Results your clients can verify.

Every report carries its test plan, its checks, and its method. Your client reads the same conclusion you do and can trace it back to the calculation.

Create your workspace Explore the demo

Free shipping progress bar

Cart · all devices · launched Oct 23 by Camille · day 18

Conclusive. Proven gain

Showing progress toward free shipping: revenue per visitor +6.0% measured, +4.9% expected once corrected for the winner’s curse. The effect on margin isn’t established yet: +3.0% measured, range from -0.7% to +6.7%.

Report checks

Planned sample size: 104%
A/B split: 50.0 / 50.0
Sample ratio mismatch (SRM) test: p = 0.81
Saw both versions: 0.3%
Guardrails breached: none
Boundary crossed at checkpoint: day 18 · z = 3.1
Orders reconciled: 98.6%
Known costs (margin): 96.7%
Winner’s curse: corrected, +4.9%

Maison Lune, a fictional demo store · report computed by the engine

A result has to hold up in front of the client.

The most common problems when you test on behalf of a store.

“Where does this number come from?”A gain presented without a test plan or a range doesn’t survive the client’s first question.
A gain that never materializesAnnounced winners overestimate their effect on average, so the client never finds the gain in their revenue.
A test stopped when convenientStopping on the day the numbers look good casts doubt on every later result, even the good ones.
Reports rebuilt by handResults get copied into slide decks, with a chance of error at every copy.

You and your client work from the same method.

What your client sees in a report, and where they can check each step.

Checks next to the verdictSample reached, traffic split, boundary crossed, bias corrected: listed in every report.9 checks in the “Free shipping progress bar” report.Learn more A rule set before launchThe test plan, checkpoint days, and guardrails are locked at launch: nobody picks the moment to conclude.“Free shipping progress bar” test plan: checkpoints on days 6, 12, and 18.Learn more A cautious gainThe reported gain is corrected for the winner’s curse and given with its range, for both revenue and margin.+6.0% measured, +4.9% reported; gain range +€41,600 to +€209,800.Learn more Published methodsEvery calculation relies on a published method with its source, and the engine is checked against reference values.O’Brien-Fleming, mSPRT, Chi-square, Holm, and Bayesian shrinkage.Learn more A report your client can verify aloneOne read-only link per period: sealed plans and their fingerprints, verdicts, decisions, impact report, and decision quality, under your agency’s brand.Demo report, Q3 2026: 11 tests, 11 with a sealed plan.Learn more Decision quality, not just winnersThe share of tests that taught you something comes before the share of winners, with criteria counted test by test: sealed plan, traffic split, sample, decision, false positive, confirmation at rollout and, for subscription tests, whether the decision held over time.Demo, Q3 2026: taught something 55% (6 of 11), winners 27% (3 of 11).Learn more A demo to show before signingThe demo is open without an account: your client can read reports computed by the engine on a fictional store.Running tests, winners, guardrail stops, and rollouts.Learn more An audit backed by numbersIdeas from an audit get an expected effect and a duration based on the client’s traffic, before anyone makes a promise.Top of the demo’s list: “Free shipping from €50”, +2.5% expected.Learn more

Maison Lune, a fictional demo store

Your client reads the report as the engine writes it.

Conclusion, metrics, guardrails, amounts, and checks fit on one page. Here, the demo store’s “Free shipping progress bar” report.

Free shipping progress bar

Cart · all devices · launched Oct 23 by Camille · day 18

Conclusive.

Result. Showing progress toward free shipping: revenue per visitor +6.0% measured, +4.9% expected once corrected for the winner’s curse. The effect on margin isn’t established yet: +3.0% measured, range from -0.7% to +6.7%.

Why. More orders reach the threshold: customers add an item rather than pay for shipping.

Next. Expected revenue: +€125,700 per month after correcting for the winner’s curse. Then test a €50 threshold, measured on margin.

Metrics

Metric	Original	Variant	Difference	95% range
Revenue per visitorprimaryOriginal €24.58 · Variant €26.06	€24.58	€26.06	+6.0%	+2.2% to +10.0%
Margin per visitorafter product costs, discounts, and shippingOriginal €10.03 · Variant €10.33	€10.03	€10.33	+3.0%	-0.7% to +6.7%
Conversion rateOriginal 20.0% · Variant 20.8%	20.0%	20.8%	+3.9%	+0.7% to +7.2%
Average order valueOriginal €122.90 · Variant €125.40	€122.90	€125.40	+2.0%	0.0% to +4.1%
Orders with free shippingexplains the gap between revenue and marginOriginal 41.0% · Variant 57.0%	41.0%	57.0%	+16 points	Expected effect of the variant

Guardrails

Guardrail	Original	Variant	Difference	95% range
Return rateprovisional, 30-day value on Dec 9Original 1.6% · Variant 1.7%	1.6%	1.7%	+2.6%	Within limit, limit +25.0%
Mobile speed (LCP)Original 2.41 s · Variant 2.42 s	2.41 s	2.42 s	+0.4%	Within limit, limit +4.0%
JavaScript errorsshare of visitors with an error · limit +25%Original 0.12% · Variant 0.12%	0.12%	0.12%	+0.2%	Watch

Amounts

Expected revenue / month+€125,700+€41,600 to +€209,800

Expected margin / month+€27,700-€5,400 to +€60,900

Free shipping-€23,800per month, already deducted from margin

Checks

Planned sample size104%

A/B split50.0 / 50.0

Sample ratio mismatch (SRM) testp = 0.81

Saw both versions0.3%

Guardrails breachednone

Boundary crossed at checkpointday 18 · z = 3.1

Orders reconciled98.6%

Known costs (margin)96.7%

Winner’s cursecorrected, +4.9%

Maison Lune, a fictional demo store · report assembled by the engine

One report per period that your client verifies alone.

For a quarter or a month: each stopped test with its sealed plan and the fingerprint that proves it, its verdict and the decision taken, decision quality, and the impact report, corrected and checked against the holdout. Your client can recompute a plan’s fingerprint; the method is published.

Open the demo report How to verify it

Branding: Your agency’s name and logo at the top, set by the workspace owner (here, Studio Ondée, a fictional agency).
Link: Read-only, no account needed, nothing else from the workspace. It expires after 7, 30, 90, and 180 days, your choice, and can be revoked at any time; every creation and revocation is logged. A viewer can see links but not create them.
Tests: Demo, Q3 2026: 11 stopped tests, 11 sealed plans, 36% (4 of 11) with no caveats.
Printing: The report prints or saves as PDF from the browser, laid out for paper.

One workspace per client, one role per member.

Each store has its own workspace, created with its account. The store invites you from its Settings with the role it chooses, and can invite its own teams as viewers. Declare in your account Settings that you manage your clients’ sites, and “My clients” appears in your menu, with one row per workspace (running tests, latest verdict, gains and quality for the quarter). You switch between them from the menu, the current workspace stays marked, and access is checked on every request.

Owner: creates the workspace and has every permission in it.
Editor: prepares, launches, and decides tests.
Contributor: prepares tests but can’t launch, stop, or roll them out.
Viewer: views everything without changing anything.

For a new client, the store opens its account and invites you, or you create the workspace from yours. Each workspace has its own plan, or draws on the pool of tested visitors in your Agency plan, shared across your clients’ workspaces on a single invoice.

A typical engagement, from audit to rollout.

How an engagement with a client runs, and what the demo store shows at each step.

Audit
Size the audit’s ideas.
Each idea gets an expected effect and a duration based on the client’s traffic. Ideas you couldn’t measure are dropped before anyone promises them.
Demo storeTop of the list: “Free shipping from €50”, “Free returns noted in the cart”, and “Secure payment methods shown in the cart”.
Present
Present the test plan to the client.
Test length, target effect, and guardrails are known before launch, so the client knows when the answer will come.
Demo storeFree shipping progress bar: 18 days, target effect 5.5%.
Test
Let it run with guardrails in place.
The plan’s guardrails stop a test on their own if it harms something. The client doesn’t have to take your word for it.
Report back
Show the report as is.
A one-sentence conclusion, the range, the gain in euros for revenue and margin, and the checks alongside.
Demo store+€125,700 in expected revenue a month, margin per visitor +3.0%.
Roll out
Roll out in steps.
The winner goes live with the same guardrails and rolls back on its own if one of them degrades.

Frequently asked questions.

Can my client verify a result?

Every report lists its checks next to the verdict: sample reached, traffic split test, visitors who saw both versions, boundary crossed at a checkpoint, and winner’s curse correction. Each test plan is sealed at launch: your client can recompute its SHA‑256 fingerprint from the text shown in the report. The methods and their sources are published on this site.

How do I show Probant to a client before they sign?

The demo is open without an account: a client can read reports computed by the engine for Maison Lune, a fictional store, before installing anything.

How are workspaces organized?

Every account gets a workspace at sign-up and can create more from the app. A workspace has members, invited from its Settings, each with a role: owner, editor, contributor, and viewer. An agency that declares in its account Settings that it manages its clients’ sites sees all its workspaces under “My clients,” one row per workspace, and switches between them from the menu; access is checked on every request.

How do I send a report to a client?

In the Library, “Client report” builds the report for a quarter or a month: stopped tests, sealed plans, verdicts, decisions, decision quality, and the impact report, under your agency’s name and logo. A read-only link, with no account needed, expires on its own and can be revoked at any time; the report also prints to PDF.

Will the reported gain show up in my client’s revenue?

It’s an estimate, and the report presents it as one: corrected for the winner’s curse and given with its range, for both revenue and margin. The low end of the range is the cautious figure to present.

Other use cases

The same engine, for other site areas and other teams. The demo opens on the screen this page describes.