For CxOs / Capacity, choices and results
Could your function deliver five days’ work in three, with better results? STEP 1 creates time and confidence. STEP 2 redesigns the work so that capability can turn into financial value. Explore what the evidence supports for your function.
The prize / Evidence by function
Choose a function, inspect the evidence and open a job to see how effort and outcomes could improve.
Each diagram keeps the source’s units and activity detail. Time, cost and productivity measure different things.
McKinsey future operating model
55%of human time redirected
Routine service and programme work falls from 70% to 15% of human time. The model redirects the difference into strategy and managing agent capability.
Of five HR effort-days, 2.75 shift from routine work into a different contribution.
Source: McKinsey, August 2026, Exhibit 2 ↗
This is a change in the allocation of human time. The future model still totals 100%: the 55 percentage points released from routine work are reassigned, not removed from the workforce.
Select any activity to explore how the work and its results improve. Click, double-click or tap.
McKinsey’s proposed future allocation ↗. The 55 points move into 35 additional points for strategy and 20 for agent ownership. Human time still totals 100%.
5% People service
10% People programmes
15% Problem-solving
50% Strategy and innovation
20% Agent capability management
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
4 August 2026 · Proposed operating model
Strategic engagement and innovation move from 15% to 50% of human time; agentic capability management takes 20% in the proposed future mix.
Human time allocation in a model of HR by 2030.
A redistribution of human work in a future model, not a measured reduction in total HR effort. Agent ownership is substantial new work.
IBM
Case study reviewed 13 October 2026 · Self-reported company case
IBM reports that AskHR contributed to a 40% reduction in HR operating cost over four years.
IBM HR operations across a multiyear transformation.
A company cost result, not a controlled estimate of hours released or a replication of McKinsey’s allocation model.
Gartner
28 October 2025 · Survey
Only 7% of organisations provided guidance on how employees should use the time AI saved.
July 2025 survey of 114 HR leaders for the time-reallocation finding.
An organisation-level management finding, not an HR productivity estimate. It supports the need to allocate time deliberately.
Sources reviewed 13 October 2026. Download the evidence register.
McKinsey quantifies four broad work categories, not the individual jobs. The named activities within them are TIP’s decomposition for exploration; no per-job saving is asserted. Agent capability management is new work in the future allocation.
Open an activity to see the steps, outcomes and measures. Work designs are TIP proposals.
Can we move HR from processing demand to improving workforce capability, without shifting hidden work to managers or making employees feel abandoned?
Establish the function’s actual time allocation. Redesign one connected people journey, then compare service, manager capability and employee outcomes across complete cycles.
Measure: Total HR and manager effort; employee readiness and trust; useful monthly one-to-ones; progress on development commitments; capability and retention outcomes.
Change course when: Revise the proposal if work simply moves to managers, errors rise or employees cannot reach a person. Expand it if quality holds and people demonstrably spend the time on agreed priorities.
Apply PERSONAL and TEAM practice to a real people-service workflow, then use FUNCTION to redesign ownership and handoffs. The test above is a proposed application, not a published HR client result.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Managers who develop their people, employees ready to contribute, and a workforce shaped around the business’s future needs.
35%More strategy and innovation
20%Own and improve agent capability
Routine service and programme work falls from 70% to 15% of human time. The model redirects the difference into strategy and managing agent capability.
Of five HR effort-days, 2.75 shift from routine work into a different contribution.
A proposed future allocation, not measured hours removed. The human working week still totals 100%.
| Work | Before | Retained | Released |
|---|---|---|---|
| People service delivery | 40 | 5 | 35 |
| People programmes | 30 | 10 | 20 |
| People problem-solving | 15 | 15 | 0 |
| Strategy and innovation | 15 | 15 | 0 |
| Total | 100 | 45 | 55 |
Service: 40 to 5; programmes: 30 to 10; problem-solving: 15 to 15. Strategy retains its original 15 and gains 35, reaching 50. Agent capability management receives 20. The 55 points released from routine work are all reassigned: 35 + 20.
Read the original source ↗McKinsey published allocation model
30%capacity available after more steering
The model releases 50% of capacity from preparation and analysis. It puts 20% more into business steering, leaving 30% for leaders to allocate.
Five effort-days become 2.5 on the original activities, plus one extra day steering the business and 1.5 days to allocate.
Source: McKinsey, July 2026, Exhibit 2 ↗
This is a share of the original FP&A capacity. The model releases 50 points, adds 20 to business steering and leaves 30 available. Those three figures describe different stages of the same allocation, not three separate savings.
Select any activity to explore how the work and its results improve. Click, double-click or tap.
McKinsey’s proposed future allocation ↗. Of the 50 points released, 20 go into additional business steering and 30 remain available.
5% Collection
10% Validation
5% Reports
15% Analysis and advice
35% Business steering
30% Available capacity
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
24 July 2026 · Allocation model and separate company cases
The illustrated allocation releases 50 points from existing activities, puts 20 into additional business steering and leaves 30 available. A separate telecom case reports more than 40% of FP&A capacity reallocated.
The FP&A function, with a model of future capacity allocation.
The model and company case are distinct. Neither establishes the saving across all finance activities.
BCG
4 June 2025 · Survey
Median reported ROI was 10%; only 45% of executives could quantify ROI from their initiatives.
More than 280 finance executives experienced with AI, surveyed in March 2025.
ROI is a financial return measure, not time released. Self-reported results show that implementation cost and controls matter alongside speed.
Gartner
18 November 2025 · Survey
59% reported AI use in finance. Gartner also reports low or moderate initial impact for 91% of respondents.
183 CFOs and senior finance leaders, surveyed May–June 2025.
Adoption and perceived impact are not measured hours saved. Early deployment does not establish an immediate capacity return.
Sources reviewed 13 October 2026. Download the evidence register.
All five source activities are retained. The flow separates effort continuing in the original work from the 50 points released, then shows the source’s reinvestment and final allocation.
Show me the reconciled numbers, control quality and cost of running this. An FP&A example cannot establish the saving across accounting, tax, treasury and every business unit.
Baseline the FP&A work from data collection through the business decision. Test a complete planning cycle and the actions it enables, including control, corrections and technology costs.
Measure: Net human hours per completed cycle; reconciliation exceptions; forecast usefulness and accuracy; earlier decisions; total cost and payback. Separate operating capacity from cash released.
Change course when: Stop expanding if controls weaken or checking consumes the gain. Increase the scope when reconciled outputs arrive earlier and the business can show which decisions improved.
Use the pathway to connect personal analysis, team challenge and a finance workflow experiment. Five-in-four practice is a starting point for testing, not a finance-wide audited result.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Earlier choices on price, margin, cash and investment, while there is still time to change the result.
20%More business steering
30%Available for agreed priorities
The model releases 50% of capacity from preparation and analysis. It puts 20% more into business steering, leaving 30% for leaders to allocate.
Five effort-days become 2.5 on the original activities, plus one extra day steering the business and 1.5 days to allocate.
FP&A scope. The model does not measure savings across accounting, tax or treasury.
| Work | Before | Retained | Released |
|---|---|---|---|
| Collect information | 25 | 5 | 20 |
| Validate information | 20 | 10 | 10 |
| Produce reports | 15 | 5 | 10 |
| Analyse and advise | 25 | 15 | 10 |
| Steer the business | 15 | 15 | 0 |
| Total | 100 | 50 | 50 |
Collection 25 to 5; validation 20 to 10; reporting 15 to 5; analysis/advice 25 to 15; existing steering 15 remains. 50 points are released: 20 adds to steering (15 to 35), 30 remains available.
Read the original source ↗McKinsey illustrated transformation office
39.2%capacity released across ten activities
Ten recurring activities in one transformation-office illustration. Monthly team effort falls from 65 to 39.5 days, releasing 25.5 days for higher-value work.
Five days of this recurring work become about three, releasing almost two for delivery.
Source: McKinsey, September 2026, Exhibit 2 ↗ · Original ten-activity exhibit ↗
This source measures combined team effort in days per month. It can therefore be expressed as the same work taking 3.04 instead of five effort-days. It does not measure elapsed programme duration or cash savings.
Select any activity to explore how the work and its results improve. Click, double-click or tap.
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
16 September 2026 · Early implementation reports and illustration
The illustrated ten recurring activities total 65 effort-days a month before and 39.5 after: 25.5 days, or 39.2%, released.
Core transformation-office reporting, insight and steering activities.
Task and office examples, not a measured enterprise-wide saving or proof of programme benefits. The chart above retains the original graphic’s ten activities and values.
BCG
30 September 2026 · Cross-company survey and analysis
BCG classifies 7.5% of companies as its most AI-mature group and 41% as scaling and creating some value.
More than 1,300 CxOs and senior leaders across more than 20 sectors.
Enterprise maturity and value categories, not transformation-office hours released. Associations with strategy and execution do not establish a causal saving.
Project Management Institute
9 July 2024 · Survey
93% of high adopters reported greater productivity, compared with 58% of low adopters.
500 project professionals, all already using GenAI in their project work.
Shares of people reporting improvement, not the amount of improvement. Self-selection, self-report and the absence of non-users limit causal interpretation.
Sources reviewed 13 October 2026. Download the evidence register.
All ten source activities are shown separately. Ribbon widths represent monthly effort-days; figures are approximate and rounded in the original.
A smaller reporting effort is useful, but did it improve delivery? Faster packs can make an unchanged programme look healthier while dependencies and difficult decisions remain unresolved.
Compare a complete steering cycle and the follow-up work, tracking the same initiatives. Include data corrections, false alarms, meeting preparation and whether escalations lead to completed actions.
Measure: Net reporting and coordination hours; evidence accuracy; time to resolve decisions and dependencies; completed commitments; independently checked benefits. Count each benefit once across functions.
Change course when: Change the approach if reporting gets faster but decisions and delivery do not. Expand when capacity is demonstrably used for coaching, resolving dependencies and achieving agreed outcomes.
Use transcript evidence, where agreed, alongside workflow results to examine how decisions and commitments change. STRATEGIC and ENTERPRISE address allocation and incentives; this chart is not a measured TIP transformation-office result.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Resolve dependencies, coach initiative owners and turn commitments into verified business results.
25.5 daysMonthly team effort released
Leadership choiceCommission the delivery work
Ten recurring activities in one transformation-office illustration. Monthly team effort falls from 65 to 39.5 days, releasing 25.5 days for higher-value work.
Five days of this recurring work become about three, releasing almost two for delivery.
One illustrative office. Effort days combine the team’s work; they are not elapsed calendar days or realised programme benefits.
| Work | Before | Retained | Released |
|---|---|---|---|
| Fortnightly steering reports | 12 | 3 | 9 |
| Monthly steering preparation | 4 | 3 | 1 |
| Additional insight generation | 4 | 1 | 3 |
| Monthly financial reporting | 5 | 3 | 2 |
| Performance walk | 5 | 3 | 2 |
| Monthly workforce reporting | 10 | 6 | 4 |
| Domain updates and follow-ups | 8 | 6 | 2 |
| Review initiative measures | 5 | 5 | 0 |
| Recalibrate targets | 10 | 8 | 2 |
| Valuation guidance | 2 | 1.5 | 0.5 |
| Total | 65 | 39.5 | 25.5 |
Before: 12 + 4 + 4 + 5 + 5 + 10 + 8 + 5 + 10 + 2 = 65. After: 3 + 3 + 1 + 3 + 3 + 6 + 6 + 5 + 8 + 1.5 = 39.5. Released: 25.5 days, or 39.23%. Five equivalent effort-days × 39.5 / 65 = 3.04.
Read the original source ↗McKinsey marketing operating-model analysis
45–60%percentage points of human time redirected
McKinsey describes execution moving from 60–70% of marketers’ time to 10–15%. That creates 45–60 percentage points for a different contribution. The flow shows the lower boundary: 60 to 15.
The lower-bound illustration redirects 2.25 of five effort-days into customer understanding and growth.
Source: McKinsey, June 2026, always-on orchestration ↗
The source compares execution’s share of marketer time: 60–70% before and 10–15% after. The difference is 45–60 percentage points of total time, not a measured 45–60% reduction in the cost of the function.
Select any activity to explore how the work and its results improve. Click, double-click or tap.
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
22 June 2026 · Operating-model analysis and company experience
The article describes execution falling from 60–70% to 10–15% of marketers’ time. The chart uses the conservative 60-to-15 boundary.
Human time in marketing with always-on orchestration.
An opportunity from consulting experience, not a controlled whole-function trial. The change is 45–60 percentage points to reallocate, not a guaranteed cost saving.
BCG
9 September 2024 · Survey and operating-model analysis
74% expected GenAI to improve key business metrics, while 47% were unconfident or neutral about their current strategy.
900 executives across marketing, sales and service.
Expectations and confidence, not achieved capacity gains. BCG argues for redesign across the customer journey rather than isolated content tools.
Gartner
19 May 2026 · Survey findings
Gartner reports an average 4.8 hours saved per seller per week, yet 72% of organisations report low reinvestment into high-value selling.
Sales organisations; the release cites 210 CSOs and senior sales leaders surveyed January–February 2026.
Reported hours and reinvestment practices, not a controlled productivity trial. The release’s outcome associations do not prove causation.
Sources reviewed 13 October 2026. Download the evidence register.
Only the two overall time allocations are quantified in this source. The five marketing capabilities below explain the work in more detail; they do not have published time weights.
Open an activity to see the steps, outcomes and measures. Work designs are TIP proposals.
I do not need more average content or more low-quality leads. Where is the evidence that faster preparation produces better customer conversations, stronger margins and commercially useful experiments?
Choose a segment and a recurring discovery or proposal workflow. Record preparation effort and deliberately reallocate released blocks to customer discovery and GTM experiments. Compare opportunities of similar maturity.
Measure: Net preparation hours; time in useful customer conversations; qualified conversion; win rate and margin; experiment learning. Track brand distinctiveness and claim accuracy alongside volume.
Change course when: Revise the approach if content grows but customer relevance, conversion or margin deteriorates. Expand when better preparation changes a customer decision or a tested commercial assumption.
Keep customer understanding and GTM experiments at the centre of the work. The pathway connects better thinking to delivery; this is not a claim that TIP has measured a sales or marketing saving.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
More relevant offers and experiences, with evidence of customer demand before spending more to scale.
Customer valueDiscovery, propositions and GTM experiments
Commercial returnConversion quality, retention and margin
McKinsey describes execution moving from 60–70% of marketers’ time to 10–15%. That creates 45–60 percentage points for a different contribution. The flow shows the lower boundary: 60 to 15.
The lower-bound illustration redirects 2.25 of five effort-days into customer understanding and growth.
A potential allocation under a redesigned operating model. The 45% flow is the lower boundary, not an observed result or a sales-function forecast.
| Work | Before | Retained | Released |
|---|---|---|---|
| Marketing execution | 60 | 15 | 45 |
| Customer and commercial judgement | 40 | 40 | 0 |
| Total | 100 | 55 | 45 |
Lower boundary: execution 60 minus 15 = 45 points released. Upper boundary: 70 minus 10 = 60. The chart retains the other 40 at the lower boundary; 40 + 15 + 45 = 100. Sub-job time shares are not published.
Read the original source ↗BCG cross-company practice evidence
15–30%reported productivity improvement
BCG reports 15–30% productivity improvements from scaled use cases and describes a broader redesign across service, quality, knowledge and demand prevention.
Apply the capacity gain across the service, with resolution quality and customer effort as the tests.
Source: BCG, September 2024, customer service transformation ↗
BCG reports a productivity improvement from scaled customer-service use cases. It supplies neither a common hours denominator nor an activity-level time budget, so 15–30% cannot responsibly be labelled “days saved”.
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
Stanford Digital Economy Lab
Research first released 2023; linked abstract as reviewed · Field study
The linked study reports 15% more issues resolved per hour on average, with substantial differences by worker experience.
Staggered rollout to 5,172 customer support agents at one company.
A throughput measure, not 15% of the function’s hours released. The most experienced workers saw small speed gains and small quality declines.
McKinsey
23 February 2026 · Survey
40% of the most mature group reported significantly improved customer experience scores, compared with 12% of the least mature group.
440 customer care leaders and executives; maturity groups defined by McKinsey.
Percentages of respondents reporting improvement, not the size of improvement or capacity released. The association does not isolate AI’s causal effect.
BCG
6 September 2024 · Practice synthesis
Some organisations had achieved 15–30% productivity improvement. BCG argues that further gains require changes to the service operating model.
Customer service operations with scaled GenAI use cases.
No common denominator or controlled sample is supplied for the range. The larger ambition discussed in the article is not an achieved result.
Sources reviewed 13 October 2026. Download the evidence register.
The detailed work comparison is TIP’s application of the research. Each row opens its efficiency, effectiveness and measurement plan. No numerical saving is assigned to an individual row.
Open an activity to see the steps, outcomes and measures. Work designs are TIP proposals.
I am accountable for resolved customer needs, not shorter calls. Does the saving survive repeat contacts, difficult cases, demand peaks and the cost of correcting a confident wrong answer?
Compare one customer journey using a matched case mix. Follow customers through resolution and repeat contact, including escalations and human review. Assess new and experienced agents separately.
Measure: Net human hours per resolved need; first-contact resolution; repeat contacts; complaints; service quality and experience; peak-time coverage. Track cost per resolved need separately.
Change course when: Change the design if faster handling produces unresolved cases or moves effort to customers. Reallocate capacity only when service holds across the full journey and the rota can use it.
Use TEAM to surface assumptions and handoffs, then FUNCTION to redesign a live customer journey. This is a proposed test; TIP is not presenting an audited customer-operations result here.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Customers get their needs resolved with less effort, while the business prevents repeat demand and supports difficult cases.
Better serviceResolve the need at first contact
Less failure demandFix the recurring causes
BCG reports 15–30% productivity improvements from scaled use cases and describes a broader redesign across service, quality, knowledge and demand prevention.
Apply the capacity gain across the service, with resolution quality and customer effort as the tests.
Published productivity range, not an hours-saved percentage. The source does not give each job’s share, so the flows show the work without invented weights.
No job-level percentages or five-day conversion are claimed. BCG does not provide a common output-per-hour denominator or before/after task mix for the reported range.
Read the original source ↗McKinsey illustrative software-delivery model
~50%less programme effort in the source illustration
The source illustration combines fewer people with a shorter programme: about 200 person-years become about 100. The article separately reports 3–5× productivity in early implementations.
For a comparable body of software work, the illustration halves total team effort. The commercial test is whether accepted, useful changes reach customers safely.
Source: McKinsey, May 2026, Exhibit 4 ↗
Person-years measure total team effort. Calendar months measure elapsed delivery time, while team size measures staffing. They are shown separately because a faster release, a smaller team and more useful output are different results. None measures the whole Technology & Data function.
McKinsey Exhibit 4 · approximate, rounded figures
Before200person-years
After~100person-years
Before~100full-time equivalents
After~60full-time equivalents
Before24months
After18months
Product owner, business analyst, tech lead, engineers and testers.
Smaller expert teams supervise agent execution, with product ownership and technical judgement.
The article separately reports 3–5× delivery productivity in early implementations. That is not calculated from these illustrative bars or a forecast for the whole IT function.
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
28 May 2026 · Reported company experience
McKinsey reports three- to fivefold productivity improvements when companies redesign the delivery model.
Redesigned software delivery across multiple companies.
No common productivity protocol or job allocation is provided. This is not a measured reduction in the total technology function’s hours; contrasting studies remain relevant.
BCG
6 January 2026 · Survey and company examples
BCG reports a 25% productivity boost in software delivery, with 44% expected at full scale.
2025 study of 1,250 companies; software delivery is one use-case category.
Reported productivity and a future expectation are different measures. Neither establishes a reduction in total technology-function effort.
METR
10 July 2025; update 24 February 2026 · Randomised trial; later methodological update
With early-2025 tools, tasks took 19% longer with AI. METR’s later update suggests the effect may have improved but says selection effects make the new estimate unreliable.
Initial trial: 16 experienced open-source developers and 246 tasks on familiar repositories.
A specific population, task mix and tool period. Neither the slowdown nor later uncertain speedup is a universal conclusion about current software development.
Read METR’s 2026 update ↗Sources reviewed 13 October 2026. Download the evidence register.
Exhibit 4 is illustrative and rounded: about 100 to 60 FTEs; 24 to 18 months; about 200 to 100 person-years; 10 teams of 8–12 become 16 teams of 3–4. The 3–5× figure is a separate reported productivity finding.
Open an activity to see the steps, outcomes and measures. Work designs are TIP proposals.
Code generation is only part of delivery. Does the gain survive integration, testing, security, production support and the next engineer maintaining it? The tool and codebase also change the result.
Compare representative changes on the actual stack, including familiar legacy work and new work. Record total human effort through review, release and a follow-up period for defects and support.
Measure: Net human hours per accepted change; lead time; escaped defects; failed changes; service reliability; maintenance effort and running cost. Retain product and customer outcome measures.
Change course when: Reduce or redirect use when review, rework or defects erase the gain. Update the conclusion as tools improve, but rerun comparable work instead of treating yesterday’s result as permanent.
From Customer to Code joins customer discovery, specification, architecture, implementation and deployment. Our own website is a practical example, not a controlled benchmark for an enterprise technology function.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Deliver changes customers use, test more business assumptions and maintain reliable services as the pace increases.
Customer valueMore validated product changes
ReliabilityStronger tests and dependable services
The source illustration combines fewer people with a shorter programme: about 200 person-years become about 100. The article separately reports 3–5× productivity in early implementations.
For a comparable body of software work, the illustration halves total team effort. The commercial test is whether accepted, useful changes reach customers safely.
Illustrative software programme and selected early implementations. These are not whole-IT savings or a guarantee for a data platform, service desk or infrastructure team.
The exhibit reports approximately 200 to 100 person-years: about 50% less total effort. It separately shows 24 to 18 months and about 100 to 60 FTEs. Its rounded quantities are not an exact resourcing plan. The article’s 3–5× productivity claim is not derived from this illustration.
Read the original source ↗McKinsey insurance function estimate
10–30%potential productivity improvement
McKinsey expects a 10–30% productivity increase across insurance risk and compliance through GenAI. The opportunity spans evidence, review and monitoring.
Use additional capacity to identify material risks sooner and help the business make better decisions.
Source: McKinsey, July 2024, insurance risk and compliance ↗
The 10–30% is a potential productivity increase for insurance risk and compliance. The source does not give activity hours or an observed before-and-after result, so it is shown as an estimate rather than a time-saving claim.
These sources measure different things. Together they show the opportunity, the limits and why reallocation matters.
McKinsey
1 July 2024 · Industry analysis and estimate
McKinsey expects AI to enable 10–30% productivity improvement across these functions.
Risk and compliance functions in insurance.
A potential improvement, with no job-by-job effort split. Not an observed departmental saving or a general legal benchmark.
Schwarcz et al. / Journal of Law and Empirical Analysis
2026, volume 3(1), pages 220–250 · Randomised controlled trial
Table 10 estimates 25.1% less complaint-analysis time with o1-preview. Effects vary by task; NDA drafting has no statistically significant time benefit.
137 law students across six tasks; 126 observations in the complaint-analysis time comparison.
Student work using specific tools, not a practising department’s saving. Estimates use participant fixed effects and are not adjusted for multiple hypothesis testing. Quality and source verification still matter.
Stanford RegLab / HAI
23 May 2024 · Pre-registered tool benchmark
The evaluated systems produced incorrect information in more than 17% of queries for two tools and more than 34% for another.
More than 200 legal queries tested against the then-current legal research systems.
Historical versions and a specific query set, not current error rates or measured lawyer productivity. The result supports verification and local evaluation.
Sources reviewed 13 October 2026. Download the evidence register.
The detailed work comparison is TIP’s proposed application. Activity-level savings need a local baseline; the source does not publish them.
Open an activity to see the steps, outcomes and measures. Work designs are TIP proposals.
The cost of a missed exception can outweigh many hours saved. Where does expert accountability sit, and how much work is needed to verify sources, privilege, context and the system’s recommendations?
Select one repeatable review class. Compare assisted and current work using a blind expert review where practical, including difficult exceptions. Use authorised material and retain accountable human decisions.
Measure: Net reviewer hours per cleared item; material errors and missed exceptions; source accuracy; escalation quality; turnaround; audit trail. Include the cost of validation and ongoing monitoring.
Change course when: Do not scale a speed gain that increases material error or weakens accountability. Expand a bounded workflow only when quality thresholds hold and senior judgement is directed to the right cases.
Use the pathway to make verification, assumptions and decision ownership explicit before extending a workflow. This is a proposed risk and legal application, not a TIP-verified departmental saving.
Proposed application and leadership questions. TIP practice does not independently validate the external findings.
What the business gains
Give leaders earlier, evidence-based challenge and a sound route to pursue opportunities responsibly.
Earlier assuranceChallenge before commitment
Wider coverageEmerging risks and difficult cases
McKinsey expects a 10–30% productivity increase across insurance risk and compliance through GenAI. The opportunity spans evidence, review and monitoring.
Use additional capacity to identify material risks sooner and help the business make better decisions.
An expected function-level range in insurance, not a realised legal-department result. Job widths are unscaled because the source supplies no effort split.
No before/after hours are inferred from the productivity estimate. The work map is TIP’s proposed application; each job needs a local effort baseline and effectiveness test.
Read the original source ↗Time and productivity are different measures. The sourced allocation models show how a function’s work can shift. Where a publisher gives only a productivity range, the diagram maps the jobs without inventing their share of time.
Five in three is a useful ambition, not a universal forecast. The transformation-office illustration comes close: 65 to 39.5 effort-days. Other functions have different evidence and starting points. Include review, exceptions and running costs when testing yours.
Reallocation makes the difference. Decide which capacity improves the function and which serves an enterprise priority. Name an owner and an outcome; financial value follows only when cost, revenue or margin changes.
Each function has three external sources plus TIP, covering different kinds of evidence. Reviewed 13 October 2026.
Download the evidence registerHow the pathway makes the prize practical
People use AI on real work, check what improves and discover what is possible. The time they create makes room to question familiar assumptions and tackle a bigger change together.
Create time →Bring the workflow owners together to redesign the work around a customer or business outcome. This is where capability can start turning into financial results through lower cost, released spend or better commercial performance.
Change the work →Decide where capacity goes, align roles and incentives, and follow the result into the numbers. Finance verifies the benefit, after implementation and running costs, before it becomes a claim.
Align the business →TIP’s own reported practice includes five days’ work in four. We are now working with teams towards five in three. These are stages of learning and application, not a promised saving for every function. Time released becomes money only when a cost, revenue or margin outcome actually changes.
Evidence and leadership judgement
If AI changes the effort, cost or expertise a task needs, the assumptions behind a business may need revisiting. That can be harder than learning the tools, particularly when a familiar model has served you well.
Disagreement deserves examination. A source may be weak, the work may be different, or the risk may be unacceptable. Equally, a familiar explanation can become a reason never to test a new possibility. Agree the evidence that would make you expand, revise or stop the idea before seeing the result. Apply that standard to TIP’s proposal too.
Read the paper: What would change your mind?The STaRS lens / Michael D. Watkins
In The First 90 Days, Michael D. Watkins uses STaRS to help leaders match their approach to the business situation. A turnaround and a growing business need different choices. Different functions may also be in different situations.
Our application below uses that lens to decide where newly available capacity could make the greatest difference.
Put capacity into customer discovery, proposition tests and go-to-market experiments. Look for evidence of demand before scaling delivery.
Look for: Customer commitments and repeat use.
Use capacity to address cash, service failures and the few decisions that determine survival. Agree which costs can actually leave the business without creating a new failure.
Look for: Cash, service recovery and fewer recurring failures.
Relieve bottlenecks, improve onboarding and build the management capacity needed to serve more customers consistently.
Look for: Throughput and customer quality as demand grows.
Make time to question familiar assumptions, investigate changing customer needs and test a different offer or operating model.
Look for: Evidence that changes an investment or operating decision.
Invest in people, emerging risks and the next source of customer value while protecting the performance that funds them.
Look for: Resilience, capability and a tested pipeline of future value.
Framework: Watkins on matching strategy to the situation. Capacity applications are TIP’s interpretation, not an allocation formula from the book.
Our 40% capacity model puts retention, insourcing, emerging risks and cross-functional growth on the table. Leadership should change the mix to suit the business, rather than divide the time equally by default.
Decide what to stop, which work to improve and where people should spend more time. Name the owner, outcome and first test.
Agree which shared priorities need that capacity, how functions contribute and what budgets, measures and incentives must change. Do not count the same released time twice.
Efficiency AND effectiveness
STEP 1 creates the capacity and confidence to act. STEP 2 uses it to redesign a real workflow, demonstrate the result and build an evidence-based case for wider investment.
Choose a customer or business outcome. Record effort, waiting, checks, rework and quality before changing the workflow.
PERSONAL / TEAMChallenge the problem, redesign a manageable piece of work and compare everyday results with the baseline. Retain what others can reuse.
ADVANCED / ACCELERATEUse the evidence to change roles, priorities and incentives. Agree where capacity goes and review whether it delivers the intended value.
FUNCTION / STRATEGIC / ENTERPRISEMeasure the complete work, including review, exceptions, learning and running costs. Time released becomes a cash saving only when an actual cost changes.
Track the result that matters: a better decision, resolved customer need, stronger service or tested growth opportunity. More output alone does not establish value.
AI can help reveal assumptions and trade-offs. Your organisation owns the choices and needs to execute them.
Your next step
Bring a function, a live priority and the result you need. We will help you identify a practical starting point.
