Sold So Many
Install on Shopify

The network, live.

Network consolefeed nominalpoll—reads0age—readingjust nowsource/api/network-pulse

Proof impressions

10,905,471

cumulative · never decrements

Network status

stores running
44+0
storefronts emitting
4
shopping right now
116+0
orders, last hour
31+0
orders today
508

Today

impressions today
14,360+0
shoppers reached
15,278+0
storefronts here
4
shoppers / store
347

Engine output

4 emitting
  • velocity · SalesSelling 2 per daybeautifulqueenhair · 1 shopping this store
  • sales30d · Sales8 sold over two daysRachel Riley UK · 12 shopping this store
  • cartAddsInWindow · Interest3 added this to a cart this weekRachel Riley US · 13 shopping this store
  • sales30d · Sales6 sold over three daysSmåfolk.dk · 3 shopping this store

Impressions / second

now—/speak—/s

measuring — first readings land within a few seconds

Throughput

measuring — first readings land within a few seconds

window—
——

measuring — this window starts when you arrived

Signal families

  • Sales3
  • Interest1
  • Live now—
  • Returning—
  • Reviews—
  • Stock—
  • Patterns—

Serving path

  1. 01readredis-score-cache
  2. 02pickproof-serving
  3. 03fallbackproof-serving
  4. 04writecopy-templates
  5. 05serverecord-signal-exposure

Engine config

@56f22de
score cache
25 h
cutoff
20
truth floor
5
pdp max
3
list cache
5 min
viewers
2 s
anonymous network-wide totals · no per-store figure · no customerInstall on Shopify

Orders, last hour

placed on stores running the app

orders placed

31

countries

5

orders today

508

  • 5 min agoUnited Kingdom
  • 6 min agoHong Kong
  • 7 min agoUnited States
  • 16 min agoPoland
  • 16 min agoMalaysia

counted network-wide · never per store · never with a customer attached · the ages tick in your browser

From a product page to a sentence

per product, per visitor

  1. 01 · read · redis-score-cache

    Dimension scores, per product

    The product’s dimension scores are read from Redis, keyed per shop and product. A cold cache falls back to the scores table rather than to nothing.

  2. 02 · pick · proof-serving

    Best signal above the cutoff

    Signals scoring 20 or better are sorted, de-duplicated by dimension, and capped at three per product page, so a page cannot stack three flavours of the same evidence.

  3. 03 · fallback · proof-serving

    Else one claim at the truth floor

    If nothing clears the cutoff the engine drops to a single softer claim, still at 5 or better. Below that it prints nothing at all.

  4. 04 · write · copy-templates

    Rendered in the shop’s voice

    The chosen signal is rendered through the app’s templates in the shop’s voice and language. A template missing a value is skipped rather than filled with a blank.

  5. 05 · serve · record-signal-exposure

    Served, then measured

    The sentence is served with the storefront and the exposure is logged, so the learning loop can weigh it later. A visitor sees each signal once a day.

configuration is a snapshot, not a reading — build-time constants inside the app, read from sold-so-many@56f22de on 2026-10-16. The test suite re-reads them from the app’s source when that checkout sits beside this one.

Storefronts in this readout

4 sampled

photographs, sentences and viewer counts come from the app · a store with no claim to make is not in this list

Limits of this console — what it does not show, and why

No per-store figure.

Every network number is a total across every store running the app. The only per-store facts here are the sentence the engine is already serving in public on that store’s own product page, and that storefront’s own live viewer count.

No customer.

Orders are counted, never listed with a buyer: a country, an age, and nothing that belongs to a person.

No history.

The app publishes a running total, not a series. Every trend on this page is drawn from the polls this browser has taken since you arrived — which is the only kind of history that would be true.