What is the outsourcing type?

Respondents named ITO most often as the outsourcing type (244 of 849 respondents, 28.74%) and KPO least often (165 of 849 respondents, 19.43%). Cancellation ranged from CXM at the low end (35.71% canceled, n = 42 engagements) to KPO at the high end (38.71% canceled, n = 31 engagements), with no difference under test (chi-square 0.08, p 0.994, Cramer’s V 0.02, n = 150 engagements). Customer satisfaction on active engagements ranged from KPO at the low end (mean 60.6, n = 19 engagements) to BPO at the high end (mean 66.3, n = 23 engagements), with no difference under test (Welch F 0.78, p 0.510, omega squared −0.004, n = 94 engagements). The engagement indicator mean was lowest for BPO (mean 3.7, n = 37 engagements) and highest for CXM (mean 4.2, n = 42 engagements), with no difference under test (Welch F 0.90, p 0.444, omega squared −0.003, n = 150 engagements). Intent to leave was lowest for CXM (mean 6.5, n = 42 engagements) and highest for ITO (mean 7.1, n = 40 engagements), with no difference under test (Welch F 0.47, p 0.703, omega squared −0.011, n = 150 engagements). The answer mix differed after Holm control by process family (chi-square 121.20, Holm p < 0.001, Cramer’s V 0.57) and by vendor type (chi-square 536.53, Holm p < 0.001, Cramer’s V 0.46).

One written response was coded to the indicator, from 1 of 850 respondents on 1 of 150 engagements. The single scored response was negative, with a score of −5. It was a secondary response, written in answer to a question outside the indicator’s own survey section.

A categorical answer has no engagement mean. Each engagement took its respondents’ plurality answer, and tied engagements were left out. An answer was assigned to 150 of 150 engagements.

Distressed is not applicable, because the answer has no good or bad end on a scale.

The single response coded to the indicator was a secondary response. The response was written in answer to C135, a question outside the indicator’s own survey section.

The indicator has no good or bad end. Under the coding rulebook’s rule for such indicators, the written score records the effect each response states on the engagement or on the respondent. The survey question is “What is the outsourcing type?” The frequency, scope and consequence a response states set the strength of its score. A score of +10 marks a response that describes “this attribute of the engagement as helping the engagement at its best.” A score of −10 marks a response that describes “this attribute as harming the engagement at its worst . . . .”

The distribution was reported as counts only, because fewer than 10 responses carried a score. The single scored response carried a score of −5.

Shares are of the answered respondents in each row, which is the base beside them. A row under 10 prints counts in place of shares.

Engagement families were tested on engagement answers, and respondent role and vendor type were tested on respondents. Process family differed in the answer mix after Holm control (chi-square 121.20, Holm p < 0.001, Cramer’s V 0.57). Vendor type differed in the answer mix after Holm control (chi-square 536.53, Holm p < 0.001, Cramer’s V 0.46). No difference held after Holm control for institution type, class of work, engagement status and lifecycle stage and respondent role.

The share of engagements canceled on the study date, measured at engagement level (base 150 engagements, 56 canceled). The association is a descriptive contrast, inflated because canceled engagements were rated after cancellation, and is not predictive.

The measure by the engagement’s assigned answer.

The test across answer levels, in the study set and inside each segment. Levels under 10 engagements were left out of each test.

Cancellation in the study set did not differ across answer levels (chi-square 0.08, p 0.994, n = 150 engagements).

The satisfaction score recorded against each engagement on a 100-point scale. The primary estimate uses the 94 active engagements. A descriptive contrast uses all 150 engagements.

The measure by the engagement’s assigned answer.

The test across answer levels, in the study set and inside each segment. Levels under 10 engagements were left out of each test.

The measure in the study set did not differ across answer levels (Welch F 0.78, p 0.510, n = 94 engagements). The clean estimate ran inside active engagements only.

The mean of an engagement’s 100 indicator ratings, measured at engagement level (base 150 engagements). Where this indicator is one of the 100, the mean is recomputed without it.

The measure by the engagement’s assigned answer.

The test across answer levels, in the study set and inside each segment. Levels under 10 engagements were left out of each test.

The measure in the study set did not differ across answer levels (Welch F 0.90, p 0.444, n = 150 engagements).

The respondent’s stated likelihood of leaving the current job within 12 months, rated 1 to 10 (C019). Measured at respondent level on up to 798 respondents, with the engagement-level form reported alongside.

The measure by the engagement’s assigned answer.

The test across answer levels, in the study set and inside each segment. Levels under 10 engagements were left out of each test.

The measure in the study set did not differ across answer levels (Welch F 0.47, p 0.703, n = 150 engagements).

Steiger’s z compares two correlations, and a categorical answer yields an effect size across levels rather than a correlation, so the pairs are not tested.

Clustering reduced the 849 respondent answers to about 150 independent answers. Respondent level intervals therefore used the cluster bootstrap, and respondent level tests on intent to leave used cluster-robust errors.

The bound is the ruled |r| < 0.30 expressed as variance explained (V15-R22). Equivalent to zero: cancellation rate, customer satisfaction, engagement indicator mean and intent to leave.

Every check held. A check holds when its test lands on the same side of 0.05 as the primary test.

Each of the four measures was computed from data this indicator does not contain, so each result is a finding rather than an arithmetic consequence.

Not applicable. A categorical item has no signed correlation, so it has no profile. Its per-level means carry the comparison.

The indicator’s twin, ENG:Outsourcing Type, records the same fact under the coding rulebook’s twin rule. The twin was coded in the single response coded to the indicator. The twin is left out of the counts that follow. The response coded to the indicator was also coded to “Rate your employer’s current supply of qualified agentic AI talent against what its engagements need” (C122, 1 response), “Rate how well your employer’s training programs build financial industry knowledge” (C125, 1 response) and “Rate the depth of your employer’s knowledge of agentic AI” (C131, 1 response). The theme “Human presented as agentic” of the coding rulebook was the only theme in the same response.

Every robustness check run on this item held (9 of 9 checks). Outsourcing type sorted engagements by process family and vendor type, and it showed no association with the four study measures.

The indicator is a survey question that is not a rating, so the written responses were not set against a rating. Written responses were voluntary. They recorded what the respondents who wrote reported, and how common it was is measured by the answers. The analysis of the written responses rested on 1 response, a base of anecdote strength.

The evidence is reported experience: what the delivery team said about its own engagement, gathered in one cross-section. The figures showed associations, not causes. Indicators about the financial institution were reported at second hand by vendor staff. Cancellation is a descriptive contrast, because canceled engagements were rated after they ended.

Tests across respondent role and vendor type were run at respondent level and treated respondents as independent. The section Clustering and Effective Sample Size states whether, and by how much, clustering within engagements reduced the information in respondent level figures.

Each engagement took its respondents’ plurality answer for the measure comparisons (150 of 150 engagements).

Fewer than 10 responses carried a score, so written sentiment was reported as counts, with no statistic. Coding was performed by AI coders under a fixed rulebook. No response coded to the indicator fell in the sample second-coded under rulebook 0.5, so agreement on whether a response was coded to the indicator could not be assessed. No response coded to the indicator was second-scored under rulebook 0.6, so agreement on the score could not be assessed.