe a failure, the control had not anticipated the situation, such as a hardship case or a loose instruction, and a person found the problem.

## Limitations

The evidence is the reported experience of the vendor's delivery teams, collected in a single cross-section. The associations reported here are not causal.

Tests by respondent role and by vendor type treated respondents as independent. The section on clustering reports how far clustering within engagements reduced the information in respondent-level figures.

Customer management executives gave 2 answers, a base of anecdote strength.

Shares of written sentiment are shares of responses and not of the study set of respondents. The written score takes five fixed values, so its mean and median are coarse. Responses clustered within engagements (17 responses on 14 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.

The written responses number fewer than 30, so their shares and the written-sentiment tests rest on a small base.

## Definitions

Table 21. DefinitionsTable

| Term | Definition |
| --- | --- |
| Response scale | Ratings run from 1 to 10, from the worst outcome a respondent can report to the best. |
| Distressed | A rating of 3 or below. At engagement level, an engagement mean of 3 or below. |
| Answering base | Respondents who gave a rating from 1 to 10. |
| Non-answers | “I do not know” and “I prefer to not answer” are counted separately and excluded from the mean. |
| Indicator class | Respondent level: a value recorded for each respondent and computed across the respondents who answered. |
| Failure category | One of the five failure categories of the Study Taxonomy. The indicator belongs to AI technology. |
| Engagement indicator mean | The mean of an engagement's ratings on the rated indicators. The indicator is one of the rated indicators, so the mean is recomputed without it wherever the two are compared. |
| Written response | A written answer about the respondent's own engagement that addresses the indicator. Non-answers and answers about another engagement are excluded. |
| Primary and secondary | A primary response answered the written-answer question of the indicator's survey section. A secondary response answered a question in another section and addressed the indicator in its content. |
| Written sentiment score | Each written response carries one of five scores: −10, −5, neutral, +5 or +10. The sign records whether the response described the state of the indicator as harmful or helpful to the engagement. The magnitude records its strength. A score of −10 marks a statement of totality, constancy or serious harm. |
| Cluster bootstrap | An interval computed by resampling engagements rather than responses, so that the responses from one engagement stay together. |

## Sources and Provenance

Table 22. Sources and provenanceTable

| Source | Value |
| --- | --- |
| Study data | Study Data.xlsx, Data sheet, column C170, sha256 219eba0457274d04 |
| Statistics engine | compute_statistics.py, sha256 738319eef764602b |
| Statistics record | Computed Statistics.json, record C170, sha256 1ae50a3637fc9370 |
| Computed | October 13, 2026, 04:57 UTC |
| Template | Indicator Template v2, version 2.0 of 10 October 2026 |
| Figures placed by script | 3,048 |
| Written responses | Quotes Master.jsonl, sha256 27da38085e2b13fd |
| Written sentiment scores | Quote Coding Rulebook, version 0.6, with section 21 |
| Written-response computation | qa_stats.py, sha256 74a61f3a96169da9. wfacts.py, sha256 b85fae20adf812a1. build_indicator.py, sha256 a97ad6e74627e11e |
| Written-response records | C170.qa.json, sha256 4b08f4d7717cc3ab. C170.wfacts.json, sha256 b768fe38637fbc04. Hand-written text handcraft/C170.json, sha256 b552ee92e4af223d |