▤The Test Group experiment running Open the partner account
paid
Affiliate disclosure. The partner link in the masthead and in the band beside the copy on this page is a sponsored link to a partner operator, and this site may be paid if you open an account through it, at no extra cost to you. It carries rel="sponsored noopener" and opens in a new tab. A desk about how a screen is measured and tested should not leave its own funding unsaid: one link funds the site, no operator and no product is named, rated or recommended anywhere on it, and this site runs no analytics of its own on its readers.
The Test Group / About
How this desk was built, and the eight things it hands off

About this project

This desk is one of a series, each of which owns one mechanism in depth. This one owns the measurement loop. It was built from ten invented samples, it names no operator, vendor or product, and it hands eight adjacent subjects to the desks that already own them rather than repeating them badly.

Desk spec
desk
75 of the series
concept
45
design
54
samples
10 invented
the eventOne interaction, written down. A click carries a name, a time, an account, a session and what was on the screen, and it is kept whether or not the reader chose to be measured.
the funnelThe order the steps happen in. A landing visit becomes an account, an account becomes a deposit page, and 100,000 visits end in 2,074 first bets - a 2.1% path the whole loop is aimed at.
the armsTwo versions of one screen shown at the same time, and a rate for each. 4.30% against 5.16% is a 0.86-point lift, and the interval decides whether it is a result or a coincidence.
Direct answer

This desk explains how a gambling interface is measured and tested: the event, the funnel, the arms, the replay, the holdout, the metric and the guardrail. It was built from ten invented samples, it names no real operator, vendor or product, it publishes no ranking, and it hands eight adjacent subjects to the desks that own them.

How it was built

  1. One subject was chosen that no earlier desk in the series owns, after scanning 1,563 files for it.
  2. Ten samples were invented - the arms, the log, the funnel, the replay, the holdout, the metric, the sample size, the consent line, the limits and the money - so every figure could be derived rather than quoted.
  3. The arithmetic of each sample was written out on the page that uses it, so a reader can check it.
  4. Each page was given one direct answer, one worked example, one comparison table and one reader checklist.
  5. The build was verified externally against the live origin, page by page, before anything was published anywhere.

The eight subjects this desk hands off

what a screen says and permits - the result message, the speed control, the money control and the reported number belong to the interface desk; this one owns the loop that chooses and measures them.
the rights over stored data - the subject-access request, the consent record and the retention period are a different desk; this one treats them as the limit on what can be measured.
the operator's service layer - chat, e-mail and the authority of a support agent belong elsewhere; here a contact is only another event.
the industry's own reporting - turnover, gross win and the hold are the measurement layer above whole businesses, not the telemetry of one screen.
the device the product runs in - the app, the store gate and the permissions belong to the device desk; this one measures what happens after it loads.
the company behind the brand - the licence, the entity and the register are covered elsewhere and are not repeated.
the controls a reader sets - deposit limits, loss limits, session timers, reality checks and self-exclusion are a subject of their own, and they are this desk's answer to its own findings.
the advertisement - what a paid promotion may claim and who is paid to place it is another desk; nothing on this site is a promotion beyond the one disclosed link.
the samples, and what each one is for A the two arms control 4.30% against variant 5.16%, a 0.86-point lift B the event log 1,240 events in 14 minutes, 240 bytes each C the funnel 100,000 visits to 2,074 first bets, 2.1% end to end D the replay 10,000 a week at 3.4 MB, 2.1% watched E the holdout 5% of a million, a 3.6% realised effect F the metric 4.30% to 5.16%, against an 11.8% guardrail move G the sample 9,550 an arm required, 12,000 run H the consent line 68.4% accept at 5.10%, population rate 4.22% I the limits 11 of 24 shipped changes aimed outside the page J the money 860 extra deposits, 29,412.00 against 6,192.00 ten samples; every other figure on the desk is derived from them.

What the desk refuses to do

Three refusals are structural rather than stylistic, and each is asserted by the build's own verifier rather than left to good intentions.

The desk's three refusals, and how each is enforced
RefusalWhyHow it is enforced
no ranking of operatorsa ranking is a commission in another formno Product, Offer, Review or AggregateRating markup on any page
no named operator or vendorthe desk describes machinery, not a companyevery figure traces to one of the ten invented samples
no method of evading anythingthe subject invites it, and the desk refuses itno page explains how to defeat a check, a limit or a control
the disclosed linkone partner link funds the siterel="sponsored noopener" and a disclosure band on every page
the desk in numbers pages = 18 concept = 45 of the series design = 54 samples = 10 invented affiliate placements = one disclosed partner link, on every page markup the desk refuses = Product, Offer, Review, AggregateRating runtime JavaScript = none; JSON-LD is the only script element files scanned for novelty = 1,563, across cycles, scheduler and .agents

The two numbers that decide a ship, in the desk's own words

the primary metric
the one number a change must raise, here the first-deposit rate
the guardrail
the number it must not move beyond a threshold fixed in advance
the sample
how many visitors the answer needs before the result may be read
the holdout
the group kept on the old version so the series can be checked as a whole
This desk is educational and is aimed at adults. It describes how a measurement loop works in general; it does not tell a reader what a named operator does, it does not help anyone evade a control or deceive a bank or an operator, and it gives no advice about any product. Every figure on it is illustrative.
What this desk does not claim
  • It does not claim any named operator runs a holdout, a replay or a guardrail.
  • It does not claim any rate on any page is observed rather than invented.
  • It does not claim that a change is always harmful, only that its direction is the operator's.
  • It does not claim that a reader can see any of this from outside; the record is internal.
  • It does not offer a method of avoiding measurement, because none of its findings require one.

Read next

The numbersThe ten samples, figure by figureLegalThe risk warning and the disclosure in fullThe limitsWhy the desk is careful about what a gap meansQuestionsThe twelve questions readers ask most