Financial Institution Readiness and the Governance of Agentic Work

Financial institution readiness was closely associated with the reported control environment. Low ratings were common across the institution's data governance, management, access, accountability, and oversight responsibilities. The governance, oversight, and control score links those responsibilities with controls elsewhere in the engagement. Conditional cross-category results were less consistent, with sign changes under alternative comparison means. These qualifications limit distinct-effect claims without removing the institution's substantive responsibilities from the analysis.

Readiness indicators correlated with the score excluding the tested indicator at r = 0.75–0.80, and with governance in the other four categories at r = 0.72–0.76 (n = 150). Five adjusted cross-category estimates were small, negative, and nonsignificant, from −0.15 to −0.11. Response speed was negative and nominally significant in the main model but sensitive to category removal. Data governance had a positive estimate of 0.22. Within canceled engagements, item-to-score correlations were 0.54–0.62. These subgroup comparisons do not establish a change caused by cancellation.

The readiness ratings separated failed from higher composite scores less sharply than the management ratings of the two firms. Among respondents who rated a readiness indicator 1 to 3, 54.26% to 57.31% had a score of 3 or below. Among those who rated one 4 to 7, 7.99% to 10.00% still did, against 2.65% to 5.30% for the co-management indicators. Table 3.9 sets out the seven indicators against these measures.

What the scale does not capture is that the finance team we were given had no mandate. Their data itself was, I would say, in reasonable order; it was the absence of anyone on their side empowered to decide that ended this.”

PORTFOLIO DIRECTOR · KPO, Corporate and back-office functions

1 Financial-Institution-Provided Data

Financial-institution-provided data connect governance of inputs with manual correction inside delivery. The rated data-governance indicator and written accounts of absent validation answer related but different questions. One describes perceived capability, while the other identifies reported occasions when input checks were missing before handover. Reading them together clarifies the control boundary without treating the accounts as an independently verified input-error rate.

1.1 Maturity of the Financial Institution's Data Governance

Indicator C097 · Rate the maturity of the customer's data governance over the data this engagement uses · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 1.3 · Full statistical analysis ›

Weak data-governance ratings were associated with a weaker wider control environment, with a positive adjusted cross-category estimate in the main model. The financial institution's control over the data used by the engagement provides the substantive connection. Data definitions, accountability, and preparation are part of the conditions under which the shared work operates. The adjusted estimate is model dependent and its significance must be read under the stated test family.

Data governance was associated with the score computed without it at r = 0.75 (95% CI 0.67 to 0.82, n = 150 engagements), the lowest of the 47 governance indicators. It was associated with governance in the other four categories at r = 0.74, and after that adjustment the association remained at partial r = 0.22 (p = 0.006).

Of the 401 respondents who rated data governance 1 to 3, 229 (57.11%) had a score of 3 or below. Among the 377 who rated it 4 to 7, 33 (8.75%) still had a score of 3 or below, and none of the 19 who rated it 8 to 10 did (chi-square = 215.42, p < 0.001, Cramér's V = 0.52). Respondents who rated it distressed recorded a score of 2.89 (n = 401), against 5.11 (n = 396) among the rest (cluster-robust p < 0.001).

The association was the weakest of the 47 within the canceled engagements, at r = 0.54, against 0.64 to 0.71 across the active stages. After Holm adjustment, it was significant in 84 of the 87 segment levels tested. Within role it stood at 0.74 among individual contributors and 0.71 among engagement managers and fell to 0.51 among the 31 portfolio directors, the lowest within-role value of any governance indicator. Among individual contributors alone, 59.38% (193 of 325) of those who rated it 1 to 3 had a score of 3 or below.

Data governance tracked the manual alteration of financial-institution-provided data across its rating bands. Of the respondents who rated data governance 1 to 3, 75.19% (300 of 399) altered the data, against 24.39% (90 of 369) at ratings of 4 to 7 and 1 of 19 at ratings of 8 to 10 (chi-square p < 0.001). Part 5 (Leading Dynamics of Manual Alteration, section 1.3) reports that comparison in full.

Data governance drew 172 written answers from 155 respondents on 99 engagements, and 152 were negative, 77 from canceled and 75 from active engagements. Of the 152, 57 (37.50%) described one of the 26 forms of a control bypassed or not working, and 50 of the 57 described data handed over with no validation by the financial institution. Respondents described fields with no owner of their meaning, classifications that changed from month to month with no approver, and governance rules written after the work had gone live.

“vendor master made me angry from first week… same supplier written three different ways in three systems, legal name in one, short name in other. nobody owns which is correct” -Individual Contributor, CXM, Corporate and back-office functions

Part 2 (Financial Institution Readiness, section 1.4) reports the association of data governance with the four study measures.

1.2 Absence of Validation by the Financial Institution Before Handover

Written pattern · All written answers to the open questions · 50 answers from 47 respondents on 37 engagements · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 1.4

Accounts of data handed over without validation came from respondents with lower mean scores. This connects a specific input-control gap with the broader assessment of governance. The operational concern is that checking and correction may occur after handover through the delivery workforce, with responsibilities less visible than a validated input process would make them. The written pattern identifies reported experience, not the incidence of all unvalidated data deliveries.

The 47 respondents who described it recorded a score of 2.88, against 4.06 among the other 803 (cluster-robust p < 0.001). Among individual contributors alone, the 47 who described it recorded 2.88 against 3.94 among the other 620 (cluster-robust p < 0.001).

The engagement share describing unvalidated data correlated with the score at r = −0.24 (p = 0.004, n = 150). After adjustment for the comparison mean, partial r was 0.01 (nominal p = 0.922). The adjusted analysis did not detect a separate association. It does not establish that absent validation has no operational relationship with governance.

The pattern differed across engagement status and gathered where the composite score was low. It was described by 2.86% of respondents in Build-Out, 4.00% in Transition/Cut-Over, 6.80% in Production, and 7.76% on canceled engagements, and by 9.09% in the lowest indicator quintile against 1.88% in the highest. By the score, it was described by 9.19% (26 of 283) of respondents at 3 or below, 3.98% (21 of 528) above 3 to 7, and none of the 39 above 7. All 47 who described it were individual contributors.

The share describing the pattern differed after Holm adjustment by management's understanding of the governing obligations, by the financial institution's discussion of them, by the manual alteration methods used, by disclosure to the financial institution, and by intent to leave. It differed by none of the structural attributes of the engagement.

The 50 answers came from all written answers to the open questions, and 26 of them were written on canceled engagements. Part 5 reports the same 50 answers from the same 47 respondents on 37 engagements, 49 of them answers on the manual alteration of data, so the two counts describe one body of testimony. Respondents described correcting the data on every file before the agent could run, tickets to the financial institution that went unanswered, and a delivery team that had become the only validation the data received.

“the problem is nothing is validated on their side before it reaches us, so i fix data myself every time, further raising ticket does nothing at all, nobody replies to it” -Individual Contributor, ITO, Financial crime and client onboarding

“I never raise this with anyone above me, just sort duplicate myself and move on… And this happens on most files now, on top of that source system clearly has no dedup on their side, we are the dedup” -Individual Contributor, KPO, Credit, underwriting and suitability

Part 5 (Leading Dynamics of Manual Alteration, section 1.4) analyzes the pattern as a factor in manual alteration.

2 Management of the Financial Institution's Side of the Engagement

Management of the financial institution's own work and its response to issues were associated with the wider control environment. These responsibilities concern actions that the vendor cannot complete solely through its own delivery process. The adjustment results show model dependence, particularly for response speed. The operational interpretation rests on the reported responsibilities and evidence, rather than a claim that the coefficients isolate their causal effects.

2.1 Effectiveness of the Financial Institution's Management of Its Side

Indicator C105 · Rate the effectiveness of the customer's management of its side of this engagement · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 3.2 · Full statistical analysis ›

Weak management of the financial institution's side was associated with low composite scores, while the adjusted cross-category analysis did not detect a clear association. This indicator concerns the institution's ability to perform its own engagement responsibilities. It extends the governance question beyond vendor delivery, because shared work also depends on decisions and actions inside the institution. The conditional result does not establish that those responsibilities have no distinct operational role.

Of the 398 respondents who rated it 1 to 3, 227 had a score of 3 or below. Among the 390 who rated it 4 to 7, 39 (10.00%) still did, the highest share at those ratings of the seven readiness indicators, and none of the 23 who rated it 8 to 10 did (chi-square = 209.26, p < 0.001, Cramér's V = 0.51).

The indicator was associated with the score computed without it at r = 0.78 (95% CI 0.71 to 0.84, n = 150 engagements) and with governance in the other four categories at r = 0.73. After adjustment for the mean of the other 53 indicators, the partial r was −0.13 and was not significant (p = 0.119). Respondents who rated it distressed recorded a score of 2.87 (n = 398), against 5.09 (n = 413) among the rest (cluster-robust p < 0.001).

The association was observed within the canceled engagements at r = 0.58 and across the active stages at 0.68 to 0.79, and after Holm adjustment it held in 85 of the 87 segment levels tested. Within role it stood at 0.77 among individual contributors, 0.73 among engagement managers, and 0.75 among portfolio directors. Among individual contributors alone, 59.44% (192 of 323) of those who rated it 1 to 3 had a score of 3 or below.

The indicator drew 196 written answers from 173 respondents on 104 engagements, and 184 of them were negative, the most uniformly negative testimony among the governance indicators with 100 or more answers. Of the 184 negative answers, 105 came from canceled and 79 from active engagements, and 48 (26.09%) described one of the 26 forms of a control bypassed or not working, most often problems raised with no response (21), undefined handoffs between the financial institution's teams (14), and raising problems seen as futile (9). Respondents described a financial institution with no single owner of the agent work, questions that passed among three functions, and staff who held no authority to approve anything.

“My honest frustration — and it is the only one worth writing down — is that there is no single owner on their side for the agent work. Procurement operations, their vendor risk function and the technology group each hold a piece, so a question about whether an agent may close a supplier onboarding case without a human goes round all three and comes back a fortnight later, usually unanswered. Everything else about them is genuinely solid.” -Engagement Manager, BPO, Corporate and back-office functions

“The worst of it is nobody over there is actually running their end. I can't get into the Custodian portal half the time and break files land in three different layouts. But their people know the tech fine. They just aren't set up to run this.” -Individual Contributor, BPO, Trade, settlement and reconciliation

Part 2 (Financial Institution Readiness, section 5.1) reports the association of the indicator with the four study measures, and Part 5 (Leading Dynamics of Manual Alteration, section 3.2) its association with manual alteration.

2.2 Speed of the Financial Institution's Response to Issues Raised

Indicator C109 · Rate how quickly the customer responds when this engagement raises issues that need customer action · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 3.1 · Full statistical analysis ›

The financial institution's response to issues was closely associated with the wider control environment and drew extensive written discussion. Issues requiring financial institution action need to be resolved while work continues. Its adjusted coefficient changed sign when the institution's other readiness indicators were removed from the comparison mean. That sensitivity limits a distinct-effect interpretation and gives the operating accounts more explanatory value than a ranking of comment volume.

Accounts of response speed included unresolved decisions while delivery continued, linking the indicator to the availability of timely financial institution action. It was addressed in 498 written answers from 382 respondents on 142 engagements, more than any other governance indicator.

The indicator was associated with the score computed without it at r = 0.79 (95% CI 0.72 to 0.84, n = 150 engagements) and with governance in the other four categories at r = 0.75. After adjustment for the mean of the other 53 indicators, the partial r was −0.18 (p = 0.033), the one nominally significant inverse association among the readiness indicators. The 53 include the financial institution's other readiness indicators. Leaving them out of the control moved the partial r to 0.13 (p = 0.111), while leaving out the vendor capability block left it at −0.35 (p < 0.001), the AI technology block at −0.04 (p = 0.594), and the third-party block at −0.15 (p = 0.065). These alternative specifications show sensitivity to the comparison measure. They do not establish that response speed has no relationship of its own.

Scores were 3 or below for 229 of 404 (56.68%) at indicator ratings 1–3. The corresponding counts were 30 of 367 (8.17%) at indicator ratings 4–7 and 0 of 18 (0.00%) at indicator ratings 8–10. The reported band results were chi-square = 214.22, p < 0.001, Cramér's V = 0.52. Respondents who rated it distressed recorded a score of 2.91 (n = 404), against 5.18 (n = 385) among the rest (cluster-robust p < 0.001).

The association was observed within the canceled engagements at r = 0.61, the second-strongest of the seven readiness indicators, and across the active stages at 0.65 to 0.76. After Holm adjustment, it was significant in 85 of the 87 segment levels tested. Within role it stood at 0.77 among individual contributors, 0.73 among engagement managers, and 0.63 among portfolio directors. Among individual contributors alone, 57.82% (196 of 339) of those who rated it 1 to 3 had a score of 3 or below.

Of the 498 written answers, 437 were negative, 217 from canceled and 220 from active engagements. Of the 437, 181 (41.42%) described one of the 26 forms of a control bypassed or not working, most often problems raised with no response (120), raising problems seen as futile (35), and undefined handoffs between the financial institution's teams (21). Respondents described requests for a supplier list that went unanswered for weeks, and exceptions held open late into the evening while the financial institution did not say which record was correct.

“i felt like i was working for nobody. we asked for the vendor master and the supplier list for weeks, no reply. two months on that desk and their side never turned up once” -Individual Contributor, CXM, Corporate and back-office functions

“This is not my job but I am the one doing it. Their side does not answer… I sit with PO exceptions until late and nobody says which record is correct. My team is tired, two already left. Then they ask why the queue is old” -Individual Contributor, BPO, Corporate and back-office functions

Part 2 (Financial Institution Readiness, section 5.2) reports the association of response speed with the four study measures, and Part 5 (Leading Dynamics of Manual Alteration, section 3.1) its association with the manual alteration of processes.

3 Governance of Agentic AI Within the Financial Institution

Internal ownership, oversight capacity, and change control each connect the financial institution's organization with governance of agent work. The indicators distinguish naming responsibility, having the capacity to exercise it, and handling required changes. Their similar score relationships should not obscure those separate functions. The adjusted results qualify a claim of a distinct statistical association for each control.

3.1 Assignment of Ownership and Accountability for AI

Indicator C112 · Rate how clearly ownership and accountability for AI are assigned inside the customer organization · Full statistical analysis ›

Unclear assignment of AI ownership and accountability inside the financial institution was associated with low scores. The indicator addresses an internal responsibility boundary that joint vendor-institution procedures cannot settle by themselves. It asks whether the institution has identified who owns AI-related decisions and consequences within its organization. The findings concern reported clarity, with no independent organizational audit of those assignments.

Of the 417 respondents who rated ownership 1 to 3, 232 had a score of 3 or below. Among the 371 who rated it 4 to 7, 34 (9.16%) still did, and none of the 14 who rated it 8 to 10 did (Monte Carlo chi-square = 198.34, p < 0.001, Cramér's V = 0.50).

Ownership was associated with the score computed without it at r = 0.78 (95% CI 0.71 to 0.83, n = 150 engagements) and with governance in the other four categories at r = 0.73. After adjustment for the mean of the other 53 indicators, the partial r was −0.11 and was not significant (p = 0.201). Respondents who rated it distressed recorded a score of 2.95 (n = 417), against 5.12 (n = 385) among the rest (cluster-robust p < 0.001).

The association was observed within the canceled engagements at r = 0.56 and across the active stages at 0.67 to 0.74, and after Holm adjustment it held in 85 of the 87 segment levels tested. Within role it stood at 0.76 among individual contributors, 0.74 among engagement managers, and 0.65 among portfolio directors. Among individual contributors alone, 57.60% (197 of 342) of those who rated it 1 to 3 had a score of 3 or below.

Forty-six respondents on 40 engagements wrote about ownership of AI, one answer each, and 41 of the answers were negative, 18 from canceled and 23 from active engagements. Of the 41, 10 described one of the 26 forms of a control bypassed or not working, most often problems raised with no response (7) and undefined handoffs between the financial institution's teams (5). Those five answers described ownership divided among functions rather than absent, with risk, technology, and operations teams each assigning the agent to another.

“There was no named owner on the bank side for the screening model. Compliance, IT and the AML operations team each said it belonged to the other.” -Engagement Manager, ITO, Financial crime and client onboarding

“We spent most of the POC working around a surveillance stack that predates all of us, and their engineers are as frustrated with it as we are. The harder part is that no one on their side owns the AI decisions, so approvals wander” -Engagement Manager, ITO, Regulatory, audit, tax and legal

Part 2 (Financial Institution Readiness, section 6.1) reports the association of ownership with the four study measures.

3.2 Capacity to Govern and Oversee Agent Risk

Indicator C111 · Rate the customer's capacity to govern and oversee the risks of AI agents doing this work · Full statistical analysis ›

Limited capacity to oversee agent risk was associated with low scores. This concerns whether the financial institution can examine and govern the risks introduced by agent work, in addition to meeting its ordinary management responsibilities. The main adjusted model did not detect a clear cross-category association. That result qualifies the statistical claim while preserving the distinction between nominal responsibility and reported capacity to exercise oversight.

Of the 417 respondents who rated oversight capacity 1 to 3, 239 had a score of 3 or below. Among the 363 who rated it 4 to 7, 29 (7.99%) still did, the lowest such share of the seven, and none of the 17 who rated it 8 to 10 did (chi-square = 220.35, p < 0.001, Cramér's V = 0.53).

Oversight capacity was associated with the score computed without it at r = 0.79 (95% CI 0.72 to 0.84, n = 150 engagements) and with governance in the other four categories at r = 0.74. After adjustment for the mean of the other 53 indicators, the partial r was −0.15 and was not significant (p = 0.068). Respondents who rated it distressed recorded a score of 2.95 (n = 417), against 5.16 (n = 380) among the rest (cluster-robust p < 0.001).

The association was weakest in Build-Out, at r = 0.60, against 0.79 in Transition/Cut-Over and 0.73 in Production, and held within the canceled engagements at 0.62, the strongest of the seven. After Holm adjustment, it was significant in 84 of the 87 segment levels tested. Within role it stood at 0.78 among individual contributors, 0.74 among engagement managers, and 0.71 among portfolio directors. Among individual contributors alone, 60.00% (201 of 335) of those who rated it 1 to 3 had a score of 3 or below.

Forty-five respondents on 38 engagements wrote about oversight capacity, and 37 of the answers were negative, 19 from canceled and 18 from active engagements. Only 3 of the 37 described one of the 26 forms of a control bypassed or not working. Respondents described audit frameworks that did not exist for the agents and model-risk governance built after launch. A portfolio director on a canceled payments engagement reported that the financial institution's AI oversight forum had been in place before the vendor arrived, while ownership of the data it oversaw never settled.

“The client was not an unserious counterparty. On governance they were ahead of most acquirers we have dealt with – their AI oversight forum was standing before we arrived, which was not true of two other payments accounts I have been near – and their process documentation was usable, which is rarer still. What did not move in three months was ownership of the transaction data itself. Nobody on their side could tell us definitively who signs off a change to a merchant reference field, and so the change went to a forum that met when it met. Data governance is the item I marked hardest and I would mark it the same today. The rest of it, access and formats, is ordinary and fixable inside a quarter. That one is not.” -Portfolio Director, ITO, Technology engineering and operations

Part 2 (Financial Institution Readiness, section 6.2) reports the association of oversight capacity with the four study measures.

3.3 Change Control for the Changes the Engagement Requires

Indicator C110 · Rate how well the customer's change control process handles the changes this engagement requires · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 3.4 · Full statistical analysis ›

Weak financial institution change-control ratings were associated with a weaker reported control environment. This concerns changes the institution must handle on its own side, distinct from the joint process used between the firms. The main adjusted cross-category estimate was negative and not clearly distinguishable from zero. The substantive interpretation remains a reported readiness constraint, without treating the coefficient as evidence that better change control worsens governance.

Change control was associated with the score computed without it at r = 0.80 (95% CI 0.74 to 0.85, n = 150 engagements), the highest of the seven. After adjustment for the mean of the other 53 indicators, the partial r was −0.14 and was not significant (p = 0.088).

Of the 411 respondents who rated change control 1 to 3, 223 had a score of 3 or below. Among the 369 who rated it 4 to 7, 34 (9.21%) still did, and none of the 20 who rated it 8 to 10 did (chi-square = 190.63, p < 0.001, Cramér's V = 0.49, the smallest conditional share of the 47). Respondents who rated it distressed recorded a score of 2.93 (n = 411), against 5.16 (n = 389) among the rest (cluster-robust p < 0.001).

The association fell further on cancellation than for any other readiness indicator, from 0.63 to 0.77 across the active stages to 0.55 within the canceled engagements. After Holm adjustment, it was significant in 85 of the 87 segment levels tested. Within role it stood at 0.77 among individual contributors, 0.76 among engagement managers, and 0.72 among portfolio directors. Among individual contributors alone, 56.63% (188 of 332) of those who rated it 1 to 3 had a score of 3 or below.

Change control drew 85 written answers from 80 respondents on 64 engagements, and 69 were negative, 33 from canceled and 36 from active engagements. Of the 69, 15 (21.74%) described one of the 26 forms of a control bypassed or not working, most often problems raised with no response (7) and habit in place of a control (3). Respondents described a change board that sat on fixed days or once a month against agents that changed continuously, and a form to add one benefit code that was still open while the payroll cycle ran on.

“Their change board sits once a month. Nothing gets approved between… And they want same desk process as before, only with agent doing it.” -Individual Contributor, ITO, Technology engineering and operations

“Question about change control assumes there is change control to rate. For payroll there is none we can point to. We raised a form in March for adding one benefit code and it is still open, and meanwhile the Cycle runs anyway. Same for data question, score of 1 does not separate file that is late from file that is wrong, and we get both, from hris and from the tax side, with definitions that do not agree between them. Their people who own AI are at least named, that part is clear enough. For the rest the scale is too narrow.” -Engagement Manager, BPO, Corporate and back-office functions

Part 2 (Financial Institution Readiness, section 6.3) reports the association of change control with the four study measures, and Part 5 (Leading Dynamics of Manual Alteration, section 3.4) its association with manual alteration.

4 Provisioning of System Access and Agent Identities

Provisioning system access and agent identities was associated with the wider control environment and was repeatedly described in accounts of manual work. This responsibility connects authorized access with the ability to operate the intended workflow. The finding concerns reported provisioning capability and does not independently verify permissions, identity security, or the reason for each manual intervention.

4.1 Ability to Provision Access and Identities for Agents

Indicator C108 · Rate the customer's ability to provision the system access and identities that AI agents require · Crossed with manual alteration in Part 5, Leading Dynamics of Manual Alteration, section 2.1 · Full statistical analysis ›

Difficulty provisioning agent access and identities was associated with low scores. Access is a governance issue as well as a delivery dependency, because the institution must supply the identities and permissions through which agent work is performed. The written accounts connect this responsibility with manual handling of work. The findings describe reported ability and experience, without verifying the security or appropriateness of each identity and permission.

Of the 410 respondents who rated provisioning 1 to 3, 232 had a score of 3 or below. Among the 376 who rated it 4 to 7, 32 (8.51%) still did, and none of the 15 who rated it 8 to 10 did (Monte Carlo chi-square = 212.66, p < 0.001, Cramér's V = 0.52).

Provisioning was associated with the score computed without it at r = 0.77 (95% CI 0.69 to 0.83, n = 150 engagements) and with governance in the other four categories at r = 0.72, the lowest of the seven. After adjustment for the mean of the other 53 indicators, the partial r was −0.12 and was not significant (p = 0.133). Respondents who rated it distressed recorded a score of 2.90 (n = 410), against 5.14 (n = 391) among the rest (cluster-robust p < 0.001).

The association was weakest in Build-Out, at r = 0.58, against 0.77 in Transition/Cut-Over and 0.74 in Production, and it held within the canceled engagements at 0.57. After Holm adjustment, it was significant in 86 of the 87 segment levels tested. Within role it stood at 0.76 among individual contributors, 0.74 among engagement managers, and 0.62 among portfolio directors. Among individual contributors alone, 59.88% (197 of 329) of those who rated it 1 to 3 had a score of 3 or below.

Provisioning drew 437 written answers from 348 respondents on 135 engagements, the second-largest written base of the 47 governance indicators, and 403 were negative, 181 from canceled and 222 from active engagements. Of the 403, 67 (16.63%) described one of the 26 forms of a control bypassed or not working, most often problems raised with no response (29), raising problems seen as futile (13), and unclear liability or regulatory accountability (9). The answers recorded the control used outside its design. Agents ran under shared service accounts or under the credentials of named people, credentials lapsed with no person assigned to renew them, and access management treated an agent identity as a new joiner with a line manager and a leaver date.

“The difficulty, and it is the only one I would put in front of their leadership, is identity… Every agent we place on the trust administration work needs its own credential on the client's core, and their access management still treats a credential as a person, with a joiner form, a line manager and a leaver date. We have waited eleven weeks for one service account. Everything else on this account I would call reasonable, the process documentation in particular is better than I usually see… but access is the item that quietly sets our timeline.” -Portfolio Director, BPO, Corporate and back-office functions

“Their merchant settlement files arrive in three formats from three platforms, and two of them still come as an SFTP drop with no schema attached. Agent identities were requested in March and half were still pending in July, so the agents run under a shared service account, which their own risk people have flagged. Their managers are reasonable to deal with. The estate underneath them is not.” -Engagement Manager, BPO, Trade, settlement and reconciliation

“Problem is only access. Agent still dont have own login on core servicing platform so we run half the account amendments under my team id, then audit come back and ask who did it. Data itself is fine mostly” -Individual Contributor, BPO, Payments, cards and disputes

Part 2 (Financial Institution Readiness, section 7.1) reports the association of provisioning with the four study measures, and Part 5 (Leading Dynamics of Manual Alteration, section 2.1) its association with manual alteration.

training was two recorded sessions and a quiz, so my team keep our own note file, the composite rules and the checks that catch bad output are all in it. i wrote most of it. accordingly nobody ask the training team for anything now.”

INDIVIDUAL CONTRIBUTOR · KPO, Trade, settlement and reconciliation