Track search visibility, useful link clicks, approved commissions and cash received separately. Then choose one improvement the evidence supports. More traffic is an observation; it is not proof that your affiliate business earned more.
This guide is for publishers promoting other businesses. Start with one reader decision, one page or program, and a comparison period. You do not need a paid dashboard to make a useful review.
Match the question to the evidence
| Question | Evidence to inspect | What it does not establish |
|---|---|---|
| Are readers finding this page in search? | Search Console impressions, clicks and CTR, with the selected page, dates and filters. CTR is clicks divided by impressions.1 | Affiliate sales, unique customers or cash received |
| Are readers following the intended link? | In a configured GA4 implementation, outbound click events and their destination. Enhanced measurement fires for links leaving the domain; configured cross-domain destinations are excluded.2,3 | A merchant accepting or crediting the referral; event count is not a buyer count |
| Has the referral earned commission? | The provider's report, accepted terms and status: pending, approved or reversed. For example, Gardener’s Supply's public template separates action locking from payment scheduling.4 | Receipt of that money or page-level attribution where the report does not provide it |
| Has the money arrived? | Reconcile the provider's payment statement with the actual receipt and currency. | Profit, which also needs costs; a site's share of payment when attribution is Unknown |
Keep direct affiliate, second-tier and advertising results on separate lines. A direct customer's purchase does not prove a downstream affiliate relationship, and a search click does not prove either reward.
Make the comparison usable
- Fix the scope
Page or program, reader task, report, complete date range, filters, currency and relevant attribution or payment clock.
- Check the evidence
Which values are observed, unavailable, preliminary or from a different scope? Preserve Unknown instead of forcing a total.
- Pick one next action
Write the signal, your hypothesis, another possible explanation and the evidence you will check next.
Compare similar periods and inspect affected pages, queries, devices and countries. Google lists seasonality, technical issues, search changes and reporting problems among possible causes of traffic drops; a decline alone does not identify its cause.5
An exported zero can mean unavailable. Search Console says values displayed as ~ or – become zeros in downloaded data. Check the report before treating every exported zero as an observed result.1 If impressions are unavailable or zero, leave your own CTR calculation undefined rather than claiming a measured 0% rate.
Search Console and Analytics are different observations. Google's troubleshooting guidance lists privacy omissions, processing, reporting lag, time zones and JavaScript measurement among reasons their data can differ. Query-table totals can also omit anonymized queries; do not invent the missing terms or force totals to agree.6
Read outbound clicks carefully
If you already use GA4, Google's documented outbound exploration uses Link domain, Link URL, Outbound and Event count. It counts an event each time the relevant link is clicked, so repeated clicks are not unique customers.3,2
Group the routes you intended to measure. A source citation, application link and compensated product link answer different questions. A network-domain destination may also need your own route label to identify the offer. Use provider-supported labels only where permitted; do not rewrite issued links just to build a report.
Check that your own setup captures the intended action before comparing periods. Keep personal details out of the worksheet and tracking parameters; Google requires that personally identifiable information not be collected in Analytics.2 This guide does not install tracking or claim an account-level tracking test.
Worked example: more clicks, lower CTR
Every number below is fictional. Imagine the same patio-sizing page, the same search filters and two equally long, complete periods. No edit, ranking gain or commercial result is being claimed.
| Search observation | Period A | Period B |
|---|---|---|
| Impressions | 1,000 | 2,000 |
| Search clicks | 50 | 60 |
| CTR | 5% | 3% |
| Affiliate click events | Unknown | Unknown |
| Page-attributed approved commission | Unknown | Unknown |
Search clicks increased by 10, or 20%, while CTR fell by 2 percentage points. Neither “CTR fell, so the page failed” nor “clicks rose, so earnings grew” follows from this table. Inspect which queries and impressions changed before choosing a revision.
Now consider a separate fictional provider ledger: 80 USD approved before a 20 USD reversal leaves 60 USD approved, with 0 USD received. Its page and click cohort are unmatched. You cannot assign that 60 USD to the patio page or divide its orders by the page's search clicks to manufacture a conversion rate.
In a real offer, cash timing depends on its actual terms. Gardener’s Supply's public Impact template locks actions 27 days after the tracked month ends, then schedules approved payment 20 days after the lock month ends. That is a template schedule, not our accepted contract, a tested deposit date or a complete withdrawal policy.4 Read the profile's full limits before applying it to a forecast.
Copy the performance worksheet
Download the CSV worksheet. It includes the fictional search and provider rows above plus a replacement row. Open it in your spreadsheet editor; clear the examples before entering your own aggregate observations. Unknown means missing evidence, not zero.
For each row, retain the report and scope so you can explain where a number came from. Keep currencies separate. Record paid amounts independently of approved balances and keep costs elsewhere if you want to assess profit.
Choose the next useful action
| Signal | Investigate next | Avoid this conclusion |
|---|---|---|
| Search clicks or impressions change | Compare affected queries/pages and check technical issues, seasonality and relevant search changes.5 | One date change or new keyword caused the movement. |
| Few intended outbound events are recorded | Check event capture, destination grouping, page purpose and whether the reader had enough information to act. | Automatically add more affiliate buttons. |
| Outbound activity appears but commission is missing | Check provider reporting, eligible customers/orders, attribution, approval lag and any permitted route labels. Preserve unmatched page attribution. | All those clicks were buyers, or the program owes payment. |
| Commission is approved but receipt is absent | Reconcile the accepted schedule, holds, threshold, reversals, fees and payment statement.4 | Approved balance equals cash available today. |
Write one review note: observation → hypothesis → alternative explanation → next check. For the fictional patio page, the observation is more clicks with lower CTR. A reasonable next check is the query mix; rewriting the whole page is not yet justified. Google also advises against radical changes for small position fluctuations when a page already performs well.5
Use the free SEO lesson for page and search-readiness decisions, then the content-maintenance lesson when an actual claim or page needs correction.
Sources and open questions
Official Google documentation was checked October 8, 2026. The traffic-drop guide retains its December 10, 2025 update.5 Gardener’s Supply's public template was checked October 8 and reused without changing its source date. Other publication or effective dates were not established.
The worksheet is an original teaching resource with fictional observations. No Affiliate Orchard search baseline, merchant approval, tracked referral, payout, experiment or revenue result is claimed. Your report configuration, available attribution and accepted provider terms still determine what can be concluded.