THE PUBLISHED RUBRIC
The standard, before you pay for it.
The first deliverable of a GTM Reset sprint is scored against the eight criteria below. Five of them are blocking: if any one of those five fails, there is no invoice and you keep the work. This page is the whole standard, published so you can judge it before buying rather than after.
SCORED AT HANDOFF · BEFORE ANY INVOICE · EVIDENCE WRITTEN INTO THE SHIP LOGFIVE BLOCKING CRITERIA
Fail any one of these and there is no invoice.
Each one is written to be checked rather than argued: a named competitor set, a dated baseline, a falsifier present in the document at handoff.
- 01BLOCKING
The argument is decided, not described
The deliverable states one claim and names what it rejects.
- How it is checked
- A reader can say, without prompting, what the company is no longer claiming. If the deliverable only adds a claim, it fails.
- Why this is the bar
- Production is cheap now, so adding a claim costs nothing and changes nothing. A position is what you refuse to say.
- 02BLOCKING
No borrowed language
No sentence in the rebuilt asset appears, in substance, on a named competitor's equivalent page.
- How it is checked
- Three competitors are named at kickoff and their equivalent pages are read. Shared phrases are listed in the deliverable and removed from it.
- Why this is the bar
- Interchangeable language is why buyers price-shop. If the copy would work on a competitor's site unchanged, it is not positioning.
- 03BLOCKING
The problem comes before the product
The asset names the buyer's problem before it names the product or the company.
- How it is checked
- Read the asset top to bottom. The problem appears above the first mention of the product. Not in the same sentence, before it.
- Why this is the bar
- Buyers do not hire a product, they hire a fix for something making their work worse. Naming the product first asks them to translate.
- 04BLOCKING
One metric, with a baseline that exists today
The measurement plan names qualified pipeline created, or the explicit proxy standing in for it, plus the current number and the expected direction.
- How it is checked
- The baseline is a figure that can be pulled today, from a named source, with the date it was pulled. A target with no baseline fails.
- Why this is the bar
- Without a number that already exists, nobody can tell afterwards whether the work did anything, which makes the next decision guesswork again.
- 05BLOCKING
A falsifier is written down before launch
The deliverable states what result would prove the approach wrong.
- How it is checked
- The falsifier is in the document at handoff, dated. A reason invented after the numbers arrive does not count.
- Why this is the bar
- An approach that cannot be wrong cannot be learned from, and a plan with no falsifier quietly becomes unfalsifiable spend.
THREE QUALITY CRITERIA
Scored and reported. Not tied to the guarantee.
These change how well the work performs, but they involve judgement. Tying a refund to a judgement call would turn the guarantee into an argument.
- 06SCORED
The handoff is specified
The deliverable names who does what next, by when, for the asset to work in the real world.
- How it is checked
- Every dependency outside our control is listed with an owner and a date. Routing, follow-up timing, and approvals count as dependencies.
- Why this is the bar
- The most expensive failure is a good asset with a broken handoff: the work is fine, the response time around it is not, and the result reads as a content problem.
- 07SCORED
Every claim traces to a source
Each factual or comparative claim carries a source, a figure, or gets cut.
- How it is checked
- Superlatives and rankings are verified live at handoff, not from memory, and the check is dated in the deliverable.
- Why this is the bar
- Unsourced claims are the ones a prospect disproves in one search, and ranking claims decay faster than the page does.
- 08SCORED
It is extractable by machines
The asset answers its question in the first two sentences of the relevant section, in a passage that stands alone.
- How it is checked
- Each key passage is read in isolation. If it opens with an undefined 'this' or 'these', or buries the answer under setup, it is rewritten.
- Why this is the bar
- Buyers increasingly arrive through an answer engine that quotes a paragraph, not a page. A passage that needs its neighbours cannot be quoted.
HOW THE SCORING WORKS
The mechanics.
01When is the rubric scored?+
At handoff of the first deliverable, before any invoice is issued. The score is written into the ship log with the evidence for each line, so the result is auditable rather than asserted.
02What happens if a blocking criterion fails?+
There is no invoice for the sprint. You keep the deliverable and the ship log either way. This applies to the first deliverable of the GTM Reset sprint.
03Who decides whether a criterion passed?+
Each blocking criterion is written to be checkable rather than judged: a named competitor set, a dated baseline figure, a falsifier present in the document. If we disagree on a line, the failing reading stands and there is no invoice.
04Do the quality criteria affect the guarantee?+
No. Criteria 06 to 08 are scored and reported because they change how well the work performs, but they involve judgement, so tying a refund to them would make the guarantee an argument instead of a standard.
JUDGE THE STANDARD FIRST