Rate the stability of the third-party APIs and data structures the agents depend on, from one provider release to the next

The mean rating was 4.7 and the median 5.0 (n = 796 respondents). Ratings of 3 or below, the distressed range, made up 30.53% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.65, 95% CI −0.73 to −0.54, 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 0.81 to 0.89, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.65, cluster bootstrap 95% CI −0.71 to −0.58, p < 0.001 with standard errors clustered by engagement, n = 747 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 37.55, Holm p < 0.001, omega squared 0.404), respondent role (Welch F 8.25, Holm p = 0.003, omega squared 0.020) and vendor type (Welch F 3.40, Holm p = 0.028, omega squared 0.017).

Of the 106 written responses, nearly all from C198, the comment field on the third-party ecosystem, 90 were negative and 15 positive. Every negative response described a provider changing a file layout, field, interface, page or model version that the agents depended on, and several said it happened with each release or each cycle, leaving the team to remap or re-enter data by hand. No written outcome test reached significance.

Ratings of 3 or below, the distressed range, accounted for 243 of the 796 answers (30.53%). The engagement mean fell in the distressed range in 36 of 150 engagements.

The primary responses numbered 96, all written in answer to C198, “Add any comments on the third-party ecosystem that the ratings in this section do not capture.” The remaining 10 responses, 9.43% of the total, were secondary and came from 7 other questions. That secondary share ranked 45th 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 1. Negative responses made up 84.38% of primary responses (81 of 96) and 90.00% of secondary responses (9 of 10).

The negative responses described renamed fields, shifted or added columns, changed date formats and decimals, restyled portal pages, changed status codes and retired model versions. Nine said the change came every cycle, week or release, or nearly so: one respondent said a market data feed changed shape about once a quarter officially and about once a month in practice. Twelve responses described re-entering, remapping or cleaning data by hand as a result, in one case rebuilding a mapping in successive weeks when the provider renamed a field and then restored the old name. Respondents on the questions about altering data and processes described a vendor component that broke the mapping most releases, and follow-ups to the provider that went unanswered.

Three responses placed part of the instability on the engagement side: no one had set out what a clean file looked like, no one asked the custodians for a format suited to a machine reader, and no one owned the integration when the provider changed its interface version. Several respondents added that the people who used to catch these changes by hand had left without writing their checks down.

The fifteen positive responses included six that described interfaces or feeds that had been stable, one card scheme feed for fourteen months without an unannounced change, while noting the absence of a fallback, an unmapped chain of sub-providers or weak notice.

Of the 106 responses, 90 (84.91%, 95% CI 77.42% to 91.67%) were negative and 12 (11.32%) carried the strongest negative score of −10. Positive responses numbered 15 (14.15%), and 1 was neutral. The mean written score was −3.8 (95% CI −4.7 to −2.9, n = 106 responses), and the median was −5.0.

Of the 106 responses, 97 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.29, 95% CI 0.05 to 0.51, p = 0.005, n = 97 responses). The written responses corroborated the rating. Of the 81 negative responses with a rating, 35 (43.21%) came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 97, with a mean rating of 4.3. The 699 respondents who rated the indicator without writing about it had a mean of 4.7. The difference, −0.4 (95% CI −1.0 to 0.1), included zero. Ratings in the distressed range made up 40.21% of the ratings from respondents who wrote (39 of 97) and 29.18% of the ratings from respondents who did not (204 of 699).

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 37.55, Holm p < 0.001, omega squared 0.404). Active engagements in Transition/Cut-Over rated highest (mean 5.7, n = 237 respondents) and canceled engagements lowest (mean 3.1, n = 297 respondents). Ratings also differed by respondent role (Welch F 8.25, Holm p = 0.003, omega squared 0.020). Portfolio directors rated highest (mean 5.9, n = 31 respondents) and individual contributors lowest (mean 4.5, n = 613 respondents). Ratings also differed by vendor type (Welch F 3.40, Holm p = 0.028, omega squared 0.017). Pure-play agentic AI firms rated highest (mean 5.3, n = 72 respondents) and BPO specialists lowest (mean 4.2, n = 180 respondents). Outsourcing model, institution type, process family and class of work showed no difference.

Written sentiment did not differ by engagement status and lifecycle stage across the 73 engagements with a written response (Kruskal-Wallis H 2.14, df 3, p = 0.544). Of the groups with 10 or more responses, responses from active engagements in Production were the most negative (mean −4.5, 10 responses) and responses from active engagements in Build-Out the least negative (mean −1.9, 18 responses).

No test across respondent roles was run, because fewer than two roles had 30 or more written responses. Individual contributors wrote 81, engagement managers wrote 21 and portfolio directors wrote 4 responses.

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 = −0.65, 95% CI −0.73 to −0.54, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in thirteen 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 = 0.65, 95% CI 0.52 to 0.76, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in thirteen 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 = 0.86, 95% CI 0.81 to 0.89, 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 = −0.65, 95% CI −0.69 to −0.60, Holm p < 0.001, n = 747 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 88 other engagements, 77 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 62 engagements with a negative response, 26 (41.94%) were canceled, against 30 of the 88 other engagements (34.09%). The difference did not reach significance (Fisher exact p = 0.392). Customer satisfaction among active engagements did not differ between the two groups (mean 63.78% on 36 engagements against 64.29% on 58, Welch t −0.21, p = 0.833).

Two correlations test written sentiment directly. Across the 73 engagements with a written response, the engagement mean written score was not associated with cancellation (point-biserial r −0.14, p = 0.239). Across 102 respondents, the respondent mean written score was not associated with intent to leave (Spearman rho −0.15, p = 0.126). Of the four comparisons, none 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 intent to leave the weakest. The associations with customer satisfaction and intent to leave did not differ (Holm p = 0.491).

Answers clustered within engagements (ICC 0.67). Clustering reduced the 796 respondent answers to the equivalent of about 204 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 0.86 to 0.86.

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, 84.91%, ranked 27th from the highest, against a median of 80.82%. Its mean written score, −3.8, ranked 51st from the most negative, against a median of −4.7.

The indicator most often present in the same responses was C181, “Rate the quality of advance notice third-party providers give before changes that can affect the agents on this engagement.”, in 64 of the 106 responses (60.38%). C178, “Rate how well the agents on this engagement are protected from breaking silently when a third party changes something without warning.” (52 responses), C183, “Rate the engagement's monitoring for unannounced changes in the third-party APIs, data feeds, and models its agents depend on.” (20 responses), C182, “Rate how reliably word of an upcoming third-party change reaches the people who operate the affected agents in time to act.” (17 responses) and C179, “Rate the quality of the fallback the engagement has ready when a critical third-party feed, platform, or model stops working.” (16 responses) followed.

The most frequent theme was “Detection only by failure”, in 46 responses (43.40%). The second was “Schema or format changed without notice”, in 44 responses.

Close to a third of all ratings of the stability of third-party interfaces and data structures fell in the distressed range (243 of 796 answers, 30.53%). 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 3.1, n = 297 respondents).

The written evidence shows that the agents depended on inputs that changed on the providers' schedule, and several of those inputs were built for human readers. Each change became manual work for the delivery team, and much of that work was invisible outside it. The positive accounts show that some providers held their interfaces steady for a year or more.

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 (106 responses on 73 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.