You ran Link Research on one domain, and the activity log shows five lines:
| When | Feature | What ran | Credits |
|---|---|---|---|
| Aug 18 4:24pm | Backlinks | anchors | −0.3 |
| Aug 18 4:24pm | Backlinks | referring_domains | −0.3 |
| Aug 18 4:24pm | Backlinks | backlinks | −0.3 |
| Aug 18 4:24pm | Backlinks | history | −0.2 |
| Aug 18 4:24pm | Backlinks | summary | −0.2 |
That is one action. It is not five charges for the same thing, and it did not cost you more than the estimate said it would.
A screen is not a purchase
The Link Research view answers five different questions, and each answer is a different dataset we buy on your behalf:
- summary — how many links and referring domains exist at all
- backlinks — the individual links
- referring_domains — the domains behind them, deduplicated
- anchors — the anchor text distribution
- history — how the profile changed over time
Our data provider sells those separately and prices them separately. We buy them separately, so we bill them separately. The five lines are the receipt.
The total is the number you were shown
Those five charges add up to 1.3 credits — which is the estimate the app showed you before the run, not something extra on top of it. We do not add a fee for splitting the bill into lines.
If we bundled them into a single "Backlinks — 1.3" row, you would pay exactly the same amount. You would just no longer be able to see what you paid for.
Why we itemise instead
Three reasons, all of which favour you:
You can see which part is expensive. On a domain with a large link profile, backlinks and referring_domains dominate and summary is rounding error. That is only visible because the lines are separate.
You can check our arithmetic. Every line is a real purchase with a real cost behind it. A single bundled number is something you have to take on faith.
A partial failure only costs you the parts that ran. If the anchors dataset times out, you are charged for the four that returned and not for the one that did not. That is only possible because each call is metered as it completes. A bundled charge would have to either bill you for the whole job or refund the whole job.
How to read the log
In Settings → Usage, the What ran column is the dataset, not the button you pressed. Charges from a single action share a timestamp — the five rows above are all 4:24 PM. Group by the timestamp and you are looking at one thing you did.
The Biggest spends panel above the log already groups by feature, so if you want the "one line per action" view, that is the one to read.
Keeping the count down
The number of lines is not something to optimise. The number of runs is:
- Re-opening a result you already ran costs 0 credits — it comes from cache, and no line appears at all.
- The estimate is shown before every run, so nothing starts without you seeing the figure.
- If a particular kind of research is spending more than you want, a credit budget caps it per project, per user, or per workspace.
Related
- What a credit is, and what uses them
- How pricing works — plans, credits, and caps
- Credit-safe workflows for busy months