Your teams produce evidence faster than anyone checks it.

Efestra checks the experiments and research behind decisions you sign off, and reports the findings to you in confidence.

Request a call Take the self-assessment

24 pieces of evidence this quarter

3 checked outside the team that made them

Illustration, not client data

AI changed how evidence is made. The checking stayed the same.

	Two years ago	Now
Test analysis	An analyst, over days	An AI assistant, in hours
Research synthesis	A researcher with transcripts	AI summaries of transcripts
Record keeping	One shared tool	A tracker per team
Who checks it	The team that made it	Still the team that made it

What we check behind the claims you are given

What we check

Whether it ran for its planned time, on a large enough sample, against the metric agreed before it started.

What we often find

Tests stopped as soon as the result looked good.

What you receive

A written finding for each decision, with the reasons and the source evidence behind it.

Evidence audit report, decision 2 of 3Illustrative example, not a client audit

Expand the checkout redesign to all markets

Reported conclusion: Conversion improved 8%
Evidence status: Insufficient support
Sources reviewed: Test plan, results export, two summary decks

Why

The records do not show that the planned sample size was reached, or that the result was judged against the success measure agreed before launch.

Recommended action

Verify the original analysis before using this result to justify a wider rollout.

Choose. Trace. Audit. Report.

Each audit is scoped by the number of decisions examined, not the size of your organisation. The fee is fixed and agreed before work begins. We work from read-only access, and nothing in your tools is changed.

decisions

weeks from access to report

£4,500

starting fee

Choose
You name three recent decisions that mattered and relied on experiments, research or analysis.
Trace
We follow each one to the work underneath, in Jira, Airtable, documents or wherever it is kept.
Audit
We check how each piece of evidence was designed, run and reported.
Report
A written finding on each decision, sent to you in confidence. You decide who else sees it.

If you received our printed guide, this is the audit described in Part Four.

How exposed are you to the evidence you approve?

Ten questions about what you can see from your position. Answer from what you know today. If you would need to ask someone else, mark Not sure.

There is no score and no email needed. At the end you see the questions you could not answer, and what an audit would examine for each.

Where the evidence comes from1 of 10

For the last major decision you approved, could you name the specific experiment, study or analysis it relied on?

Yes Not sure No

Who carries out the audit

Manuel da Costa

Founder, Efestra (formerly Effective Experiments)

For more than a decade, I’ve worked on how experimentation and research evidence is documented, governed, challenged and used to make decisions.

Efestra grew from software built around quality gates, evidence integrity checks, approval workflows and programme-level oversight. This service brings that experience directly to an independent review of the evidence behind decisions, without requiring teams to change tools.

Where a review requires specialised statistical or research-methodology expertise, that work is scoped appropriately rather than implied to be covered by a generic checklist.

Manuel on LinkedIn

More than a decade working in experimentation and research governance
Enterprise software and operational experience across multiple sectors
Review conducted separately from the team producing the evidence
Confidential findings provided to the commissioning organisation

Questions leaders ask before an audit

Will you tell us whether our decisions were right?

No. The audit reports on the evidence behind the decisions, and whether it supported the confidence placed in it.

Will my teams feel audited?

The audit examines evidence, not people, and does not assess individual performance. Where documentation is unclear, we ask short factual questions. You decide who reviews the findings.

What access do you need?

Read-only access to the places where the relevant experiments and research are recorded, such as Jira, Airtable, research repositories and shared drives. We do not change any data.

Could we not do this internally?

Your teams can build a tracker or an AI workflow in an afternoon. What they cannot build is independence: the people whose work is examined would be setting and applying the standard.

What happens after the audit?

You keep the report. If you want the evidence behind future decisions checked in the same way, periodic reviews are available after the first audit.

What the review covers

Method and design
Evidence provenance and changes
Interpretation and reporting
Claimed commercial impact
Governance and approvals
Findings and recommendations

What isn't included

Replacing your tools
Implementing new workflows
Continuous monitoring
Formal certification
Unbounded data re-analysis
Security or privacy certification

Start with three recent decisions.

For product, data, marketing and finance leaders who approve decisions based on evidence they did not produce.

Leave your details and Manuel will reply by email to arrange a 15-minute call. Or write to manuel@efestra.com.