An external estimate can help you choose a product to investigate. It cannot establish a competitor's private sales, profit or monthly income. Your own sales records remain the reference for your business.
This article explains the historical examples published on InsightRaider, with a calculation you can reproduce and the information the archive does not contain. It is not a validation study of the paid product. The public methodology and source archive provide the underlying evidence.
A calculation you can reproduce
The archived entry for contains:
| Field | Archived value | How to read it |
|---|---|---|
| Listed price | $129 USD | Stored price, not a verified historical average transaction value |
| Sales proxy | 15,093 | Preserved model input, not an audited order count |
| Modeled revenue | $1,946,997 | Listed price multiplied by the sales proxy |
| Archive date | March 17, 2026 | Date the example was preserved, not the original collection date |
| Revenue period | Not recorded | The amount cannot be labeled monthly revenue |
The calculation is $129 脳 15,093 = $1,946,997. Download the source archive as JSON to inspect the product URLs, stored inputs and source revision.
Reproducing this multiplication establishes how the displayed amount was calculated. It does not validate either input against the seller's accounts. Historical discounts, price changes, refunds, bundles and payment terms can make actual receipts different.
Where the evidence stops
The archive preserves the sales proxy but does not record its original derivation. We cannot reconstruct that upstream inference from this file, and we do not relabel it as verified sales.
The original collection date and revenue period are also missing. Updating this article does not refresh those inputs. A six-month-old archive date does not tell us how many months of activity a stored value represents.
Public listings can help verify the offer currently being sold: buyer, deliverables, listed price, license and support. They do not reveal the complete transaction history behind an estimate. A present-day listing cannot retroactively verify all past prices or purchases.
For the same reason, adding several modeled product amounts does not establish the total addressable market. Products may have different observation periods, the sample may omit competitors, and some offers may overlap.
What would establish accuracy?
An accuracy claim needs a published comparison with appropriate reference data. At minimum, that means:
- Matching products and dates: each estimate and seller record refer to the same offer and period.
- A defined amount: gross sales, receipts after refunds and net profit are different measures.
- Documented adjustments: discounts, currencies, bundles, taxes and refunds are handled consistently.
- A disclosed sample: readers can see which products were included and where coverage is weak.
- An error calculation: the method and distribution of errors can be inspected, including poor predictions.
The public archive does not supply such a study. We therefore publish no validated accuracy percentage for these examples. Earlier versions of this article described a four-model system and a numerical accuracy range without the evidence needed to substantiate them; those claims have been removed.
A model can produce a precise-looking dollar amount without having precise knowledge of the seller's business. Use the amount as a research lead, alongside the limits above.
Compare offers before comparing their estimates
For an existing seller, a useful comparison begins with the buyer's problem. Three products in the same category may solve three different problems.
Our software research walkthrough illustrates this distinction. A battery utility, a display-brightness tool and an editor theme can all appear under software, but they are not interchangeable price benchmarks.
For each candidate, record:
- The customer and problem it serves.
- Deliverables, compatibility, license and support.
- Listed price, currency, billing period, source URL and observation date.
- Which values are directly observed, which are estimated and which are missing.
Keep direct competitors for a price comparison. Keep adjacent offers only for a stated reason, such as investigating license design. Exclude offers that do not help answer your decision.
From one snapshot to a weekly review
The historical public archive has no dated observation series, so it cannot demonstrate weekly growth. To evaluate a change, compare matching fields from observations taken on different dates.
A higher revenue estimate could reflect a new listed price, a changed input or an updated model. It is not, by itself, proof of additional sales. Similarly, an unchanged estimate does not prove no sales occurred.
In a current report, inspect observation dates and coverage before drawing a trend. If you start tracking a product today, treat today as the baseline; do not imply that tracking created a missing earlier history.
End the review with one decision: keep the offer, clarify a verified difference, or test a price or tier. Evaluate the outcome with your own visits, orders, net proceeds, refunds and customer feedback. A competitor's estimate cannot establish that your change worked.
Choose the right evidence for the job
| Your question | Start with |
|---|---|
| How much did I sell? | Your platform's sales and transaction records |
| Why did my own conversion change? | Your measured journey, traffic mix and buyer feedback |
| Which competing offers should I inspect? | Comparable product pages and public market signals |
| Did the market around my offer change? | Dated observations with consistent fields and coverage |
The Gumroad analytics and market research tools guide separates these jobs and includes a public walkthrough. Public examples are free to read. InsightRaider dashboard access is $49/month or $490/year, with a 30-day money-back guarantee on the first payment; review the offer and access conditions before subscribing.
Editorial correction: September 23, 2026. Product example archive: March 17, 2026. The editorial date does not change the age or completeness of the underlying data.
