Practical affiliate measurement · October 2026
Can you match an affiliate sale to its source? Build one evidence row first
Your website shows a click. The affiliate dashboard shows a sale. It is tempting to put them next to each other and call the campaign successful. Unless the reports share a usable identifier, you may only know that two things happened during the same period.
This worksheet helps you find the evidence you already have before paying for another tracking tool. It does not inspect your accounts, create referral links or generate a test sale. Use a private copy of the record and omit customer details.
Follow one journey through five records
Start with an existing campaign. Give the review a clear date range and use the same time zone across the reports. Keep owner tests, previews and genuine prospective-customer activity in separate groups. A test visit is useful for checking a route; it is not customer demand.
| Record | What to write down | What it establishes |
|---|---|---|
| Source visit | Channel, landing page, timestamp and permitted campaign label | A recorded visit to a page you can measure |
| Outbound referral | Approved destination and recorded outbound event | Someone followed the referral route, subject to the report's counting rules |
| Vendor event | Signup, trial or sale, with the event name and reporting time | Only the specific event the vendor reports |
| Approved commission | Eligible amount and approval status, if reported | The program has accepted a commission for payment under its rules |
| Paid commission | Payout record and receipt status | Payment progress; distinguish a sent transfer from money received |
An empty field stays unknown. It should not silently become zero, and a page visit should not fill the sale field. Your affiliate network's accepted sales and commission records remain essential even when a tracker reports more events.
The identifier decides how far you can go
Look for the program's documented Sub-ID, tracking reference or another permitted identifier. Do not invent a parameter or append one to an issued affiliate URL simply because another vendor accepts it.
If both the click report and sale report preserve a usable identifier, you can investigate the match. Also check the time window, duplicate events and the program's attribution rules. A shared identifier supports a stronger connection than a similar timestamp, but still does not override a rejected or refunded commission.
If you only have channel-level totals, mark the relationship as aggregate. You can compare patterns cautiously; you cannot assign a particular sale to a particular post from that alone.
Google's campaign-URL documentation explains how UTM labels describe traffic arriving at a measured destination. Such labels on your own acquisition link do not automatically become purchase evidence inside another company's checkout. Keep incoming traffic labels separate from the affiliate program's referral mechanism. [Source: Google Analytics URL builders.]
A worked example with a deliberately missing answer
Suppose an educational post generates eight recorded landing-page visits. Your site logs three outbound events. The vendor reports four clicks and one trial during the same week, but no sale identifier or source breakdown.
The truthful report is: eight recorded visits, three site-recorded outbound events, four vendor-counted clicks and one reported trial. The different click totals need investigation. The trial's source remains unconfirmed. There is no basis for claiming a paid conversion or calculating a sale conversion rate.
Different counting windows, repeat-visit rules, consent settings and previews are possible explanations to inspect. None is a diagnosis until you have evidence. Keep the raw totals rather than forcing them to agree.
Check the program's credit rule before interpreting a match
The same referral journey can receive different treatment under different programs. Fillout's current official program documentation describes first-link credit and an account association that cannot later be changed. Tapfiliate's own affiliate program describes a last-cookie rule. Those are program-specific examples, not settings to apply to all offers. [O08, O13]
Record the rule beside the campaign. A later link interaction does not automatically replace an earlier affiliate association, and your own analytics cannot override a vendor's eligibility decision. If a report only says “conversion,” keep its definition next to the count. Do not change the label to “approved commission” without the relevant approval record.
For money, record the commission base as well as the percentage. Payhip's partner terms share a percentage of Payhip's fees, not the seller's full product revenue. Multiplying a commission percentage by the wrong base produces a convincing but incorrect earnings figure. This worksheet deliberately does not forecast commissions. [O11]
When a paid tracker may become useful
ClickMagick's official product page describes server-to-server postback integrations for affiliate offers. That can be relevant when your actual network supplies a supported event and the identifier survives the journey. It does not give you ownership of the merchant's checkout or guarantee access to unshared transactions. [Source: ClickMagick product documentation.]
Before evaluating paid tracking, write the decision you expect it to improve. “Find which placement produces accepted sales” is a useful question if accepted-sale data is available. “Find every missing purchase even though the vendor shares none” is not a realistic test.
You may be ready to evaluate when the data path exists and reconciling multiple campaigns has become recurring work. Keep existing reports when they already answer the question or when the missing dependency is vendor data. There is no fixed number of clicks that makes another subscription worthwhile.
Copy this review record
Use one row per event or reconciliation question. Record the source system, date range, time zone, event definition, identifier availability, observed count or amount, test/customer classification and unresolved issue. Keep commission amounts and visitor counts in separate columns.
For a first review, copy or print this blank record. Enter the reporting window and time zone above it. An empty value means you have not established it; use zero only when the relevant report actually shows zero for that window.
| Record | Observed value and unit | Source report and window | Match or limitation |
|---|---|---|---|
| Landing-page visits | ______ visits | ______ | ______ |
| Site-recorded outbound events | ______ events | ______ | ______ |
| Vendor-counted clicks | ______ clicks | ______ | ______ |
| Vendor signups or trials | ______ named events | ______ | ______ |
| Eligible paid sales | ______ sales | ______ | ______ |
| Approved commission | ______ currency/amount | ______ | ______ |
| Received payout | ______ currency/amount | ______ | ______ |
Keep a separate line for controlled tests and note which rows may include them. Do not subtract a guessed test count from a vendor total. Mark the uncertainty until the report supports an exclusion.
The count-comparison tool below can help with one narrow check: whether two counts differ after you have confirmed that they describe the same event, window and counting basis. It does not match people, diagnose missing sales or connect to an account. Leave a confirmation unchecked when you do not know. A precise difference between incomparable reports would be misleading.
Use counts only. Keep money and currency in the separate commission rows. For example, three site-recorded outbound events and four vendor clicks should remain incomparable until you have established that both reports count the same activity in the same way. Even a valid difference of one is a question to investigate, not proof of a lost customer.
Give every proposed fix an expected result. For example: “The approved source label appears in the network's report.” A later check can then pass, fail or remain untested. A collection of screenshots without a defined event is harder to interpret.
Choose the next step your evidence supports
If you cannot name the missing record yet, use Demandward's Tracking Gap Assessment. If a compatible sale feed exists and you are comparing software, read ClickMagick for affiliate marketers for plan and buyer-fit checks.
Affiliate disclosure: Demandward has a ClickMagick affiliate relationship and may earn a commission from an eligible purchase through the disclosed referral link on its buyer guide. This worksheet contains no direct vendor referral link. No software purchase is needed to use the method.
Research limits
Official program sources refreshed October 21, 2026, U.S. Central. The Google campaign-label reference was carried forward from the October 20 source review; the ClickMagick capability reference was refreshed October 21. The worked example is fictional arithmetic, not Demandward traffic. This worksheet was not connected to a merchant, network or analytics account. It cannot validate a commission, reconstruct unreported sales or establish that a third-party integration works in your account.
Additional sources: O08, Fillout/Zite affiliate program; O13, Tapfiliate's own affiliate program; O11, Payhip partner terms. Their rules illustrate why an evidence record needs the actual program name. They are not endorsements or permission to use a new referral link.
Compare two counts without inventing a match
Counts only. Do not enter money, customer information, account numbers or identifiers. These fields do not connect to any report.
A blank count means unknown. No difference has been calculated.
This tool processes values in this page. Its code uses no network calls or storage. Site-wide hosting and analytics behavior require a separate release check.
Official reference links
Documentation checked October 21, 2026, except the carried-forward Google reference as stated in the manuscript. No product integration or customer outcome is claimed.