Rate how well joint human review capacity keeps up with the volume of output the agents produce

The mean rating was 3.7 and the median 4.0 (n = 805 respondents). Ratings of 3 or below, the distressed range, made up 47.33% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.62, 95% CI −0.71 to −0.52, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.59, 95% CI 0.43 to 0.70, p < 0.001, n = 94 engagements), a higher engagement indicator mean computed without the indicator (r = 0.82, 95% CI 0.76 to 0.87, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.67, cluster bootstrap 95% CI −0.72 to −0.60, p < 0.001 with standard errors clustered by engagement, n = 755 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 39.47, Holm p < 0.001, omega squared 0.390) and respondent role (Welch F 10.87, Holm p < 0.001, omega squared 0.022).

Written responses addressing the indicator numbered 145, from 135 respondents on 89 engagements. Of these, 134 (92.41%) were negative. More than half came from C029, where human review was named as the bottleneck of the engagement in nearly the same words each time. Respondents described agents that produced more in a morning than two or four people could read, approvals released without reading, and analysts whose work became checking. Responses to the question on manipulated reports described dashboards that hid the review backlog. Respondents whose written responses were more negative reported higher intent to leave.

Ratings of 3 or below, the distressed range, accounted for 381 of the 805 answers (47.33%). The engagement mean fell in the distressed range in 65 of 150 engagements.

The primary responses numbered 53, all written in answer to C156, “Add any comments on the joint management of this engagement that the ratings in this section do not capture.” The remaining 92 responses, 63.45% of the total, were secondary and came from 6 other questions. That secondary share ranked 14th from the highest among the 71 Part 2 indicators with 30 or more written responses, against a median of 18.92%. Respondents who wrote both a primary and a secondary response numbered 2. Negative responses made up 79.25% of primary responses (42 of 53) and 100.00% of secondary responses (92 of 92).

All 77 responses to C029 named human review, oversight, validation, sign-off or the human-in-the-loop step among the challenges that reduce job satisfaction, most often as the bottleneck of the engagement, one that no one had planned for, staffed or fixed.

Among the 53 responses to C156, the comment field on joint management, 28 negative responses stated that agent output exceeded what the reviewers could read, or that review rested on one or two people: four hundred exceptions in one night for two reviewers, one reviewer for forty provisioning outputs a day, a month-end validation queue still unfinished two weeks into the next month, eleven thousand records in a payroll cycle checked by sample. Five described approvals clicked through because of volume, including a reviewer who opened a queue of forty and released thirty-eight. Eight responses described the respondent's own work turned into review: reading generated test scripts instead of building, checking one's own models in the evening, engineers reading generated code until nine at night, and a formal check that sampled about two percent of output.

Against these, seven positive responses said review capacity kept pace or had improved, in one case after the two firms split the review roster so that seniors could sample. Two more said capacity was adequate only because volume was still small. The eight responses to C055, the question on how the vendor manipulates performance reports and dashboards, described dashboards that hid the review backlog: closure time measured from the agent's triage, a pilot-phase review time reported after volume had tripled, records counted as reviewed once opened, and throughput rolled up before the review queue was counted.

Of the 145 responses, 134 (92.41%, 95% CI 87.97% to 96.35%) were negative and 37 (25.52%) carried the strongest negative score of −10. Positive responses numbered 11 (7.59%), and no response was neutral. The mean written score was −5.4 (95% CI −6.1 to −4.7, n = 145 responses), and the median was −5.0.

Of the 145 responses, 138 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.28, 95% CI 0.09 to 0.45, p < 0.001, n = 138 responses). The written responses corroborated the rating. Of the 127 negative responses with a rating, 67 (52.76%) came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 130, with a mean rating of 3.7. The 675 respondents who rated the indicator without writing about it had a mean of 3.7. The difference, 0.0 (95% CI −0.5 to 0.4), included zero. Ratings in the distressed range made up 49.23% of the ratings from respondents who wrote (64 of 130) and 46.96% of the ratings from respondents who did not (317 of 675).

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.

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 39.47, Holm p < 0.001, omega squared 0.390). Active engagements in Build-Out rated highest (mean 4.7, n = 166 respondents) and canceled engagements lowest (mean 2.2, n = 304 respondents). Ratings also differed by respondent role (Welch F 10.87, Holm p < 0.001, omega squared 0.022). Portfolio directors rated highest (mean 5.2, n = 29 respondents) and individual contributors lowest (mean 3.5, n = 624 respondents). Outsourcing model, institution type, process family, class of work and vendor type showed no difference.

Written sentiment did not differ by engagement status and lifecycle stage across the 89 engagements with a written response (Kruskal-Wallis H 2.54, df 3, p = 0.467). Of the groups with 10 or more responses, responses from canceled engagements were the most negative (mean −6.3, 43 responses) and responses from active engagements in Production the least negative (mean −4.5, 22 responses).

The role test compared individual contributors and engagement managers, the roles with 30 or more responses. Written sentiment did not differ (Kruskal-Wallis H 1.30, p = 0.255, n = 132 respondents): the mean written score was −5.7 for individual contributors (109 responses) and −4.5 for engagement managers (33 responses). Portfolio directors wrote 3 responses, reported as counts only.

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.

Across the study set, higher ratings came with less cancellation (r = −0.62, 95% CI −0.71 to −0.52, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in sixteen of the 20 segment levels tested.

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

Across the active engagements, higher ratings came with higher customer satisfaction (r = 0.59, 95% CI 0.43 to 0.70, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in nine of the 16 segment levels tested.

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.

Higher ratings came with a higher engagement indicator mean (r = 0.82, 95% CI 0.76 to 0.87, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in twenty-four of the 24 segment levels tested.

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.

Respondents who gave higher ratings reported lower intent to leave (r = −0.67, 95% CI −0.70 to −0.62, Holm p < 0.001, n = 755 respondents). The association held after Holm correction in thirty-six of the 37 segment levels tested.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 66 other engagements, 61 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 84 engagements with a negative response, 28 (33.33%) were canceled, against 28 of the 66 other engagements (42.42%). The difference did not reach significance (Fisher exact p = 0.308). Customer satisfaction among active engagements did not differ between the two groups (mean 64.05% on 56 engagements against 64.16% on 38, Welch t −0.04, p = 0.967).

Two correlations test written sentiment directly. Across the 89 engagements with a written response, the engagement mean written score was not associated with cancellation (point-biserial r −0.18, p = 0.097). Across 128 respondents, a more negative respondent mean written score came with higher intent to leave (Spearman rho −0.23, p = 0.008). Of the four comparisons, one reached significance at 0.05.

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 = 0.620).

Answers clustered within engagements (ICC 0.69). Clustering reduced the 805 respondent answers to the equivalent of about 201 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.

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.

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.

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 0.83 to 0.82.

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.

Among the 71 Part 2 indicators with 30 or more written responses, the indicator's share of negative responses, 92.41%, ranked 8th from the highest, against a median of 80.82%. Its mean written score, −5.4, ranked 25th from the most negative, against a median of −4.7.

The indicator most often present in the same responses was C136, “Rate how effectively the customer and vendor jointly manage this engagement now that agents perform part of the work.”, in 19 of the 145 responses (13.10%). C149, “Rate how much genuine scrutiny agent outputs receive at human review points rather than routine approval.” (16 responses), C147, “Rate how well human review points are placed across the two firms at the moments where agent errors would matter most.” (15 responses), C098, “Rate how well the customer's legacy systems support the integration this engagement requires.” (11 responses) and C157, “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.” (11 responses) followed.

The most frequent theme was “Liability or regulatory accountability unclear”, in 6 responses (4.14%). The second was “The scale or question does not fit”, in 6 responses.

Nearly half of all ratings of how well joint human review capacity keeps up with agent output fell in the distressed range (381 of 805 answers, 47.33%). 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 2.2, n = 304 respondents).

The written responses describe a control sized for the work people produce, applied to the volume agents produce. Review either fell behind, became a signature, or moved to the analysts who built the output. The reports described in answer to C055 show that the backlog was in some engagements removed from the dashboards the institution read.

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 (145 responses on 89 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.