# How to examine a startup's traction evidence

> Read startup traction metrics with clear definitions, cohorts and limitations. Separate registrations, usage, customer commitments and revenue.

Canonical HTML: [https://orbitrum.io/academy/commercial-analysis/startup-traction-evidence/](https://orbitrum.io/academy/commercial-analysis/startup-traction-evidence/)

Prepared for Orbitrum Academy

Published: 2026-10-08

Source-based educational synthesis prepared with AI assistance. Proposed worksheets and fictional examples are labelled. No named expert review is claimed.

Educational material, not personalized investment, legal, tax or safety advice. Research access on Orbitrum depends on the expert's offer.

Startup traction is evidence of how a defined group of customers or users adopts, uses or pays for a product over time. Assess the underlying definitions, period, cohort and supporting records before comparing headline metrics. Registrations, demonstrations, pilot agreements, recurring revenue and cash received answer different questions; none should silently stand in for the others.

## Define the claim before comparing the number

The SEC's 2020 KPI guidance asks issuers to explain a metric's definition and calculation, why it is useful and how management uses it, with relevant context about assumptions or methodology changes. This guide adapts that disclosure discipline to reading startup claims; it does not assert that every private startup is subject to the SEC guidance. [SEC Release 33-10751, pp.3-5](#source-sec-kpi-2020)

Start with a sentence such as "This metric counts paying customer organizations at the end of the stated month." Compare that with "This metric counts all user accounts ever registered." Both may be accurately reported, but they do not describe the same phenomenon.

Ask for a calculation you could reproduce from authorized supporting records. If you cannot reproduce it, preserve the publisher's label and explain what remains unverified.

## Separate the kinds of evidence

The distinctions below are an Academy reading framework, not a universal accounting policy. Contracts and accounting treatment require their own context.

| Reported item | Initial question | Do not assume |
|---|---|---|
| Registrations | Which accounts are counted, and over what period? | That all accounts actively use or pay for the product |
| Active usage | What action and time window define active? | That usage is retained or commercially valuable |
| Pilot | What has been agreed, delivered and paid? | That a pilot is an ongoing customer deployment |
| Signed contract | What commitments and conditions apply? | That the full amount has been earned or collected |
| Reported revenue | Which period and recognition basis apply? | That revenue equals cash received |
| Cash received | What payment and obligation does it relate to? | That it represents recurring revenue |
| Customer logo | What relationship does the underlying statement describe? | That the customer independently endorses every claim |

The point is not to distrust every number. It is to avoid combining unlike evidence into a single growth story.

## Examine a cohort rather than only the total

For this guide, a cohort is an explicitly defined starting group observed over a stated interval. Follow the same group when asking whether customers remain or spend more. Keep newly acquired customers separate when the question concerns the original group's behavior.

Here is a **fictional learning exercise**, not an Orbitrum result or an industry benchmark. The definitions below are chosen for the exercise.

At the start of a month, 10 customer organizations each pay 100 units per month, for 1,000 units in total. At month end, two have left. The remaining eight generate 800 units at their original amounts plus 300 units of expansion. New customers are excluded.

| Calculation in this exercise | Result | Interpretation boundary |
|---|---|---|
| Customers retained / starting customers | 8 / 10 = 80% | Counts organizations, not their revenue |
| End recurring amount from starting cohort / starting amount | 1,100 / 1,000 = 110% | Includes expansion from retained customers; excludes new customers |

The second result does not erase the loss of two customers. The first does not describe expansion. Reporting both makes the difference visible. No conclusion about sustainable growth follows without the period, definitions and wider context.

When the starting denominator is zero or unavailable, mark the ratio as undefined or unavailable. Do not display a reassuring zero or invent a growth percentage.

## Ask whether the measurement changed

If an "active customer" definition changes between periods, an apparent increase may include a measurement change. Record the old and new definitions, the date of the change and whether earlier periods were recalculated. The SEC guidance specifically discusses providing context when metric calculations or presentation change. [SEC Release 33-10751](#source-sec-kpi-2020)

This also applies when reading excerpts from presentations. A graph's labels and footnotes may contain information that a copied headline omits. Preserve the original period and unit in your notes.

## Look for evidence that could change the commercial thesis

Camuffo and colleagues' 2024 replication study concerns hypothesis-driven decisions by entrepreneurs, not a formula for investor returns. Its relevance here is the discipline of making a testable assumption explicit rather than treating a preferred story as settled. [Camuffo et al., 2024](#source-camuffo-2024)

Write a focused hypothesis, for example: "Customers in this specified group continue using the product after the pilot under the stated terms." Then identify what observations would support or weaken it. The exact measurement depends on the product; this guide does not prescribe a universal retention threshold.

A failed hypothesis can tell you where the claim needs narrowing. A positive result should retain the same boundaries. Do not broaden the customer group after seeing a favorable result without explaining that change.

## Build a traction memo

Use a compact structure: customer and use case; metric definitions; period and cohort; supporting records; findings; contradictory or missing evidence; and next questions.

For an expert, attach sources to the conclusions rather than displaying an unexplained growth adjective. For a reader, check whether the memo distinguishes company statements, reproduced calculations and interpretation.

This is a proposed writing format. It is not a claim that Orbitrum stores confidential customer records or independently verifies every metric displayed in a project profile.

## Apply this to public research

Read the available project information and published analyses on Orbitrum, then return to their underlying sources where accessible. Treat an unavailable private contract or customer dataset as a limit on the public review.

The [first-pass due-diligence guide](/academy/general-method/first-pass-project-due-diligence/) helps connect these commercial questions with other review areas. The [NVIDIA filing case](/academy/applied-cases/nvidia-filing-analysis/) shows how reported financial numbers can be separated from calculations and interpretation in a public-company example.

## Questions about traction

### Is there a universal revenue or retention level that proves product-market fit?

No such threshold is established by the sources used in this guide. State the product, customer group and measurement context, rather than importing an unexplained benchmark.

### Does a rising total user count demonstrate retention?

Not by itself. A total can include new users while earlier users leave. Examine the defined starting group if retention is the research question.

### Should an expert request private customer data for a public article?

This guide does not require that. Work only with material you are authorized to access and publish. Describe the limits of a public-source assessment without disclosing confidential or personal information.

## Explore the research on Orbitrum

Browse public project profiles and the research available from their authors. Public previews are free; full research access depends on the expert's offer.

- [Explore projects](https://orbitrum.io/)
- [Discover experts](https://orbitrum.io/?mode=experts)
- [Browse sectors](https://orbitrum.io/sectors/)

## Sources and reading notes

<a id="source-sec-kpi-2020"></a>
### Commission Guidance on Management's Discussion and Analysis of Financial Condition and Results of Operations

Publisher/authors: US Securities and Exchange Commission; Release 33-10751
Date: January 30, 2020
Type: Official disclosure guidance
Locator: Printed pp.3-5; definition, calculation, purpose, assumptions and changes in metrics
Official source: https://www.sec.gov/files/rules/interp/2020/33-10751.pdf

Supports: Transparent definitions and context for KPIs; consistency when calculations change.
Scope and limitations: US issuer disclosure context. Its application to private-startup research is explicitly an educational adaptation, not a statement that every startup is subject to this guidance.
Sources checked: 2026-10-08

<a id="source-camuffo-2024"></a>
### A scientific approach to entrepreneurial decision-making: Large-scale replication and extension

Publisher/authors: Camuffo, Gambardella, Messinese, Novelli, Paolucci and Spina; Strategic Management Journal 45, 1209-1237
Date: 2024
Type: Peer-reviewed study; four randomized controlled trials, 759 firms
Locator: Abstract; sections 4 and 5.5; discussion
Official source: https://sms.onlinelibrary.wiley.com/doi/pdf/10.1002/smj.3580
DOI: 10.1002/smj.3580

Supports: Hypothesis-driven entrepreneurial learning; effects on idea termination and patterns of strategic pivots.
Scope and limitations: Not a trial of investor returns or a proof that a single traction metric predicts success; do not assert guaranteed revenue or funding improvements.
Sources checked: 2026-10-08

## Continue learning

- [How to conduct a first-pass project due diligence](https://orbitrum.io/academy/general-method/first-pass-project-due-diligence/)
- [An annotated project analysis: NVIDIA's FY2026 filing](https://orbitrum.io/academy/applied-cases/nvidia-filing-analysis/)