NOWMethodology in development

NOW

The Commerce Acceleration Index

How fast does this website want you to act?

NOW is being assembled in the open. It records what commerce pages actually said at a stated moment, with the evidence attached, so a claim about pressure can be checked rather than taken on trust.

Five dimensions are now measured across the whole panel in one cross-sectional pass: every retailer against the same snapshot, the same frozen ontology and the same frozen detector. It still does not issue a score. Nothing here has been through independent human review, and the bar a dimension must clear was written down before the review was sent.

Higher pressure is not a synonym for worse. A deadline can be real, a stock count can be accurate. NOW measures presence, form and placement. Whether a claim is truthful is a separate question this instrument does not answer.

The corpus, as counted

114
retailers in the universe
241/265
pages observed, open

The rest were hard-blocked, errored or 404. A block is a coverage gap, never an absence of pressure.

241
pages scanned

Across 96 retailers, every one against the same pinned snapshot.

474
signals recorded

Each carries its mechanism, a verbatim window and the hash of the page it came from.

Corpora dated 2026-09-06 · baseline run 2026-09-06T18-15-baseline-01 · dataset now-dataset-2
Ontology now-observation-v1 · detector now-v1-detector-1 · split now-v1-split-1

Detector observed — independent human review pending.


What is measured

Holdout figures below are author-adjudicated, blind to detector output and made against a frozen ontology. They are not independent and do not satisfy the acceptance criteria; they set expectations for what independent review might find.

DimensionPages with signalRetailersPrecisionRecallF1
Urgency

A claim that adds a reason to act sooner through TIME. The pressure is a deadline, a clock, or a bounded event.

20/24113100%83%91%
ScarcityExperimental — evidence base currently limited.

A claim that adds a reason to act sooner through SUPPLY: how much is left, or how fast it is going.

23/2411380%80%80%
Promotion

A commercial inducement on price or value. Presence of a discount, code, bundle or gift. Explicitly NOT a time claim.

169/24172100%96%98%
Shipping Pressure

Fulfilment used as an inducement: a spend threshold that unlocks shipping, a cut-off that protects a delivery date, or expedited delivery offered as a benefit. Ordinary logistics copy is excluded.

118/24157100%80%89%
Payment Softening

The price is made to feel smaller by changing how it is paid: split into instalments, deferred, or expressed as a small periodic figure.

16/24111100%92%96%

What is not measured

Transaction speed — not measured

The three-page model observes home, category and product. Checkout and cart are not observed at all, and the earlier cart pass reached 9 of 63 observable retailers against a 60% threshold predeclared before that run. There is no honest way to score it.

Decision friction — not measured

No observation method exists. Reserved, not measured.

Shown as unmeasured rather than as zero. A dimension nobody has looked at is not a dimension scoring nothing.


Why no composite is issued

No dimension has been through independent human review. The acceptance criteria were recorded before the 181-item blind package was sent, and until it comes back no dimension can be shown to clear them.

A dimension may enter a composite only if independent review supports precision of at least 90%, recall of at least 75%, an F1 of at least 80%, at least 10 independently judged positive examples, and no unresolved ontology defect that materially changes the construct. A composite needs 4 of 5 dimensions to clear all five.

Those thresholds were recorded before the review package was built, and the file is hashed. A bar chosen after seeing the score is not a bar.


Retailer records

Detailed records exist for ten retailers, chosen before any signal was read. The v1 pass covers all 96 scanned retailers; the rest are listed with their coverage.