Define which kind of variety you are measuring
Backlink diversity describes variation in a set of references. Domain variety, destination spread, anchor wording and editorial context are different dimensions. Start with a question such as whether one publisher supplies much of the recorded evidence, or whether useful resources receive little attention. A larger count of categories is not a ranking guarantee, and this worksheet does not calculate a quality or safety score.
Keep a stable denominator and domain boundary
This exercise reuses the automation example: twelve supplied rows become nine distinct evidence records after two duplicate copies and one invalid row are accounted for. Use nine as the denominator for every dimension here. There are eight distinct source–target pairs and six root-domain groups; those are separate counts. The Public Suffix List explains why boundaries such as co.uk require more than taking the last two labels.
The analyzer uses public-suffix-aware grouping. A domain group is not proof of independent ownership, editorial independence, DNS availability or quality. Google’s Links report also documents root-domain grouping and sampling; do not directly compare that report’s totals with this differently collected exercise.
Read all three distributions without mixing their totals
The complete CSV has 16 groups: six domains, four destinations and six contexts. Each dimension sums to nine observations. The magazine domain has three records across two source pages, the lab has two, and four other domains have one each. The planner destination receives five, the dataset two, and the homepage and calculator one each.
Manual case notes classify two editorial, two resource, two research, one discussion and one paid-review observation. One remains unknown, so it stays in the denominator as 1/9. These labels were authored for the fictional case; the analyzer does not infer them from page content. The six-row table highlights decisions; the downloadable concentration file contains every group.
| Check | Fictional evidence | Interpretation | Next action |
|---|---|---|---|
| Counting unit | 9 distinct evidence records | One denominator for each dimension | Retain two duplicates and one invalid row in the raw log |
| Domain concentration | magazine.example: 3/9 | Three records across two source pages | Inspect context; do not infer three independent endorsements |
| Destination concentration | https://shop.example/planner: 5/9 | Five records point to this resource | Compare with the resource’s actual purpose |
| Known contexts | 2 editorial; 2 resource; 2 research; 1 discussion; 1 paid review | Eight manually classified records | Keep the supporting case notes |
| Unknown context | 1/9 | Unknown remains part of the total | Investigate without guessing |
| Complete distribution | 6 domains + 4 destinations + 6 contexts | 16 groups; each dimension totals nine | Do not sum dimensions into 27 links |
Investigate concentration in its actual context
The magazine’s 3/9 share is an inspection lead, not an automatic problem. Read those sources and ask whether repeated navigation, independent articles or repeated exports explain the count. Review the planner’s 5/9 share against its purpose: a widely useful planner may reasonably attract more references than a contact page. Do not add the three dimension totals together and call the result 27 backlinks.
Relationship attributes are another layer. Google’s guidance on outbound links distinguishes sponsored and user-generated references. An attribute alone does not describe an article’s complete editorial context. Countries, IP addresses, top-level domains and third-party scores also cannot establish independent ownership or relevance by themselves.
Turn a genuine gap into useful work
Load the shared input into the CSV analyzer to reproduce domain and target counts. Use the separate complete concentration sheet for the manually supplied contexts; a blank sheet supports your own categories. Keep raw evidence and explain unclassified rows. The six-step review CSV documents decisions rather than serving as analyzer input.
If an important research resource has few references, first check that it answers a real reader need and is accessible, accurate and worth citing. Examine relevant independent publications where that resource would improve an existing explanation. Record the reason for relevance and any actual editorial interaction, then verify later observations. Do not buy unrelated placements or manufacture anchor variants merely to fill a quota. A concentration pattern can suggest a question; it does not dictate a fixed acquisition plan.
Download the completed review · Download the blank review · Download the 12-row input · Download all concentration groups · Download a blank concentration sheet
Open the CSV backlink analyzer
Report findings with their limitations
Describe the collection, unit, duplicate rule, unknowns and next checks alongside the ratios. This case is not a live portfolio, and its nine observations are separate from the velocity guide’s 100-record starting inventory. There is no universal ideal domain mix, anchor percentage or safe follow ratio supplied here. The useful output is a transparent explanation of where evidence is concentrated and what additional information would change the decision.