teas.co.uk
The online British tea shop, with coffee too
Room 02 of 07 · the claim, stated narrowly
What we can prove.
A server log can establish that a named machine asked, that content was sent, and that something later came back carrying the trace. It cannot see inside an answer engine. This room says exactly where the evidence stops.
Every figure below carries its source, window and rule; the rungs count different events and are never added together.
7 SERVED READS IN 1 DAYS0 AI-REFERRED SESSIONS1 PAID PLACEMENT, 0 CREDITED EFFECT
Section 01
From a signpost to a token.
Each step down is rarer and harder to earn than the one above it. The counts shrink because the rules tighten — not because anything was lost.
| Step | Count | What it is | Why it is never added to the row above |
|---|---|---|---|
| Signposts published | 696 | Citation anchors offered across 158 pages | Counts the offer, not the uptake |
| Machines asked | 16 | Logged attempts, written down as they happened | Includes failures, probes and empty answers |
| The server delivered | 7 | Served reads - status 200 or 206, bytes above zero, credit rules applied | Delivery, never proof of consumption |
| Something came back | 0 · 0 | Traced requests at strong match - AI-referred sessions | Different source tables with no shared key |
| Thanks recorded | 0 | GBP 0 tokens across all four stages | Records its evidence class, never upgrades it |
Each row keeps its own source, window and rule. The rows count different events over different windows and share no key; no cross-table total exists.
Sources: anchor index, wp_crackers_citations (26 Sep 2026 to 27 Sep 2026), followback table (26 Sep 2026 to 27 Sep 2026), wp_crackers_sessions (26 Sep 2026 to 27 Sep 2026), the mint ledger - live.
Section 02
Where the evidence stops.
Not verified organisations.
Every agent name is a user-agent claim. We record what was presented, not who presented it, and withdraw credit on behaviour.
Not people.
A session is a browser visit. It is not a unique human, and it is not a customer.
Not proof of consumption.
Positive bytes mean we sent the content. They do not prove an answer engine read or used it.
Not a citation.
Whether our name appeared inside an answer happens beyond our server, so no figure is published for it. That is Stage 5, and it shows "no figure" until an import can evidence it.
The gap is published rather than filled. Our logs end at the door: we can prove delivery and we can prove returns, and we say plainly that the step in between — our name inside an answer — is visible only to the answer engine.
Section 03
The one paid placement.
One sponsorship exists — URSY. Its trail is tracked with the same instruments as everything else, published in full, and walled off from every figure in the study. Sponsorship evidence is graded on its own page and never mixed into a study total.
No count on this instrument proves that URSY sponsorship caused traffic, a citation, or a sale. Legacy origin imports and later same-family reads are held in the trail's own tables, graded separately, and never enter the study's figures.
Keep reading
Where this leads next.
One study, one style, seven rooms. The legacy room addresses now land on the room that absorbed their content, and every machine mirror keeps its original path.