Understand What Delivery Really Costs

Task

Calculate delivery effort, expert dependency, and profitability by engagement type.

Summary

Measure the time, bottlenecks, margins, and specialist help required for each type of work.

Measure the Real Economics of Delivery

Task ID: S1-07

A service business cannot know which offers deserve investment until it knows what each type of engagement consumes. This article explains how to measure delivery hours, senior-expert dependence, bottlenecks, handoffs, and profitability by engagement type—and how to turn that baseline into decisions about scope, pricing, staffing, process improvement, and eventual productization.

Executive summary

A consulting firm can appear healthy while its delivery economics are deteriorating. Revenue rises, calendars stay full, customers remain satisfied, and referrals continue. Yet the founder reviews every deliverable, senior specialists rescue poorly scoped projects, employees record only billable work, and fixed-fee engagements quietly consume twice the hours assumed in the proposal.

S1-07 addresses that problem. Within this methodology, it is an internal operating task: calculate delivery effort, expert dependency, and profitability by engagement type, producing evidence on hours, margin, bottlenecks, and handoffs. Its primary measure is expert hours per deal, and its immediate target is to capture a baseline—not to hit a universal industry ratio.

The central operating principle is simple:

Measure the resources each engagement actually consumes, not merely what was quoted, invoiced, or entered as billable time.

A credible analysis should connect five kinds of evidence:

1. recognized or economically earned revenue;
2. actual delivery time by activity and role;
3. the cost of supplying each role’s practical capacity;
4. non-labour delivery costs, rework, and post-delivery support;
5. the commercial and operational conditions that explain variation.

The analysis should separate engagement types because time-and-materials, fixed-price, retainers, managed services, and outcome-linked contracts place different risks on the supplier. Accenture, for example, reports that it uses time-and-materials, fixed-price, transaction-priced, and mixed-fee contracts; it recognizes many service revenues over time and continuously revises estimates of contract revenue, cost, and profitability. That is a useful reminder that contract value, invoices, bookings, and earned revenue are not interchangeable measures.

The most useful first baseline is not a single company-wide margin. It is a distribution for each engagement type: median and range of total hours, expert hours, loaded delivery cost, contribution margin, rework, cycle time, handoffs, and estimate error. The analysis should also show data coverage and uncertainty. One completed project is a case; it is not yet a reliable baseline for a category.

Expert hours per deal deserves special attention because senior people are usually both expensive and capacity-constrained. A profitable-looking engagement may still be difficult to scale if every sale requires the same founder, architect, clinician, lawyer, security specialist, or domain expert at several critical points. The objective is not to remove expert judgment. It is to distinguish work that genuinely requires judgment from coordination, review, correction, information gathering, and routine production that could be handled differently.

No external benchmark can determine the correct number of expert hours. The appropriate level depends on the customer’s risk, the promise being made, regulatory exposure, price, quality standard, novelty, team capability, and how much knowledge has been documented or built into software. The baseline is complete when leadership can explain where expert time goes, which engagement types produce acceptable economics, which constraints limit growth, and what decision should follow.

Scope and analytical foundations

What S1-07 means

This task assesses delivery economics by engagement type during a consulting-led baseline stage. It applies to professional and technology-enabled services, especially a founder-led company learning from real customer work before making an offer more repeatable or separating part of it into software.

Several boundaries matter:

• Delivery effort includes all work required to produce, validate, hand over, and stabilize the promised result—not only time invoiced to the customer.
• Expert dependency means the amount and timing of work that requires scarce senior judgment, approval, relationships, credentials, or system access.
• Profitability must be stated using a defined cost model. Contribution margin, project gross margin, and company operating margin answer different questions.
• Engagement type should describe a repeatable combination of commercial model, scope pattern, delivery process, and expected result. Customer names alone are not useful categories.
• Geographic scope is unspecified. Accounting, employment, privacy, tax, and recordkeeping requirements must therefore be localized. The legal examples below draw from international accounting standards, the European Union, the United Kingdom, and the United States.
• Time horizon is unspecified. The practical default is to analyze completed engagements from the most recent representative operating period while retaining older work when it is still comparable.

The analysis has practical data prerequisites. The company needs completed-project records, contracts or statements of work, invoices, time or activity evidence, compensation and contractor costs, and interviews with the people who performed the work. The analysis can begin without perfect systems; its uncertainty should not be hidden.

Where the method comes from

The analytical roots of S1-07 lie in job costing, activity-based costing, time-driven activity-based costing, project accounting, capacity management, revenue recognition, and knowledge management.

Traditional activity-based costing attempted to assign overhead to the activities, products, and customers that caused it. Robert Kaplan and Steven Anderson’s time-driven activity-based costing, first presented in a Harvard Business School working paper in 2003, simplified the model around two inputs: the cost of supplying resource capacity and the amount of capacity—usually time—required by a transaction, service, or customer.

This is especially relevant to service work. A project can be decomposed into activities, each activity can consume time from different roles, and each role can be assigned a capacity cost rate. Variations can then be expressed through “time equations”: for example, a standard implementation may require a base amount of work, while a legacy integration, regulated environment, accelerated deadline, or missing customer data adds further effort.

Revenue recognition provides a second foundation. IFRS 15 was issued in 2014, became effective for reporting periods beginning on or after January 1, 2018, and establishes principles for when and how much contract revenue should be recognized. The International Accounting Standards Board completed a post-implementation review in September 2024 and concluded that the standard was working as intended, while identifying several topics for possible further consideration.

Knowledge management provides a third foundation because expert dependency is not only a labour-cost issue. ISO 30401:2018 established requirements and guidance for organizational knowledge-management systems. As of August 1, 2026, ISO was balloting a draft second edition intended to replace the 2018 standard, which had also received amendments in 2022 and 2024.

timeline
    title Development of the methods behind S1-07
    1980s : Activity-based costing develops as a management-accounting method
    2003 : Kaplan and Anderson publish the TDABC working paper
    2014 : IASB and FASB issue the converged revenue standard
    2018 : IFRS 15 becomes effective
         : ISO 30401 is published
    2024 : IASB completes its IFRS 15 post-implementation review
    2026 : ISO begins balloting a draft second edition of ISO 30401

The timeline shows that S1-07 is a practical synthesis rather than a new accounting standard. It combines established ideas about cost, capacity, contracts, and organizational knowledge for a specific management decision.

Related methods, versions, and present status

What to measure and how to calculate it

Start with a defensible engagement taxonomy

The analysis becomes misleading when unlike work is averaged together. A two-week diagnostic, a six-month implementation, and a continuing support retainer should not share one “average project margin.”

A useful engagement type normally combines:

• the customer problem and promised result;
• the commercial model;
• the normal scope and exclusions;
• the delivery stages;
• the expected team and seniority mix;
• the usual duration;
• important complexity drivers;
• the support or warranty commitment after delivery.

Begin with a small number of categories that operators recognize. Split a category only when the difference explains economics or changes a decision. A taxonomy that produces twenty categories from twenty projects has described history without creating a repeatable management model.

The commercial model matters because it determines where uncertainty lands.

Accenture’s 2025 filing illustrates why these distinctions are operational rather than merely contractual. The company reports time-and-materials, fixed-price, transaction-priced, and mixed-fee contracts and uses different methods to measure progress depending on the nature of the service and contract.

ICF’s 2025 filing provides another example of segmenting revenue by commercial model: it reported approximately $802.0 million of time-and-materials revenue, $932.7 million of fixed-price revenue, and $138.2 million of cost-based revenue, or roughly 42.8%, 49.8%, and 7.4% of the disclosed total. These figures are evidence about ICF’s own contract portfolio, not benchmarks for a smaller consulting business.

Build the cost model from practical capacity

For each role or resource pool, calculate a capacity cost rate:

Capacity cost rate = Cost of supplying the role pool ÷ Practical delivery capacity

The numerator should include the costs that the company has decided the rate is meant to recover. At minimum, that usually includes salary or contractor cost, payroll-related costs, and benefits. A fuller delivery-cost rate may also include relevant management, equipment, software, occupancy, training, and other costs required to maintain the capacity.

The denominator should be practical capacity, not every paid hour in the year. Remove leave, holidays, training, management duties, internal meetings, business development, and a realistic allowance for unavailable or idle time. Kaplan and Anderson’s formulation emphasizes the cost of supplying capacity and the resource demand generated by the work; failing to model unused capacity can make active engagements appear more expensive while hiding the separate cost of maintaining excess capacity.

Accenture’s filing similarly identifies employee compensation and payroll as major service costs, together with subcontractors, facilities, technology, and travel. Its disclosure is useful as a cost-stack example, although a smaller firm should choose allocations that fit its own decision.

For each completed engagement:

Labour delivery cost = Σ(actual activity hours × relevant role capacity cost rate)

Then add:

Total attributable delivery cost = Labour delivery cost + subcontractors + variable tools + travel + expected rework and support cost

Two margin views are usually worth retaining:

Contribution margin % = (Revenue – attributable variable delivery cost) ÷ Revenue

Project gross margin % = (Revenue – defined cost of delivery, including selected delivery overhead) ÷ Revenue

Neither formula is universally “the” correct margin. The company must document its policy and apply it consistently. Contribution margin helps decide whether taking another similar engagement adds economic value with available capacity. Project gross margin helps assess whether the offer supports the delivery organization over time. Company operating margin then subtracts selling, general, administrative, and other operating expenses.

Measure the work the customer does not see

Time records should include more than visible production. The activity model should cover:

• discovery and preparation;
• customer meetings and follow-up;
• research and analysis;
• design, configuration, development, or production;
• internal review and quality assurance;
• project management and coordination;
• waiting and rescheduling caused by missing inputs;
• revisions and defect correction;
• handover, training, documentation, and stabilization;
• included post-delivery questions or warranty work;
• escalation and exception handling.

Founder time is often misclassified. Time spent winning the sale belongs in sales economics. Time spent doing the promised work, resolving delivery exceptions, reviewing output, or preserving the customer relationship during a delivery problem belongs in delivery economics. Recording all founder time as overhead makes the engagement appear more transferable than it is.

Treat expert dependency as a capacity and continuity risk

The primary measure is:

Expert hours per deal = Total qualifying expert delivery hours ÷ Number of completed deals

It should be accompanied by:

Expert share of delivery = Expert hours ÷ Total delivery hours

Expert demand ratio = Expert hours required in a period ÷ Expert hours practically available

Expert estimate variance = Actual expert hours – Planned expert hours

The classification must be based on the work, not merely job titles. An hour is expert-dependent when a less scarce role cannot yet perform it at the required quality, risk level, or authority. The reason should be coded:

• customer trust or relationship ownership;
• diagnosis or judgment;
• regulated approval or professional credential;
• architecture or design decision;
• access to sensitive systems;
• unusual exception;
• quality review;
• missing documentation or training;
• routine work retained by habit.

That final distinction is crucial. Some expert dependency protects quality. Some reflects a capability gap. Some exists only because the process, contract, data, or product creates avoidable exceptions.

A useful sensitivity test makes the economic consequence visible. Consider an illustrative $60,000 fixed-fee engagement requiring 40 expert hours at a loaded cost of $180 per hour, 220 other hours at $85 per hour, and $3,000 of other delivery cost. Attributable cost is $28,900 and contribution margin is about 51.8%. If expert time rises to 80 hours with everything else unchanged, cost rises to $36,100 and margin falls to about 39.8%. The extra expert time also consumes capacity that may block other sales.

Measure bottlenecks, handoffs, and variation

The average can conceal the constraint. A role may account for only 10% of total project hours but control every project’s start, approval, and completion.

For each engagement type, identify:

• the role with the highest demand relative to practical availability;
• the activities that wait for that role;
• the amount of work in progress awaiting review;
• the number of handoffs;
• wait time between handoffs;
• rework caused by incomplete or misunderstood handoffs;
• the proportion of work completed correctly on the first pass;
• the complexity factors associated with the longest or least profitable engagements.

A process map should be validated by the people who perform adjacent stages. A systematic review of 70 time-driven activity-based costing studies found that observation, surveys, and interviews were commonly used to build maps, but only 33% of studies integrated documents or operational data, while 67% did not report validation of the map by specialists. That evidence comes from healthcare, but the data-quality warning transfers directly: a process diagram based only on management recollection can look precise while missing actual variation.

flowchart LR
    A[Completed engagements] --> B[Normalize revenue and costs]
    B --> C[Map activities and roles]
    C --> D[Calculate effort and margin]
    D --> E[Group by engagement type]
    E --> F[Find expert and workflow constraints]
    F --> G[Change scope, price, staffing, process, or product]

In plain language: start from completed work, make the financial data comparable, reconstruct what happened, calculate the economics, compare like with like, locate the constraints, and use the evidence to change a decision.

Use a compact operating scorecard

Report the median, range, and important outliers, not only the mean. A profitable median with a severe loss tail may still be unacceptable when the company cannot predict which project will become the outlier.

Evidence from implementations and public examples

Time-driven costing can expose hidden resource use

The strongest broad implementation evidence for time-driven activity-based costing comes from healthcare, where researchers have mapped detailed service pathways and the time of different professional roles. A 2025 systematic review identified 32 studies across primary, secondary, acute, tertiary, and long-term care. It found that the method could identify resource use and cost drivers but also reported implementation weaknesses, including limited use of real-time tracking, inconsistent documentation of capacity cost rates, and incomplete integration into routine budgeting and decisions.

A separate systematic review of 215 cost-measurement studies found that process mapping, expert input, and direct observation were recurring practices associated with useful managerial cost information. It also warned that costing only a short or partial pathway can produce locally accurate numbers while missing important costs elsewhere in the full service cycle.

The lesson for consulting is direct: do not measure only the visible workshop, build sprint, audit fieldwork, or presentation. Include preparation, review, customer delays, internal coordination, corrections, handover, and included support. Otherwise, the cost model rewards moving work outside the measured boundary.

Technology can increase scale but does not repair a poor model. A literature review comparing manual and technology-supported costing studies found that technology-supported analyses covered substantially more cases on average, but the value still depended on process definition, data quality, and how the output was used.

Utilization cannot be managed independently of quality and margin

Accenture reported 92% utilization for fiscal 2025 and a gross margin of 31.9%, down from 32.6% in fiscal 2024, with higher payroll costs identified as the primary reason for the decline. The same filing warns that utilization that is either too high or too low can adversely affect employee engagement, attrition, work quality, and the company’s ability to staff projects.

This is a useful counterexample to simplistic utilization targets. High utilization may improve short-term absorption of payroll cost, but it can also remove the time required for training, documentation, process improvement, presales support, recovery from demanding work, and rapid response to new demand. Low utilization may indicate weak demand, poor scheduling, mismatched skills, or excessive capacity. The correct question is not “How do we maximize utilization?” It is “What utilization range supports quality, employee sustainability, response capacity, and target economics for this operating model?”

Accenture’s number is not a benchmark for a founder-led specialist firm. Its workforce scale, service mix, geography, accounting policies, use of subcontractors, and definition of utilization differ. Public-company metrics are valuable for learning which variables sophisticated operators monitor, not for copying a target without context.

Automation changes both cost and the commercial model

Automation and artificial intelligence can reduce delivery time, but that does not automatically increase profit. If the firm charges for hours, reducing hours may also reduce revenue. If it charges a fixed fee, automation may increase margin until competition, customer expectations, or contract renegotiation transfers part of the benefit to the buyer.

Accenture’s 2025 filing explicitly identifies this tension: automation may reduce demand for tasks previously performed by people, and the company may need new pricing or commercial models that reflect the value of AI-enabled services. It also identifies risks involving accuracy, bias, intellectual property, privacy, cybersecurity, and regulation.

S1-07 should therefore record both labour saved and new cost introduced by automation: software fees, model usage, verification, security review, data preparation, exception handling, and human quality assurance. A process is not economically improved merely because one visible task became faster.

What counts as a completed baseline

“Baseline captured” should mean that the company has produced an auditable management view, not that every timesheet field is perfect.

The evidence should include:

• a documented engagement taxonomy;
• a written definition of each cost and margin measure;
• a role-cost and practical-capacity model;
• engagement-level revenue, hours, costs, and margin;
• expert hours by activity and reason;
• process maps for major engagement types;
• bottleneck and handoff evidence;
• estimate-versus-actual comparisons;
• data-quality notes and unresolved assumptions;
• a management decision or test arising from the analysis.

Where the sample is thin, state it. A category represented by one or two unusual projects should be labelled provisional. Where time data are incomplete, reconstruct them from calendars, tickets, documents, version histories, meeting records, and interviews, but distinguish observed data from estimates.

Security, legal, ethical, and compliance considerations

Revenue and contract accounting

Management economics and statutory accounting must be reconciled but should not be confused. IFRS 15 centers revenue recognition on contracts, performance obligations, transaction price, and the transfer of promised goods or services. A firm may bill before revenue is earned, earn revenue before it can bill, or receive advance payment that remains a contract liability until the associated service is provided.

For fixed-price work, estimate revisions can change recognized revenue and expected profitability as work progresses. Accenture reports that it continuously monitors total contract revenue and cost estimates and recognizes an anticipated loss when it becomes probable and reasonably estimable for relevant contracts.

The S1-07 dataset should therefore retain separate fields for:

• signed contract value;
• approved scope changes;
• invoices issued;
• cash collected;
• recognized or economically earned revenue;
• deferred or unearned amounts;
• estimated cost to complete.

The specific accounting treatment should be reviewed by a qualified accountant under the company’s applicable standards.

Employment and working-time records

Time analysis may overlap with legal payroll and working-time obligations. In the United States, the Department of Labor states that employers must retain specified wage and working-hour records for covered non-exempt employees and that the selected timekeeping method must produce complete and accurate information. Requirements differ for exempt employees and across jurisdictions.

A management-costing system should not encourage unrecorded overtime or the removal of “non-billable” delivery work from employee records. When workers are pressured to fit actual effort into the sold budget, the company loses both lawful time evidence and the economic signal that the offer was underpriced or poorly designed.

Privacy and ethical monitoring

Time records and activity logs can become employee-monitoring data. The European Union’s General Data Protection Regulation requires personal data to be processed lawfully, fairly, and transparently; collected for specified purposes; and limited to what is necessary.

The United Kingdom Information Commissioner’s Office warns that excessive worker monitoring can intrude into private life, undermine trust and mental wellbeing, and create greater risks for people working from home. It recommends a clear purpose, a lawful basis, transparency, and the least intrusive means capable of meeting the objective.

For S1-07, that means:

• collect activity categories and duration needed for costing rather than keystrokes, screenshots, or continuous surveillance;
• explain what is collected, why, who can see it, and how long it is retained;
• avoid treating raw hours as an individual performance score;
• allow correction of obvious coding errors;
• aggregate reporting where person-level detail is unnecessary;
• assess local privacy, employment, and labour-relations requirements before introducing new monitoring.

Time spent is a resource measure, not a complete measure of contribution. A senior expert may prevent a major failure in twenty minutes; a junior employee may perform valuable work for several days. Ethical use requires context.

Confidentiality and access control

Delivery-economics records may reveal customer names, rates, contract terms, employee compensation, security incidents, system architecture, regulated work, and commercially sensitive margins. Access should be role-based and limited to what each person needs.

NIST defines least privilege as granting users and processes only the minimum system resources and authorizations necessary to perform assigned tasks. The NIST Cybersecurity Framework provides a broader structure for identifying, governing, protecting, detecting, responding to, and recovering from cybersecurity risk.

A practical design might allow project leaders to see hours and operational variation, finance to see cost rates and margins, and executives to see cross-engagement comparisons, without exposing every employee’s compensation or every customer’s commercial terms to all users.

Knowledge continuity

Reducing expert dependency should not mean copying sensitive knowledge into uncontrolled documents or general-purpose AI tools. Knowledge transfer must preserve customer confidentiality, intellectual-property rights, professional obligations, and access restrictions.

ISO 30401 frames knowledge management as an organizational system that must be established, implemented, maintained, reviewed, and improved. Its relevance to S1-07 is that expert bottlenecks should produce a managed response—training, documentation, role design, decision rights, software controls, or succession—not only a complaint that senior people are busy.

Limitations, controversies, and open questions

Cost precision can be false precision

Capacity rates depend on judgment. Decisions about practical capacity, overhead allocation, management time, software cost, bench time, and post-delivery support can materially change project margins. Two internally consistent models can produce different answers because they address different decisions.

The remedy is not to pretend that one allocation is objectively perfect. Document the policy, test sensitivity, and retain more than one economic view when necessary. A price decision may require full-cost economics; a short-term capacity decision may require incremental cost; a productization decision may require analysis of which costs would disappear, remain, or move into product development and support.

Timesheets are incomplete representations of knowledge work

People forget, round, code strategically, or conform their entries to the sold budget. Passive system data can also mislead: an open document does not prove active work, and an automated task may consume no visible employee time while incurring material software or review cost.

The strongest approach triangulates self-reported time with process maps, calendars, project-management records, ticket histories, document activity, invoices, and interviews. Recent systematic reviews of time-driven costing repeatedly identify process mapping, direct observation, expert input, and transparent capacity-rate documentation as important, while also finding inconsistent implementation and reporting.

A low number of expert hours is not always better

High-risk, novel, regulated, or strategically important engagements may correctly require substantial expert involvement. Removing experts too aggressively can increase error, rework, customer harm, legal exposure, or reputational risk.

The better question is:

Which expert interventions create disproportionate value or reduce material risk, and which exist because the delivery system has not yet made routine work transferable?

Measure outcomes alongside hours. A process that saves ten expert hours but increases defects, escalations, or customer churn is not an improvement.

Margin can be improved in harmful ways

A fixed-price project can show higher short-term margin if the team skips testing, limits communication, defers maintenance, or pushes unresolved work into “support.” Likewise, a manager can improve reported utilization by discouraging training and internal improvement.

Financial measures must therefore be paired with quality, customer-result, rework, support, and employee-sustainability indicators. Accenture’s warning that both overly high and overly low utilization can damage quality, engagement, attrition, and staffing illustrates why one-dimensional optimization is dangerous.

Historical data may not represent the future

A founder-led engagement completed before a new employee, workflow, product feature, AI tool, price change, or customer segment may not predict current economics. Historical baselines should be versioned according to meaningful operating changes.

When a process changes, retain the old baseline, define the change date, and compare subsequent cohorts. Do not overwrite historical data to make the new model appear successful.

Important open questions remain

The evidence supports activity and capacity costing, but several questions require company-specific judgment:

• How should the economic value of expert judgment be separated from the time it consumes?
• How much practical capacity should remain available for urgent work, training, presales, and improvement?
• How should shared software, proprietary assets, AI inference, and product-development costs be allocated?
• Which customer complexity factors reliably predict effort before a proposal is signed?
• When does process standardization improve quality, and when does it suppress necessary judgment?
• Which expert tasks can be transferred through training, decision rules, software, or better customer inputs?
• How should included support and long-tail warranty work be assigned to the original engagement?
• Does a higher project margin correlate with customer retention and successful outcomes, or merely with narrower service?
• How much data is enough before the company changes a price, scope, or staffing model?

The current literature demonstrates that detailed costing can reveal cost drivers and inefficiency, but it also shows inconsistent methodological reporting and limited evidence about whether organizations routinely convert the analysis into sustained operating improvement.

Adoption and monitoring recommendations

Build the first baseline from completed work

Use every recent completed engagement that is sufficiently comparable, then work backward from customer result to delivery activities. Do not begin by designing a complex future system.

For each engagement:

1. confirm the customer promise, scope, fee model, and approved changes;
2. determine economically earned or recognized revenue;
3. reconstruct actual activities, roles, and elapsed time;
4. calculate capacity-based labour cost and other attributable cost;
5. identify expert interventions and why they were required;
6. map handoffs, delays, rework, and included support;
7. calculate contribution and project gross margin under documented definitions;
8. record complexity factors and data confidence.

The first pass may be manual. A spreadsheet or database with a stable schema is preferable to purchasing a large system before the team agrees on definitions.

Review by engagement type, not only by project manager

The review should answer four levels of question:

Engagement: Did this project perform as expected, and why?

Type: Is this category reliably profitable and deliverable?

Portfolio: Which mix of work best uses the company’s constrained capacity?

Operating model: What should be standardized, delegated, automated, turned into software, repriced, narrowed, or stopped?

Compare median results and dispersion. An engagement type with a 45% median contribution margin and little variation may be more dependable than one with a 55% average produced by a few successes and several losses.

Turn expert-hour evidence into a work redesign decision

For each major expert activity, select a response:

Do not set an arbitrary reduction target before this classification. The desired number of expert hours follows from risk, quality, price, and process design.

Establish a recurring review cadence

Update the analysis when engagements close and perform a broader review at a cadence suited to deal volume. High-volume, short engagements may justify monthly review. A company with a few long projects may need milestone-based estimates and quarterly portfolio review.

Watch leading and lagging indicators together:

• proposal estimate versus current estimate at completion;
• constrained-role demand for the next several weeks;
• scope changes and unresolved customer dependencies;
• current rework and escalation;
• completed-project margin;
• expert hours per completed deal;
• customer outcome, satisfaction, renewal, or follow-on work;
• employee workload and unrecorded effort;
• data completeness.

The review should produce a decision, not merely a dashboard. Examples include changing an exclusion in the statement of work, introducing a paid discovery stage, raising the price for a complexity factor, replacing an expert review with a quality gate, requiring customer data before kickoff, or discontinuing an engagement type whose economics cannot be repaired.

Define completion in decision terms

S1-07 is complete enough to rely on when leadership can answer, with evidence:

• Which engagement types consume the most total and expert time?
• Which produce acceptable contribution and gross margin?
• What explains the difference between profitable and unprofitable cases?
• Where do work, information, and approvals wait?
• Which handoffs generate rework?
• Which senior interventions genuinely require senior judgment?
• Which engagement types can accept more sales without overloading a constrained person?
• Which scope, price, staffing, process, or product change should be tested next?
• How confident is the company in each conclusion?

“Baseline captured” does not mean that a target margin or expert-hour ratio has been achieved. It means the current operating reality has been measured well enough to support a decision. The next move should depend on that reality: strengthen a profitable engagement type, repair a promising but inconsistent one, narrow or reprice a poor one, or stop selling work that consumes scarce expertise without producing sufficient customer and financial value.

Sources

Primary and official sources

• Accenture, 2025 Form 10-K. Primary evidence on contract models, revenue estimation, utilization, delivery costs, gross margin, talent constraints, and automation risk. Credibility: Very high for Accenture’s operations; limited generalizability.
• ICF International, 2025 Form 10-K. Primary evidence on revenue mix by time-and-materials, fixed-price, and cost-based contracts. Credibility: Very high for the reported company figures; limited as a benchmark.
• IFRS Foundation, IFRS 15 supporting material and post-implementation review. Official history and current status of the revenue-recognition standard. Credibility: Very high.
• International Organization for Standardization, ISO 30401 and draft second edition. Official status of the knowledge-management standard and its revision. Credibility: Very high.
• European Union, General Data Protection Regulation, Article 5. Primary legal text on lawfulness, purpose limitation, and data minimization. Credibility: Very high within its jurisdiction.
• UK Information Commissioner’s Office, monitoring workers guidance. Official regulatory guidance on lawful, fair, proportionate, and transparent monitoring. Credibility: Very high for UK data-protection practice; currently noted as under review.
• U.S. Department of Labor, recordkeeping guidance. Official guidance on working-hour and wage records for covered employees. Credibility: Very high within U.S. federal jurisdiction.
• National Institute of Standards and Technology, Cybersecurity Framework 2.0 and least-privilege definition. Official security-risk and access-control guidance. Credibility: Very high.

Open research

• Kaplan and Anderson, “Time-Driven Activity-Based Costing,” Harvard Business School Working Paper, 2003. Original formulation of the method. Credibility: High; primary scholarly source from the method’s authors.
• Harvard Business School Working Knowledge, “Adding Time to Activity-Based Costing,” 2007. Author interview explaining capacity cost rates, time equations, and the method’s development. Credibility: High for conceptual history; institutional explanatory source rather than peer-reviewed evaluation.
• “Time-Driven Activity-Based Costing and Its Use in Health Economic Analysis,” systematic review, 2025. Evidence on applications, benefits, reporting quality, and implementation limits across 32 studies. Credibility: High; peer-reviewed synthesis, although sector-specific.
• “Cost Measurement in Value-Based Healthcare,” systematic review, 2022. Evidence on process mapping, observation, expert input, full-cycle costing, and managerial usefulness. Credibility: High; peer-reviewed synthesis, with transferability limits outside healthcare.
• “Is It Possible to Automate the Discovery of Process Maps for TDABC?” systematic mapping review, 2023. Evidence on process-map construction, operational-data use, and validation gaps. Credibility: High; peer-reviewed synthesis, although healthcare-focused.
• “Evaluation of Reporting in TDABC Studies on Cardiovascular Diseases,” scoping review, 2025. Evidence on inconsistent capacity and cost reporting. Credibility: High for methodological limitations; narrow application domain.
• “Improvements in Technology and the Expanding Role of TDABC,” literature review, 2024. Evidence on the scale of manual and technology-supported implementations. Credibility: Moderate to high; useful implementation synthesis, but less authoritative than a systematic effectiveness review.