Every link building agency rejects domains. It is not optional — prospecting throws up far more candidates than any standard would accept, and someone has to decide which ones fail. What varies enormously is whether that decision is recorded, whether the reasons are written down, and whether the client ever sees any of it.
In our experience, almost nobody shows it. Not because agencies are hiding something dramatic, but because the log is genuinely awkward to keep and genuinely awkward to share. Which is precisely what makes it a good question.
Why the log is uncomfortable
Consider the position of an agency in the third week of a month with a delivery target of sixteen links and eleven live so far. The prospecting queue contains a domain with decent metrics, a slightly thin topical history, and an anonymous byline. It is a marginal call.
If nobody records the decision, the marginal call quietly becomes a yes. Not through dishonesty — through the ordinary pressure of a number that has to be met. Repeat that a few times a quarter and the standard has moved without anyone deciding to move it.
A rejection log is not primarily a transparency device. It is a mechanism for making it costly to lower your own bar when nobody is looking.
That is why we publish ours to clients monthly, and it is also why publishing it is uncomfortable: it means writing down, every month, the sites we would have bought if we were being slightly less careful.
What a real one contains
A rejection log is not a list of domain names. It is a list of decisions, and each one needs a reason code specific enough to argue with.
| Reason code | What it means | Typical share |
|---|---|---|
| TRAFFIC | Below the minimum verified organic sessions threshold | ~30% |
| RELEVANCE | Topical overlap score below threshold for this client's category | ~22% |
| HISTORY | No coverage of the topic predating our outreach | ~14% |
| IDENTITY | No named author, no contactable editor | ~12% |
| FOOTPRINT | Shares host, template or outbound pattern with other inventory | ~11% |
| OUTBOUND | Excessive or commercially patterned external linking | ~7% |
| CLIENT | Excluded by the client — competitor adjacency, brand safety | ~4% |
The proportions above are roughly what ours look like across a year. Yours will differ by category. What matters is that the codes exist, that they are applied consistently, and that a client can point at any row and ask why.
What the log tells you that the delivery report cannot
Three things, none of which are visible from a list of live links.
Where the bar actually sits
Two agencies both deliver sixteen links a month. One rejected forty domains to get there; the other rejected four. Those are completely different products at potentially identical prices, and only the log distinguishes them.
Whether the bar moves under pressure
Look at the rejection rate across months rather than in one month. A rate that dips in the months where delivery was behind is the single clearest evidence that the quota is driving the standard rather than the other way round.
Whether prospecting is actually happening
An agency drawing from a fixed inventory list has very little to reject — the vetting happened once, years ago, when the relationships were established. A high rejection count is indirect evidence that fresh prospecting is genuinely running each month.
How to ask without sounding hostile
The framing that works: "I'd find it really useful to see what you turned down and why — partly so I understand the standard, partly so I can flag anything on the excluded list that I'd actually be happy with."
That second clause is genuine and it changes the temperature of the request. Sometimes a client is comfortable with a domain we rejected — a smaller publication they know is read by exactly their buyer, for instance — and the log is how that conversation starts.
What a bad answer sounds like
You will hear some version of these. None of them are disqualifying on their own; all of them are worth one follow-up question.
- "We don't really track that." Fair, if true. Then ask what percentage they estimate, and whether they would start tracking it. The willingness matters more than the current state.
- "Our vetting happens automatically." Ask to see the criteria. Automated scoring is good; unexaminable automated scoring is a black box.
- "We only work with pre-approved partners." This is the inventory answer. It means prospecting is not really happening — you are buying from a list.
- "That's proprietary." The domains they rejected are not proprietary. The scoring model might be; the outcome is not.
A caveat, in fairness
A high rejection rate is not automatically a virtue. An agency rejecting nine in ten prospects may simply be prospecting badly — casting too wide a net and filtering afterwards, rather than targeting well in the first place. What you want is a rate that is stable, explained, and consistent with the number of placements actually delivered.
Ours sits around seven in ten. If it dropped to two in ten we would take that as evidence that our prospecting had got sharper. If it rose to nine in ten we would take it as evidence that it had got lazier. Either way, we would only know because we count.
If you are already in an engagement
Ask retrospectively. "Could you send me the domains that were considered and not used over the last quarter?" If they exist, you will get them. If they do not, you will get a slightly evasive email, and that is also an answer — one worth weighing before the next renewal.