Shopify sales tracker: how the estimates are made, and what you can check

No Shopify store publishes its revenue, and Shopify does not publish it for them. Every Shopify sales tracker is therefore an inference, and almost all of them infer the same way: ask the store's own cart how many units of a variant it will accept, come back later, and read the difference. You can run that probe yourself in one request, which is the fastest way to understand what the estimates are worth. ShopLeads is our product and it sells store and contact data, so we say so; everything below was run against live storefronts on November 9, 2026.

The pages ranking for this query all describe what a tracker shows you. None of them describes how the number is produced, which is the only thing that tells you when to trust it.

What a Shopify store actually publishes about its sales

Nothing, directly. The public JSON a storefront serves tells you what is for sale and whether each variant can be bought, and that is the end of it. A product in /products.json carries published_at, created_at, updated_at, vendor, type and tags, and each variant carries a price, a SKU and an available boolean. There is no quantity field and no order count anywhere in it.

What is published is the platform total. Shopify reported $119 billion in GMV for its third quarter on November 3, 2026, alongside revenue growth of 32% and a free cash flow margin of 22%, and it defines GMV as the total dollar value of orders facilitated through the platform (Shopify). That is the aggregate of every merchant, and it is the only sales figure Shopify publishes: nothing in its investor reporting breaks GMV down to a single store. That gap is the entire market for sales trackers.

How a sales tracker produces a number

The mechanism is not secret; one vendor states it outright. Procware describes its estimator as analyzing daily order volume, best-selling SKUs and a projected annual run-rate using what it calls mathematical inventory decrement snapshots. That is the method, and here is what it looks like from the outside.

1. Read the catalog from /products.json, which gives every variant, its price and whether it is buyable.
2. For each variant of interest, ask the store's cart to add an absurd quantity. When the store tracks inventory and refuses to oversell, the error message names the number it would actually accept.
3. Repeat on a schedule. The drop in that number between two reads is units sold, minus any restock.
4. Multiply units by price, sum across variants, and extrapolate to the variants and days you did not sample.

Every weakness in a tracker lives in step four. Steps one to three produce a real observation; step four produces a revenue figure, and the distance between them is where the error is.

Run the inventory probe yourself, on one variant

curl -s -X POST https://www.beardbrand.com/cart/add.js -H 'Content-Type: application/json' -d '{"items":[{"id":VARIANT_ID,"quantity":99999}]}'

Take the variant id from /products.json. On Beardbrand on November 9, 2026 we got two kinds of answer. For a variant marked unavailable: {"status":422,"message":"The product '…' is already sold out."}. For a buyable variant: {"status":422,"message":"Only 60 items were added to your cart due to availability."}. That second message is the exact on-hand quantity for that variant, handed over by the store itself. Nothing was ordered and nothing was charged; the request creates an anonymous cart and is rejected.

Run this a handful of times to understand the method, not thousands of times to build a dataset. It is one request per variant per sample, and a store with hundreds of variants sampled daily is a lot of traffic aimed at somebody else's server.

Five things that break the estimate

• **Untracked inventory returns no number.** A variant set to continue selling when out of stock, or with inventory tracking off, accepts any quantity. Print-on-demand, made-to-order, digital products and services all behave this way, and a tracker has nothing to measure on them.
• **Restocks look like negative sales.** Inventory going up between two samples means a delivery arrived, and the tracker cannot see the difference between a restock and a cancellation. Sampling less often makes this worse, not better.
• **Sales outside the storefront are invisible.** Wholesale, retail doors, Amazon, TikTok Shop and physical stores never touch the variant inventory the probe reads, or touch it in ways that look like sales nobody paid for. Any brand that also sells through retailers will be understated or distorted.
• **Discounts and bundles distort the multiplication.** The price in the catalog is list price. Subscription discounts, bundle pricing, promo codes and shipping all move the real figure, and none of them is in the public JSON.
• **Rate limits truncate the sample.** Allbirds answered our plain /products.json request with a rate-limit message rather than data on the morning we tested. A tracker hitting the same wall samples fewer variants and extrapolates harder.

What you can measure exactly, for free

The inferred revenue figure is the weakest thing you can take from a storefront. Several other readings are exact, reproducible, and often more useful for deciding whether a store is worth your time. All three rows below came from /products.json and the product sitemap on November 9, 2026.

The ban.do row is the lesson in that table. Every one of its first 250 products reported a published_at inside the last 90 days, which cannot be true of a catalog that size. published_at is the date the product was last made visible, so it resets when a product is unpublished and republished, which this store clearly does in bulk. Treat it as a visibility date, never as a launch date, and check the distribution before you draw a cadence from it.

• **Catalog size, from the sitemap.** Three requests gave a product count for a store where the JSON API would have taken nine. The sitemap is also authoritative about what is indexable.
• **Stockout rate.** A third of Olipop's variants were unbuyable and a thirty-fifth of ban.do's were. That is a real operational difference, and it is one request to read.
• **Price distribution.** Every variant price is in the JSON, so the average order value a store can plausibly have is bounded by arithmetic rather than by a model.
• **Product mix.** Vendor, product type and tags are all there, which tells you whether a store sells its own brand or resells, and how it segments its own catalog.

What the trackers publish about their own accuracy

An accuracy claim with no methodology next to it cannot be checked, because checking it would require the one thing nobody outside the store has: the store's real numbers. Read these figures as positioning. The vendor worth more of your trust is the one that tells you how the number was made and where it breaks.

If the store is yours, none of this applies. Shopify's own reporting gives you real numbers in the admin, and its sales tracking roundup lists the tools that connect to them. Estimation is only ever for stores you do not own.

When a sales band beats a revenue estimate

For product research, a point estimate is the whole game: you want to know whether a specific SKU is selling. For selling to stores, a point estimate is a false precision you will then have to defend in a meeting. What you actually need from store size is a yes or no to three questions.

1. Can this store afford what I sell, at the price I sell it for?
2. Is it big enough to have the role I need to talk to, or is the founder still doing everything?
3. Is it small enough that my product is still a decision rather than a procurement process?

A band answers all three and survives being wrong by a factor of two. A single number answers none of them better and breaks the moment anyone checks it. ShopLeads carries a monthly sales figure and a sales band on each store, and the band is what you filter on, from 10k-100k a month up to 10M+, alongside the platform, the installed apps and technologies by exact name, the tech gaps, the country and the category, and it attaches the decision-makers to each store in the same search. Searching and counting are free and never reveal a contact. Browse the Shopify store directory, then start the free trial or read the pricing page.

For the sizing arithmetic behind those bands, how many Shopify stores there are and why the sources disagree covers why no two indexes agree, how to build a Shopify store list covers assembling and cleaning one, and what a technology checker can read off a site covers confirming a store is on the platform before you measure anything.

Where to go next

• How many Shopify stores there are explains why no two indexes agree on a number.
• How to build a Shopify store list covers assembling and cleaning one.
• The indexes that publish sales estimates, compared: ShopLeads vs StoreLeads, Store Leads alternatives.
• Stores grouped by platform, with the roles on file: the Shopify store directory.
• Watch the launch film, or read the pricing page.