See the Business as It Is

Task

Document the current revenue mix by consulting, product, and recurring revenue.

Summary

Document the current mix of service, product, and recurring revenue before deciding what should change.

Know What the Business Is Actually Selling

Task ID: S1-01

A revenue-mix baseline shows what customers paid for during the last 12 months, which parts of that revenue depended on people, which parts came from products, and which parts repeated. Done well, it replaces a vague “services versus software” story with reconciled evidence that can guide offer design, product investment, staffing, and customer focus.

A founder may describe the company as a software business while most invoices still pay for discovery, configuration, implementation, training, and ongoing support. Another company may call itself a consultancy even though a large share of its revenue now comes from retainers or subscriptions. Total revenue can rise in both cases while hiding the more important question: what, exactly, are customers continuing to pay for?

The first useful answer is not a target mix borrowed from another company. It is a reliable picture of the company’s own last 12 months. That picture should show revenue by source, customer segment, delivery model, and recurrence, with enough cost and concentration context to prevent an attractive percentage from being misread.

The baseline is a decision tool, not an accounting exercise

The operating principle is simple:

Classify 12 months of recognized customer revenue by what was delivered and whether it repeats, reconcile the result to the financial records, and then examine the mix by customer segment.

“Recognized revenue” matters because bookings, invoices, cash receipts, annual recurring revenue, and accounting revenue answer different questions. Under the common revenue-recognition model in International Financial Reporting Standard 15 and U.S. accounting standard Topic 606, revenue depicts the transfer of promised goods or services to customers, not merely the signing of a contract or receipt of cash. The standards also address contracts containing several promised items, which is common when software, implementation, support, and training are sold together.

This task belongs early in a company’s development because later decisions depend on it. A team cannot sensibly decide which service to standardize, which repeated task to turn into software, or which customer segment deserves focus when it cannot distinguish product revenue from labour-heavy delivery. The baseline does not prove that a standalone product should be built. It reveals where to investigate.

The completed work should support a specific decision: Is there enough repeated financial and customer evidence around one customer, problem, and result to build a more repeatable offer? Revenue alone cannot answer that question, but it can show whether the proposed direction is grounded in actual buying behaviour or mainly in aspiration.

The working target, “baseline captured,” is therefore a completion condition rather than an industry benchmark. At this stage, success means the company can explain and reproduce its revenue mix. It does not mean services must fall below a particular percentage.

Separate what is sold from how it repeats

The labels “consulting,” “product,” and “recurring revenue” look like three categories, but they are not naturally mutually exclusive. Consulting describes what the customer receives. Recurring describes how the commercial relationship repeats. A monthly managed service is both service revenue and recurring revenue. A software subscription is both product revenue and recurring revenue.

Putting all three labels into one column creates double counting or forces arbitrary choices. A better model uses two dimensions:

1. Delivery model: service, product, or another clearly defined source.
2. Revenue pattern: recurring, usage-based, repeat purchase, or one-time.

Start with mutually exclusive operating buckets, then calculate the headline percentages from flags attached to each bucket.

This distinction reflects the broader accounting principle that revenue should be separated into categories that show how economic factors affect its nature, timing, and uncertainty. Official implementation guidance lists type of good or service, customer type, contract type, duration, timing, and sales channel as potentially useful categories, while deliberately avoiding one mandatory scheme for every business.

A practical classification flow looks like this:

flowchart LR
    A[General ledger and billing records] --> B[Normalize customers and contract lines]
    B --> C[Split mixed product and service items]
    C --> D[Assign one operating bucket]
    D --> E[Add service, product, and recurring flags]
    E --> F[Group by month and customer segment]
    F --> G[Reconcile totals to reported revenue]
    G --> H[Review mix, margin, and concentration]

In plain language, the company starts with financial transactions, cleans the customer and contract data, separates mixed sales, assigns each amount to one base category, and only then creates service, product, and recurring rollups.

The classification policy should be written before the team studies the result. Otherwise, the desire to show progress toward “product revenue” can influence how ambiguous transactions are labelled.

What research and public accounts show

Research does not support a universal ideal service-revenue percentage. A meta-analysis of 41 peer-reviewed studies found an overall positive relationship between adding services and company performance, but the strength of the relationship changed with the way service activity and performance were measured, as well as by industry and region. That finding is useful precisely because it resists a simple rule such as “less service is always better.”

Customer-level analysis also becomes more important when revenue is unevenly distributed, customer-specific overhead is high, and service complexity varies. A 2024 case study found that customer profitability analysis became more sophisticated under those conditions and worked better when financial data was combined with non-financial information and local operating knowledge. The study was based on one manufacturing company, so its exact practices should not be treated as universal, but the decision lesson travels well: revenue by customer is not enough when different customers consume very different amounts of work.

Another open study applied recency, frequency, and monetary-value analysis to service customers at two machinery manufacturers. The method helped expose patterns in customer behaviour and service potential, but the researchers had to adapt it to the business context and add information about the installed product base. The cases used substantial transaction data, including more than 100,000 orders in one company, yet remained limited to two industrial settings.

Public company accounts make the same point in more concrete terms: revenue categories have different economics.

For the year ended March 31, 2026, hospitality-software company Agilysys reported 12.9% of net revenue from products, 64.5% from subscription and maintenance, and 22.6% from professional services. Its reported gross margins were 79.4% for subscription and maintenance and 27.3% for professional services. The revenue percentages alone would not reveal that large difference in contribution.

Salesforce provides a useful counterweight to the idea that low-margin services are necessarily a mistake. For the fiscal year ended January 31, 2026, it reported $39.388 billion of subscription and support revenue and $2.137 billion of professional services and other revenue, approximately 95% and 5% of total revenue respectively. Salesforce also stated that its professional-services organization helps customers adopt its offerings, supports larger subscription contracts, and may cost more than the professional-services revenue it generates.

The lesson is not to copy either mix. Agilysys shows why revenue needs margin context. Salesforce shows that a service line can have strategic value beyond its direct margin. A young founder-led company may reasonably rely on services to learn, implement, and win trust. The danger is not the existence of services. The danger is being unable to tell whether they are profitable work, deliberate adoption support, hidden product development, or unlimited custom effort.

Build a credible twelve-month revenue baseline

Use the most recent 12 completed months unless seasonality, a major acquisition, a change in accounting policy, or an unusual disruption makes another period more representative. Record the exact start and end dates. Do not silently mix a calendar-year sales report with a fiscal-year income statement.

Choose one accounting basis. The baseline should normally begin with recognized revenue from the general ledger because that total can be reconciled to the company’s financial reporting. Bookings and cash collections may be added as separate views, but they should not be blended into the revenue total. A signed annual contract, an annual invoice, and 12 months of recognized subscription revenue can all have different amounts in the same period. The common revenue standard exists in part to create consistent treatment of when promised goods and services transfer to customers.

Extract at the lowest useful level. The ideal source is a revenue or invoice-line export that can be connected to customer, contract, product or service code, recognition month, and amount. When the accounting system contains only broad accounts, use billing records and contracts to create a mapping, then reconcile back to the ledger. Preserve the original source field so reviewers can trace every transformation.

A practical baseline table usually needs the following fields:

Write the data dictionary before classifying exceptions. Define what counts as consulting, managed service, product, subscription, maintenance, usage revenue, pass-through revenue, and other. Define “recurring” narrowly enough that another person could repeat the classification. A good definition usually requires an active contractual or established repeating obligation, not merely the hope that a customer will buy another project.

Split mixed contracts. A single invoice may combine software access, configuration, custom development, data migration, training, and support. The accounting standards treat distinct promised goods and services as separate performance obligations and allocate consideration across them when required. The management baseline does not need to recreate technical accounting work, but it should use the approved accounting allocation where available and avoid assigning an entire mixed sale to whichever label appears first on the invoice.

Choose customer segments that explain buying behaviour. Useful segments might reflect company size, industry, use case, buyer role, implementation complexity, geography, or another factor that changes the problem, sale, price, delivery, or retention pattern. Start with a small number that management can define consistently. Accounting guidance likewise recognizes customer type and market as legitimate ways to disaggregate revenue, but it leaves the choice to company-specific facts.

Reconcile and review. The sum of classified revenue must equal the 12-month recognized revenue total, subject only to documented exclusions such as interest, grants, or non-customer income. Review all unclassified amounts, negative entries, credits, refunds, intercompany entries, and unusually large transactions. Have finance or the person responsible for the books approve the reconciliation, and have sales or delivery leaders approve the commercial classifications.

Freeze a version. Record the extraction date, source systems, definitions, mapping table, unresolved exceptions, and approvers. A baseline that changes every time someone opens the spreadsheet cannot support a decision. Corrections are allowed, but they should create a new dated version with a short change log.

Measure the mix without mistaking it for the answer

The primary measure is:

This percentage includes both one-time consulting and recurring managed services if both involve service delivery. It should therefore be accompanied by a separate recurring-revenue percentage:

The two percentages are allowed to overlap. That is a feature, not an error. A recurring retainer contributes to both. Reporting them as if they must add to 100% would erase the distinction between delivery model and revenue pattern.

Add product revenue percentage, revenue by customer segment, gross margin by bucket, and concentration by the largest customers. Public-company rules require disclosure when one external customer supplies at least 10% of revenue; that threshold is not a private-company performance target, but it is a useful prompt to investigate dependence and negotiating risk.

Annual recurring revenue (ARR) can be useful as a separate point-in-time operating measure, but it should not replace the 12-month revenue baseline. Amplitude’s 2025 annual filing explicitly says its ARR is based on subscription agreements at a point in time, should be viewed independently of accounting revenue, does not represent annualized U.S. accounting revenue, and is not a revenue forecast. The company also notes that similarly titled operating metrics lack uniform calculation standards.

A credible “baseline captured” result should include three layers:

The task is complete when the company has a reconciled 12-month dataset, documented category and segment definitions, separate service/product/recurring flags, an exception log, and a repeatable summary by source and customer segment. It is stronger when gross margin, direct delivery effort, and customer concentration are included. It is not complete merely because a chart has been produced.

Failure modes and judgment calls

Using invoice labels as truth. “Platform,” “solution,” or a product name may describe a sale that required extensive custom work. Classification should follow what was delivered, supported by the contract, work records, and accounting treatment.

Calling repeat purchases recurring revenue. A customer who bought three separate consulting projects may be loyal, but the revenue is not necessarily contractually recurring. Track repeat purchase behaviour separately from recurring obligations.

Treating all recurring revenue as scalable. A managed service can be predictable and valuable while still requiring more staff as revenue grows. Recurrence improves visibility; it does not by itself remove labour dependence.

Replacing revenue with ARR, bookings, or cash. Each can be useful, but each answers another question. Mixing them makes period comparisons unreliable. The baseline should state its basis and keep operating metrics separate.

Ignoring direct cost and senior effort. Two customer segments may generate equal revenue while one needs twice the delivery time, founder intervention, or subcontractor cost. Research on customer profitability analysis shows why financial totals become less informative as customer-specific overhead and service complexity rise.

Designing segments after seeing the answer. Segments should reflect a business distinction that changes buying or delivery behaviour. Constantly redrawing them to make a preferred segment look attractive turns analysis into advocacy.

Applying today’s customer label to historical revenue. A company that is now “enterprise” may have been a much smaller customer when the revenue was earned. Preserve the segment definition and classification date, especially when the company is studying movement between segments.

Hiding exceptions in “other.” “Other” should be small, reviewed, and explained. If it becomes material, create a new category rather than allowing it to obscure an emerging revenue source.

Assuming a high service percentage is bad. It may indicate strong consulting demand, a deliberate implementation strategy, or a market where expert work is integral to the result. It may also indicate that the supposed product cannot yet deliver value without custom intervention. The percentage identifies the question; margin, delivery effort, customer results, and repeatability answer it. Research on service expansion finds positive average effects but meaningful variation across settings, reinforcing the need for company-specific judgement.

Assuming a low service percentage is good. A company can have mostly product revenue and still suffer from weak retention, high support cost, concentrated customers, or poor margins. Salesforce’s public explanation of professional services illustrates that even a relatively small service line can play an intentional role in adoption and larger subscription sales.

The central tradeoff is between simplicity and decision value. Too many categories create endless debate and sparse data. Too few categories conceal differences that matter. The right baseline is the simplest model that still distinguishes revenue sources with different customer behaviour, delivery work, margins, timing, or risk.

The decision that should now be clearer

When this work is finished, the company should have one dated, reconciled 12-month revenue file and one concise summary showing:

• revenue by mutually exclusive operating bucket;
• service, product, and recurring revenue percentages;
• revenue by customer segment and source;
• gross margin or a clearly stated cost limitation;
• major customer concentration;
• classification rules, owners, assumptions, and unresolved exceptions.

The result should make a practical decision easier. Management can see whether one customer segment repeatedly pays for the same problem and result, whether that revenue comes mainly from custom work or a repeatable offer, and whether the apparent opportunity remains attractive after delivery cost and concentration are considered.

That is the value of “baseline captured.” It does not declare the company ready to become a product or subscription business. It establishes a shared factual starting point. The next move can then be based on what customers have already paid for, rather than on what the company hopes its revenue mix will become.

Sources

Primary and official sources

• IFRS Foundation, IASB and FASB Issue Converged Standard on Revenue Recognition and the 2024 post-implementation review of IFRS 15.
• Financial Accounting Standards Board guidance as reproduced in Deloitte’s public accounting research pages on revenue disaggregation and major-customer disclosure.
• Agilysys, fiscal 2026 Form 10-K.
• Salesforce, fiscal 2026 Form 10-K.
• Amplitude, fiscal 2025 Form 10-K.

Open research

• Wang, W., Lai, K. H., and Shou, Y. (2018), The Impact of Servitization on Firm Performance: A Meta-Analysis.
• Stormi, K., Lindholm, A., Laine, T., and Korhonen, T. (2020), RFM Customer Analysis for Product-Oriented Services and Service Business Development.
• Lueg, R., and Ilieva, D. (2024), Customer Profitability Analysis in Decision-Making—The Roles of Customer Characteristics, Cost Structures, and Strategizing.