Find the Segments That Produce Value

Task

Segment past projects by industry, buyer, ACV, margin, effort, and outcome.

Summary

Use past work to identify combinations of buyer, problem, price, margin, effort, and outcome that are worth repeating.

Find the Projects Worth Repeating

Task ID: S1-02

A busy project pipeline can hide a weak business model. This article shows how to turn completed work into evidence: classify projects by customer and buyer, calculate the real cost and margin of delivery, separate project efficiency from customer outcomes, and rank the few patterns that are most credible as repeatable offers.

A founder-led company can finish a strong year, look at a list of recognizable customers, and still be unable to answer a basic question: Which kind of project should we deliberately sell again?

The largest contract may have consumed the founder for six months. The easiest client may have bought only once. A project that finished on time may not have produced a measurable customer result. A smaller engagement may have quietly delivered the best margin, the fastest sale, the clearest outcome, and the strongest referral.

That is the problem this work solves. The company needs to stop treating its project history as a collection of stories and start treating it as operating evidence.

The central principle is simple:

Segment past projects by the customer situation, the buyer, the economics, the delivery work, and the result—not by revenue or industry alone.

This analysis belongs near the beginning of a company’s move from custom consulting toward a repeatable offer. It should happen before the company chooses a narrow target market, fixes a standard scope, or turns part of the work into software. It assumes completed client work and imperfect but recoverable records; the analysis should expose those imperfections rather than hide them.

What the project list does not tell you

Most project records were created to sell, invoice, staff, or close work—not to compare business models. A customer relationship management system may contain industry and contract amount; time sheets may omit rework; accounting may miss unrecorded founder time; and a closeout email may praise the work without naming the operational result. These records do not automatically reveal a repeatable pattern.

Revenue is especially easy to misread. Customer-profitability analysis exists because a high-revenue customer can consume enough service, exceptions, support, and overhead to become less attractive than the top line suggests. Activity-based management assigns costs through the activities that customers cause; professional accounting guidance explicitly notes that apparently “best” customers can become loss-making after those costs are assigned.

Annual contract value (ACV) can also create false comparability. ACV is not a standardized accounting measure. Public companies define it under their own operating policies. Ansys states that ACV has no comparable generally accepted accounting principles measure and should be viewed independently of revenue. Schrödinger defines ACV differently for annually billed and prepaid multiyear contracts, then tracks it by industry and customer cohort to understand sales cycles, contract duration, deployment, renewal, and expansion.

For a project business, this leads to a practical rule: record the signed contract value, realized project revenue, and ACV separately. Use ACV only where there is a documented annualization policy. Do not convert a one-time implementation into an annual figure merely to make it look comparable with a recurring contract.

Industry labels are useful, but they are only one layer. A chief financial officer buying a cost-control diagnostic, an operations leader buying workflow redesign, and a technology leader buying an integration may work in the same industry but create entirely different sales and delivery patterns. Research on business-to-business buying treats the “buying center” as the people involved in a particular purchase and emphasizes its formation, internal dynamics, outcomes, and context. Related customer-journey research distinguishes the people who buy from the people who use and judge the solution after purchase.

The unit of analysis, therefore, is not simply “a customer.” It is a project pattern:

A particular kind of buyer, in a particular customer context, paying to solve a particular problem through a recognizable scope and delivery method, with observable economics and an observable result.

That definition is narrow enough to compare work and broad enough to find repetition.

What the research changes about the analysis

The evidence across management accounting, project management, professional services, and organizational buying points to a combined method rather than a single score.

Time-driven activity-based costing is especially useful when detailed costing feels too heavy. Its core method requires two estimates for each resource group: the cost per unit of practical capacity and the time consumed by products, services, or customers. That creates a workable bridge between time records and project economics.

The project-success finding is equally important. In a study of 1,386 projects, efficiency against time, scope, and budget correlated at 0.6 with overall project success and explained 36% of its variation. The exact measures and context limit how far that statistic can be generalized, but the operating lesson is strong: a project can be well managed without producing the result that justified the purchase. Government benefits-management guidance makes the same distinction operationally by requiring teams to define expected benefits, responsibilities, measurement, and realization across the project lifecycle.

Professional-service research adds the repeatability test. A service becomes more product-like when the firm specifies and standardizes the offering, makes expertise and deliverables tangible, and systemizes its methods. A profitable project is therefore not automatically a repeatable offer. Repeatability also requires enough common work that the company can sell and deliver the next engagement without redesigning everything.

The development of these ideas can be summarized as follows:

timeline
    title Measurement ideas that support project segmentation
    2005 : Time-driven costing simplifies the measurement of customer and service effort
    2011 : Professional-service research identifies standard offers and systemized delivery as productization practices
    2014 : Large-sample project research separates delivery efficiency from broader success
    2023 : Buying-center and customer-journey research reinforces the importance of buyer and user roles
    2026 : Current public-company disclosures continue to show that operating metrics such as ACV require company-specific definitions

Build one row per completed project

Start with completed projects, not ideal customers or future market claims. Include cancelled and troubled projects when records exist. Excluding failures produces a flattering model rather than a useful one.

Give each project one row. Keep raw project data separate from segment summaries so that assumptions can be corrected without rebuilding the analysis. Use a stable industry classification where possible—for example, a North American Industry Classification System code—while retaining the company’s plain-language label. Official classifications reduce spelling variants and make later grouping more consistent.

A strong project-level table contains the following fields:

Define the financial fields before calculating them

Gross profit margin relates direct production or service costs to revenue and is commonly calculated as:

That is consistent with the standard gross-margin definition of revenue less cost of goods or services sold, divided by revenue. The company’s finance lead should reconcile the project view to the general ledger and document what counts as a direct cost.

For managerial analysis, direct delivery cost usually needs more detail than invoices provide. It should include the loaded cost of employee delivery time, contractors, project-specific cloud or software, travel, external data, materials, and rework or post-launch support that was part of the promise. Sales commissions, general marketing, office rent, and broad administration normally belong outside project gross margin, but a second contribution view can subtract customer-specific sales, account-management, and support costs. This keeps gross margin comparable while still exposing expensive customers.

Founder time requires special treatment. If the founder delivers without recording time or taking market-rate compensation, the accounting record can make a project look artificially profitable. Preserve the booked accounting view, then add a decision view that assigns founder and other unrecorded senior time a reasonable replacement-cost rate. Label the estimate. Do not quietly mix it into the ledger.

Time-driven costing offers a practical calculation:

Use practical hours rather than every paid hour, because holidays, internal meetings, administration, training, and normal downtime reduce available delivery capacity. The goal is not false precision. It is consistent comparison.

Record outcome evidence in levels

Do not force every project into a financial return-on-investment claim. Some work improves compliance, reduces risk, speeds decisions, raises capacity, or creates benefits whose financial effect is difficult to isolate.

Use an evidence ladder:

• Output only: the agreed deliverable was completed.
• Adoption: the customer used the deliverable or changed a process.
• Operational result: a relevant measure changed, such as cycle time, error rate, conversion, throughput, or support volume.
• Financial result: the customer documented revenue, savings, avoided cost, or another financial effect.
• Strategic or risk result: an authorized customer owner confirmed a defined strategic, regulatory, or risk benefit.

Store the baseline, follow-up period, source, and customer owner. A testimonial is useful market evidence, but it is not a substitute for an outcome measure. Likewise, a renewal or referral supports the conclusion that the customer valued the relationship, but it does not by itself prove the promised operating result.

Turn project rows into comparable segments

First group projects by industry + buyer role + problem or use case. Then test whether customer size, geography, technology environment, trigger event, or delivery model separates materially different economics. Create a segment only when the distinction could change whom to pursue, what to promise, how to price, or how to deliver.

flowchart LR
    A[Completed projects] --> B[Normalize customer, buyer, and problem data]
    B --> C[Calculate actual effort, cost, and margin]
    C --> D[Group similar project patterns]
    D --> E[Compare outcomes and repeatability]
    E --> F{Evidence consistent enough?}
    F -->|Yes| G[Rank candidate segments]
    F -->|No| H[Mark as a hypothesis]
    H --> I[Collect better data or test with new sales]

For each candidate segment, produce a roll-up with:

Showing both portfolio-weighted and median project margin matters. The weighted figure answers, “How much gross profit did this segment produce in total?” The median and range answer, “What did a typical project look like, and how variable was it?” A single large contract should not be allowed to conceal four small losses.

Small samples require restraint. Official statistical guidance warns that smaller achieved samples create more volatile estimates and should be treated with additional caution. A company does not need a universal minimum number of projects before learning anything, but it should label a pattern based on one or two engagements as a hypothesis rather than a proven segment. Confidence should rise with more projects, better records, more recent work, and consistent results.

Rank the top candidates with gates, not a magic formula

“Top three segments ranked” is a useful working target because it forces comparison and creates a manageable leadership discussion. It is not an industry benchmark and does not mean that every company should pursue three markets. A narrow firm may have one credible pattern. A diverse project company may need more than three candidates before choosing one.

Use a two-stage decision.

First, apply evidence gates. A segment should not rank as “proven” when its revenue cannot be reconciled, delivery effort is mostly estimated, the outcome is unknown, or one account supplies nearly all the evidence. It may still be strategically interesting, but it belongs in the hypothesis column.

Second, compare the segments across five dimensions:

1. Economics: gross profit, gross margin, pricing stability, cash and payment behavior, and customer-specific cost to serve.
2. Repeatability: common scope, predictable work, reusable assets, low rework, and limited dependence on the founder.
3. Outcome strength: frequency, magnitude, and credibility of customer results.
4. Demand evidence: repeat purchases, expansions, referrals, sales speed, and resistance to discounting.
5. Strategic fit: whether the company has a credible advantage, access to the buyers, and a practical path to serve more of them.

Weights are management choices, not scientific constants. A cash-constrained company may emphasize margin and payment speed; a company seeking a software opportunity may emphasize repeated workflow and customer outcome. Test at least two reasonable weighting schemes. If the winner changes with small adjustments, the evidence is not decisive.

Write each ranked pattern as a testable statement:

Operations leaders at mid-sized distributors buy a fixed-scope workflow diagnostic to reduce order exceptions; recent projects sold within a defined range, used a common delivery method, produced documented cycle-time improvements, and achieved a consistent gross-margin range without heavy founder delivery.

The sentence should be filled with actual ranges and evidence. It should never rely on adjectives such as “ideal,” “high value,” or “easy” without a measure.

A public example of repeated client work becoming a product

Basecamp offers a useful historical example, not a universal formula.

The company that became Basecamp began as a web-design firm. Its official history says that growing project volume made client coordination difficult, so the team built a system out of operational necessity. Independent reporting described the product as an internal client-communication system that customers also wanted.

The lesson is not “build every internal tool.” The lesson is that repeated delivery problems can reveal a product opportunity when several kinds of evidence align:

• the problem recurs across projects;
• the team already uses a common method or tool to solve it;
• customers notice and value that method;
• the solution can be separated from the custom work;
• the economics can improve when delivery is standardized.

A project-segmentation table can show whether the same buyer repeatedly pays for the same result, a common workflow absorbs much of delivery, or a reusable artifact drives outcomes. It can also show that an internal tool solves the provider’s inconvenience without solving a problem customers will buy separately.

Where the analysis usually goes wrong

A polished spreadsheet can still create a bad decision. The most common failures are conceptual, not technical.

The company segments by industry alone. Industry can affect regulation, budgets, language, and buying behavior, but it may not explain the purchase. Add buyer role, problem, trigger, and use context before concluding that an industry is attractive.

The company ranks revenue instead of economics. Signed value, recognized revenue, cash, gross profit, and contribution answer different questions. Keep them separate. Include change orders and write-offs so that weak scoping does not disappear.

Quoted effort replaces actual effort. A statement of work shows the plan. Time records, calendars, tickets, commits, meeting logs, contractor invoices, and delivery-team interviews show what happened. Preserve both planned and actual effort so estimate quality can improve.

Founder work remains invisible. This is especially dangerous in referral-led firms. The founder may sell, diagnose, calm the customer, solve exceptions, review every deliverable, and maintain the relationship. A segment that works only under that subsidy is not yet repeatable.

On-time completion is treated as the customer outcome. Efficiency is important, but research does not support treating it as the whole of success. Record delivery performance and customer benefit separately.

A single “whale” defines the segment. Show concentration, medians, ranges, and results with the largest project removed. If the conclusion collapses, the company has an important customer, not yet a repeatable segment.

Old and new business models are mixed. Pricing, team skill, automation, scope, and market conditions change. Tag the offer version and time period. Historical work may still teach useful lessons, but it should not be averaged uncritically with the current delivery system.

The table becomes too detailed to maintain. Begin with fields that change a decision. Add complexity only when it explains margin, effort, outcome, demand, or repeatability. Standardization is meant to create controllable work, not bureaucracy.

The top-three ranking is treated as a permanent strategy. Segments move, buyer priorities change, and the company’s capabilities improve. Recalculate the analysis on a regular cadence and after material changes to price, scope, staffing, or delivery technology.

There are legitimate exceptions. A low-margin project may create defensible knowledge, open a regulated market, produce a reusable asset, or reach an important buyer. A highly customized engagement may also fit a deliberate premium-project model. Label such work as a strategic bet, define the expected benefit, and review whether it occurred instead of hiding the exception in the core segment.

What should exist before the company builds around a segment

The work is complete when leadership has evidence that can survive questions from sales, delivery, and finance—not when every cell is filled.

The minimum credible package is:

• a project-level table with completed, cancelled, and materially troubled work;
• documented definitions for value, revenue, ACV, delivery cost, margin, founder effort, outcome, and confidence;
• a segment roll-up covering economics, effort, outcomes, demand, repeatability, concentration, recency, and data quality;
• ranked candidates tested against reasonable changes in assumptions;
• an exceptions log for missing records, unusual projects, strategic bets, and estimates;
• a decision note: build around one segment, test several, or collect better evidence.

The primary measure is gross margin by segment, because it connects price to direct delivery work. It should not stand alone. High margin without a documented outcome is fragile; a strong outcome that requires the founder is not yet scalable; a fast sale with heavy rework is not repeatable. Judge the full pattern across economics, effort, result, demand, and confidence.

The final decision should be plain:

Which customer situation, buyer, problem, result, and delivery pattern has enough real evidence to justify focused selling and a more standardized offer?

If one pattern is clearly ahead, the company can use it to shape the next offer. If several are close, it can run deliberate sales and delivery tests. If the records are weak, the result is not failure; it is a measurement agenda. The company now knows which facts must be captured in every new project before it makes a larger commitment.

Sources

Primary and official sources

• U.S. Securities and Exchange Commission, Ansys quarterly filing. High-priority evidence that ACV is a company-defined operating metric, separate from generally accepted accounting revenue.
• U.S. Securities and Exchange Commission, Schrödinger 2025 results and operating-metric definitions, filed February 2026. High-priority current example of an explicit ACV policy and cohort analysis by industry and customer type.
• Institute of Chartered Accountants in England and Wales, KPI glossary. High-priority implementation reference for gross profit margin and cost-to-serve definitions.
• UK Infrastructure and Projects Authority and Cabinet Office, Guide for Effective Benefits Management in Major Projects. Official framework for defining, owning, measuring, and realizing project benefits.
• U.S. Census Bureau, Economic Census and North American Industry Classification System tables. Official reference for using stable industry codes instead of uncontrolled labels.
• Office for National Statistics, regional labour-market quality guidance. Official illustration of why estimates based on smaller samples require additional caution.
• Basecamp, “Where We Came From.” Primary company history for the origin of the product in repeated client-project coordination problems.

Open research and university sources

• Robert S. Kaplan and Steven R. Anderson, “Rethinking Activity-Based Costing” and “Adding Time to Activity-Based Costing,” Harvard Business School Working Knowledge. Foundational practical explanation of assigning service cost from practical capacity and actual time demands.
• Pedro Serrador and Rodney Turner, “The Relationship between Project Success and Project Efficiency,” 2014. Open-access survey of 1,386 projects showing why schedule, scope, and budget performance should not be used as the sole measure of customer success.
• Elina Jaakkola, “Unraveling the Practices of Productization in Professional Service Firms,” 2011. Peer-reviewed study identifying offer specification, tangibilization, and process systemization as central productization practices.
• Pablo Cabanelas, Roberto Mora Cortez, and Jon Charterina, “The Buying Center Concept as a Milestone in Industrial Marketing,” 2023. Open-access systematic review supporting the analysis of buyer roles, decision processes, and organizational context.
• Arttu Purmonen, Elina Jaakkola, and Harri Terho, “B2B Customer Journeys: Conceptualization and an Integrative Framework,” 2023. Open-access framework that connects buying-center and usage-center members across purchase and use.
• Association of Chartered Certified Accountants, “Activity-Based Management.” Practical accounting treatment of customer profitability and customer-caused service costs.

Public reporting

• WIRED, “Work Smarter: 37signals,” 2010. Independent contemporary account confirming that Basecamp grew from an internal system used in client work.