Every revenue figure you have seen is a guess

Not a lie — a guess, made carefully, by people who mostly say so in their own marketing. A Shopify storefront does not publish its sales, so no tool can read them. What the category sells is a model, and the category puts that model at 70 to 85 percent accurate.

2 September 2026 · figures from the tools’ own published claims

What the tools say about themselves

This is not a claim we are making about them. It is the number in their own materials: sales estimates are typically 70–85% accurate, and PPSPY advertises 80.5%.

Sit with that for a second. The headline feature — the reason people install the thing — is described by its own vendor as wrong somewhere between one time in five and one time in three. And the error is not evenly spread: the signals these models rest on behave very differently for a two-month-old store than for an established one, which is precisely the case where a dropshipper most wants an answer.

How the estimate is actually made

From signals the storefront genuinely does publish:

• Best-seller ordering — the sequence the merchant themselves put products in.
• Publish dates — how long a product has existed.
• Review counts and their rate — reviews accumulate roughly with sales, at some unknown ratio that varies by niche, by app, and by how hard the merchant asks.
• Variant availability — what has sold out, and how quickly.
• Change between crawls — which requires a server that visited before.

A product gaining reviews quickly while staying in stock is inferred to be selling well. That is a sensible inference. It is also why two tools looking at the same store on the same day routinely disagree by a factor that would embarrass either of them if the two figures were ever shown side by side.

Why the number is displayed so confidently anyway

Because a range is hard to sell. “Between €8,000 and €47,000 a month, and we are not sure” is the honest rendering, and nobody would install it. So the interface prints a single figure in a large typeface, and the typeface does the persuading that the arithmetic cannot.

We looked at this from the inside: we unpacked forty of these extensions and read the code. The pattern was consistent — the confident number and the hedged disclaimer live in different parts of the product, and only one of them is on the screen you actually look at.

What you can measure instead

Quite a lot, and all of it counted:

• The full catalogue and its size.
• Every price, and which are discounted, and by how much.
• The merchant’s own best-seller ordering — their opinion, not a model of it.
• When each product was published, and the launch cadence over two years.
• The apps, theme and pixels in use.
• What changed since you last looked.

That set answers most of what people ask a revenue figure to answer — is this store serious, is this product new, are they discounting, what are they investing in — without a single estimate. And where a figure genuinely cannot be known, the honest move is an em-dash and a reason, not a number with the uncertainty removed.

When an estimate is still worth having

Sometimes it is. If you are screening two hundred stores to find ten worth a closer look, an 80%-accurate signal is a perfectly good filter — you are ranking, not deciding, and the errors mostly wash out at that scale.

The failure is using the same number to decide whether to commit money to one product. At n=1 an 80% accurate figure is just a figure that is probably wrong in an unknown direction, and the interface will not tell you which one you got.

Questions

See also: We took apart 40 Shopify spy extensions · Shopify competitor spy tools compared · MurmSpy · Writing