Methodology v1 · August 2026
The Product Readiness Score
A product is ready when everything that sells it can read it: Google, Meta, and the AI agents doing a growing share of product discovery. The readiness score puts one number on that, per feed, so an agency and a retailer can talk about "the feed is at 7.8" and mean the same thing. This page is the whole methodology. We publish it because a standard you cannot inspect is not a standard.
How the number is computed
Every audited item lands in one of three states. Ready items pass every check. Improvable items carry warnings: things that rarely block serving but cost visibility. Blocked items carry at least one error of the kind destinations reject outright.
The score is the share of the catalogue that is ready, with improvable items counted at 70 per cent, scaled to 0 to 10 and shown to one decimal:
score = (ready + 0.7 × improvable) ÷ total items × 10
The 0.7 mirrors how channels treat warnings: they rarely block an item, but they rarely leave its reach untouched either. The score is measured on the delivered catalogue, after your rules run, so fixes visibly move it. Your source feed is scored with the same formula beside it, which is how you see what the layer is earning you.
The seven dimensions of readiness
The checks behind the score group into seven questions, and we are explicit about which the score measures today.
-
Correct
In the scorePrices carry a currency and are not zero. Sale prices make sense against regular prices, with valid effective dates. Availability uses wording channels accept. Links and images resolve over HTTPS.
-
Complete
In the scoreThe fields that decide visibility are present: identifiers (GTIN, MPN, brand), titles, descriptions, images, prices, availability. The field inventory also tells an attribute your platform never sends from one it sends blank, which are different problems with different fixes.
-
Compliant
In the scoreProducts meet destination requirements before the destination has to complain: valid Google product categories, title and description lengths inside channel limits, sane variant group structure.
-
Understandable
In the scoreMachines can parse what the product is. Titles that truncate mid-attribute, shouting case, and missing identifiers all make a product harder for search engines and shopping agents to match and rank.
-
Current
In the scoreThe delivered feed reflects the source on schedule. Run recency and refresh cadence are visible per feed; a stale feed is a readiness failure however clean its data.
-
Governed
In the scoreNot a data check but a property of the layer: every change is previewed before it applies, every rule is versioned and reversible, and the live feed only changes when a new version passes every check.
-
Performant
Not yet scoredWhether improvements produce measurable commercial results. Not yet part of the score, and we will not pretend otherwise: it joins the methodology when outcome attribution ships, and it will be marked verified or estimated honestly.
Verdicts beat predictions
A readiness check is a prediction about what a destination will do. Where we can, we replace prediction with the destination's own answer: connect Merchant Center and SKULayer reads Google's actual verdicts, item by item, and matches them to your products beside the audit findings. When a fix ships, you see whether the issue actually cleared, not whether we think it should have.
Versioning
This is version 1, published August 2026. The formula, the check inventory and the dimension mapping only change with a version number and a dated note on this page, because a score that moves silently is a score nobody can put in a client report.
Find out what your feed scores
Connect the feed you already have and the full audit, score included, is free for your first 1,000 products. Or look at a worked example first.