Define the decision the report supports
A backlink report is a dated account of link activity, supporting evidence and relevant measurements. Start with the decision: accept a delivery, investigate a missing link or decide what to check next. One green total cannot answer all three. Keep a reported publication date alongside a separately dated observation, and name who supplied each fact.
Fictional teaching example: six records and eight recorded checks. No real campaign, publisher or traffic result is shown, and no sites are contacted.
One target page. Two inclusive 14-day periods: September 1–14 and October 1–14, 2026. Publication and check dates use UTC. Counts describe only the tracked records.
Keep the reporting window reproducible
For a repeatable comparison, save the original records, filters, extraction dates and timezone. This example deliberately compares two equal-length windows rather than whole months of different lengths. It includes a record once tracking has begun by the cutoff. Publication is cumulative; observation is the latest check inside that window. The table is an evidence exercise, not a live audit.
Use the latest check inside the selected period. Older checks are not carried forward; checks after the cutoff cannot establish earlier status. Row E’s October 16 check is excluded from October 14.
| Metric | September comparison | October review |
|---|---|---|
| Records | 4 | 6 |
| Reported published | 3 | 5 |
| Published during period | 2 | 2 |
| Observed present | 2 | 2 |
| Observed missing | 0 | 1 |
| Unresolved | 1 | 1 |
| Not checked | 1 | 2 |
Separate cumulative publication from period growth
Five publications by October 14 include three from earlier months and two new October publications. A cumulative count of five must not be reported as five new links this month.
The present share falls from 2/4 = 50% to 2/6 = 33.3% because the list expands. With the original four records it remains 50%. Neither percentage alone describes delivery quality.
The fixed cohort contains A–D in both periods. E and F enter only the full October list. The status filter changes the displayed rows, not the headline totals.
Explain changes hidden by an unchanged total
A and B were present in September; A and C are present in October. The total stays at two while the underlying links change. C is an older publication with newly resolved evidence, not a new placement.
| Record | September comparison | October review |
|---|---|---|
| A | Observed present | Observed present |
| B | Observed present | Observed missing |
| C | Unresolved | Observed present |
| D | Not checked | Unresolved |
| E | Not yet tracked | Not checked |
| F | Not yet tracked | Not checked |
All eight invented observations are retained here, including the later check excluded from the October snapshot. A check records a point in time, not continuous availability.
Read traffic and search measurements in their own scope
GA4 Traffic acquisition describes sessions by acquisition dimensions. Use a specific source and landing-page scope when discussing referrals; a visit count is not an inventory of links. Search Console’s Links report is a sample and may include historical entries. Neither export should silently replace the dated check ledger. GA4 · Referral sessions
Search Console · Read a backlink report
Search clicks answer a different question from referral sessions. Keep the search type, page filter, country and device scope fixed within each comparison. Search Console daily dates use Pacific Time; a UTC observation log has different day boundaries. Show the discrepancy instead of merging the numbers into one performance score. Search Console · Measurement scope
In this invented example, referrals rise by six and search clicks by ten. Describe those as measured differences within their stated scopes, not gains attributed to the two new publications. Timing, other activity and measurement changes remain possible explanations. No key-event data was supplied, so an ROI or conversion claim would be unsupported.
GA4-style session source / medium = alpha.example / referral, landing page /guide; UTC property; the same two 14-day calendar ranges. This covers one publisher, not the whole network.
Search Console-style exact target page, Web search, all countries and devices; the same calendar ranges in Pacific Time. These day boundaries are not aligned with UTC link checks.
No key-event measurement supplied for either period. Missing values are not zero conversions.
Assign the next action to each exception
One known missing observation needs follow-up; one unresolved check needs evidence; two unchecked rows need scheduled work. Unknown outcomes remain unknown. A zero baseline has no finite percentage growth.
| Observed link status | Owner | Next action |
|---|---|---|
| Observed present | Report analyst | Retain the dated observation; schedule the next review. |
| Observed missing | Publisher liaison | Repeat the check and compare page evidence before contacting the publisher. |
| Unresolved | Report analyst | Resolve the access or inspection problem; do not classify it as a lost link. |
| Not checked | Report analyst | Schedule a first check within the next review period. |
Request the publication URL and date, then plan a check.
Reuse the example for your own report
Use the HTML as a review handoff: both periods, the transition table and all eight checks travel together. The two CSV files are period snapshots, not a complete history. The October JSON restores the report builder’s ten-column evidence and native notes. Replace the example records and provenance before using it with a client; exporting a file does not verify its claims.
Downloads always contain the full example, independent of the controls. Restore the October JSON in the report builder to edit its scope and notes. Proposed next actions have not been performed.
Choose evidence before choosing software
For free report searches, this is a free worked example, not a free crawl of any domain. An arbitrary-site report needs a permitted data source and stated coverage. A software shortlist should be judged on reproducible exports, separate status dates, provenance and retention. This guide does not rank vendors or claim complete keyword-and-backlink discovery.