After this lesson you should be able to
- Diagnose a stalled campaign against six specific failure modes
- Use the reply and conversion ratios to locate where the pipeline broke
- Distinguish a bad list from a bad pitch before rewriting either
- Decide whether to fix, restart or abandon a campaign
Diagnose before you change anything
When a campaign underperforms, the instinct is to change the tactic. That is almost always wrong, and it is expensive, because you throw away the list, the asset and everything you learned along with the problem. Campaigns fail in a small number of recognizable ways, and five of the six are fixable without abandoning the campaign.
Two ratios locate the failure before you touch anything. Replies divided by first contacts tells you whether the list and the pitch are working. Links divided by replies tells you whether the offer is working. A collapse in the first points upstream, to prospects and pitch. A collapse in the second points at the thing you are asking people to link to. If both look fine and there are still no links, the problem is usually time, and time is failure mode six.
Work through the six in order. They are ordered by how often I see them, and by how much damage they do before anyone notices.
Failure one: the prospects were wrong
This is the most common cause by a distance, and it is the hardest to see from inside the campaign, because a bad list produces exactly the same symptom as a bad pitch: silence.
The signature is a reply rate near zero across a large sample with no pattern in who does reply. A list that is merely mediocre produces a low but non-zero reply rate with the replies clustered in one segment. A list that is genuinely wrong produces nothing anywhere.
Three ways lists go wrong. Relevance was assumed rather than checked — the sites are in a related industry but have never linked to anything like your page. The pages are not alive — sourced from an index, never verified, and a meaningful share are unmaintained, deindexed or abandoned. The site does not link out at all, which is common among corporate publishers and is trivially checkable before contact.
The fix: take 20 rows at random and manually check each one. Does the site link out editorially? Has it published on this topic in the last year? Would a reader of that site want your page? If fewer than half pass, stop sending immediately and rebuild the list. Do not rewrite the pitch first; a good pitch to the wrong people is indistinguishable from a bad one.
Failure two: there is nothing worth linking to
The signature here is distinctive and easy to miss because it looks like success at first: people reply, they are friendly, they say they will take a look, and nothing happens. Replies over contacts is healthy. Links over replies is close to zero.
What has happened is that you asked a reasonable person to do something they could not justify to their reader. The page you pointed at is a service page, or a thin post, or a competent article on a topic already covered better elsewhere. Nobody wants to say that in a reply, so they say they will look at it.
This is the failure mode that most campaigns actually have and almost nobody diagnoses, because diagnosing it requires admitting the asset is not good enough. The test I use is simple and unforgiving: would you link to this from a site you owned, if you had no relationship with the author? If the answer requires a qualification, it is a no.
The fix: stop outreach and rebuild the asset. That may mean original data, a genuinely comprehensive reference, a tool, or a piece of expertise nobody else in the market has published. Restarting the same outreach against a better asset usually converts, and you get to reuse the list, which is why this failure is worth catching before you have burned the whole list against the old asset.
Failure three: outreach that reads as templated
Signature: reply rate low but not zero, and the replies you do get are curt or hostile. Somebody occasionally tells you to stop emailing them.
Editors receive an enormous volume of this. They recognize the shape of a template instantly, and the tells are consistent: an opening compliment about a specific post that could apply to any post, a paragraph explaining who you are before saying what you want, a formulaic transition into the ask, and a signature block that suggests scale. Merging a first name and an article title into a template does not fix this. It is the structure that gives it away, not the personalization tokens.
There is a specific version of this that deserves naming: the pitch that is personalized but obviously assembled by a machine. Fluent, well-formed, oddly generic in its specificity, and increasingly common. Recipients have learned to spot it and it is starting to depress reply rates for everyone.
The fix: write ten emails by hand with no template open and compare their reply rate to the templated batch. If the handwritten ten do better, you have your answer, and the correct response is not a better template but a shorter one with more of the message written per recipient. Structure the message around what the recipient gets, be specific about their page rather than their site, and be brief enough that the ask is visible without scrolling.
Failure four and five: no follow-up, and the wrong metric
No follow-up
A large share of placements come from the second or third contact, not the first. People are busy, mail gets buried, and an unanswered first email is usually not a refusal. Yet the follow-up is the first thing dropped, because week 7 is when the workload doubles and following up feels like nagging.
The signature is a campaign whose activity chart shows a burst of first contacts and almost nothing after. The fix is scheduling rather than willpower: when the first email is sent, put both follow-up dates in the sheet at the same moment. Two follow-ups, spaced roughly a week and then two weeks, adding something small each time rather than repeating the ask. Then stop. Three touches is persistence; five is harassment and it costs you the relationship.
Measuring the wrong thing
Some campaigns are working and get killed anyway because the reported number does not show it. The classic case is reporting total backlinks, where one sitewide placement creates a spike that has nothing to do with the campaign, or where a genuine month of fifteen new referring domains disappears inside noise from a link count in the thousands.
The fix: report new referring domains attributable to the campaign, against a dated snapshot taken before it started. Module 8 covers this in detail, including why the number can move fivefold with no editorial decision behind it.
Failure six: stopping at three weeks
The last failure mode is the one that kills the most campaigns and gets the least attention, because it never produces a post-mortem. Someone runs outreach for three weeks, sees two links, decides the tactic does not work, and stops.
Three weeks is roughly when a well-run campaign is at its worst-looking. The first wave has gone out, most of it has not been answered yet, follow-ups have not landed, and the placements that will come from weeks 5 through 16 do not exist yet. Judging at week 3 is like judging a crop at the point where the field looks like mud.
The commitment that makes the difference is small and specific: send for eight weeks, follow up twice on everything, and count at week 16 rather than week 12. If you cannot commit to that, do not start the campaign, because a campaign abandoned at week 3 has all the cost and none of the return.
When to actually stop. There is a genuine version of stopping and it looks different. Stop when a rebuilt list and a rebuilt asset have both failed to move the ratios across at least two full cycles, or when the tactic turns out to depend on a supply that does not exist in your market — no resource pages, no journalists covering the subject, no existing mentions to reclaim. That is a real finding and it justifies changing tactic. Impatience does not.
Questions
How do I tell a bad list from a bad pitch?
Check the list manually first. Take 20 rows at random and verify that each site links out editorially, has published on the subject recently, and serves readers who would want your page. If fewer than half pass, the list is the problem. Only when the list survives that check does rewriting the pitch tell you anything, because a good pitch to the wrong audience produces the same silence as a bad one.
How many follow-ups is too many?
Three touches total: the first email and two follow-ups, roughly a week and then two weeks later, each adding something rather than repeating the ask. A fourth rarely converts and a fifth damages a relationship you may want later. The failure in practice is almost never excess; it is stopping after the first email because following up feels like nagging.
My campaign got replies but no links. What now?
That pattern points at the asset, not the outreach. People were willing to engage and could not justify the link to their readers. Stop sending, rebuild the thing you are pointing at until it is something you would link to from a site you owned, then restart against the same list. Reusing the list is why catching this early matters.
When is it right to abandon a tactic entirely?
When the underlying supply is absent in your market, or when a rebuilt list and a rebuilt asset have both failed across two full cycles. No resource pages exist in your subject, no journalists cover it, no mentions exist to reclaim: those are real findings. Two quiet weeks is not a finding, and switching tactic on that basis throws away the list, the asset and the learning at once.