In shortA cost per lead is only high relative to what this account has actually measured — the useful line is around 1.5× your own median, below which the movement is ordinary variance, and the first thing to check is whether the leads got more expensive or simply harder to find.
Nothing is broken. Leads are still arriving, the campaign is delivering, and the only thing that changed is the price. That makes this the finding most likely to be argued about, because whether it is a problem depends entirely on a number nobody agrees on: what a lead is supposed to cost.
Expensive compared to what
An industry benchmark cannot answer this. On a roster running a gym membership next to a plot of land, one number is wrong in both directions at once — it calls the gym’s leads a disaster and the land’s a bargain, and it is most confident exactly where it is most wrong.
The only comparison that means anything is the account’s own history: the median cost per result across its recent campaigns. That number is specific to this business, this market and this offer, which is what makes it a threshold rather than an opinion.
It also needs enough history to be a median rather than an anecdote. One or two campaigns is not a benchmark — a median drawn from a single campaign is just that campaign’s own cost, which would always judge itself as average.
The band where nothing needs doing
Cost per result moves week to week for reasons nobody controls. So a useful rule needs two lines and a gap between them: comfortably below the median is worth scaling, comfortably above is worth investigating, and the space between is noise you should not be sending anyone a message about.
In practice, about 0.8× your median at the bottom and about 1.5× at the top. Half your campaigns will always sit above the median — that is what a median is — so a rule that fires at the median alone produces an alert for half your account, every week, forever.
If the threshold fires on half of everything, it is not a threshold. It is a report with a warning icon on it.
What actually moved it
Cost per result is two things multiplied: what it costs to reach people, and how many of the people you reached converted. They have completely different causes and completely different fixes, so split them before doing anything.
- Cost to reach people rose — CPM is up. Usually competition: a seasonal surge, a bigger advertiser entering your auction, a narrower audience after an edit. Your conversion rate is fine and the leads are the same leads.
- Conversion rate fell — CPM is flat and CTR or conversion rate dropped. Something about who is being reached, or what they land on, changed. Check whether delivery moved toward a different placement or a different age band before blaming the creative.
- Both moved a little. Common, and usually the honest answer is a saturating audience: you have been running long enough that the easy conversions are already converted.
- Neither moved and the number still rose. Check tracking. A partially broken pixel or a lead form change makes results disappear while spend continues, and that arrives looking exactly like a cost increase.
How to check it
Compare two equal windows — last seven days against the seven before it, not against the campaign lifetime, which averages the change away. Put CPM, CTR and conversion rate side by side with cost per result, and read them in that order. One of them will have moved more than the others, and that one is the finding.
Then check the breakdowns for the same two windows: placement, region and age. A cost increase that is really a delivery shift shows up here and nowhere else, and it is the single most common cause that gets mistaken for a creative problem.
What to do about it
If the cause is competition, the honest answer is usually to accept the new price or reduce spend, not to rebuild the campaign. You cannot out-creative an auction that got more expensive, and the rebuild costs you the learning phase as well.
If the cause is a delivery shift, correct the delivery rather than the creative. If it is a conversion-rate fall with delivery unchanged, then the creative or the landing experience is the right suspect — and by then you will know that rather than assuming it.
And the thing worth checking before any of it: whether the leads got worse rather than dearer. A campaign whose cost per lead rose while lead quality also rose is a campaign that got better, and treating it as a problem is how a good week gets optimised away.
When Admetriq raises it
The actual rule, so you can hold this page against what you see rather than take its word for it.
- Fires when cost per result rises above 1.5× this client's own measured median (ABOVE_NORMAL_CPL_MULTIPLE = 1.5).
- The median must be this client's own, derived from at least three measured campaigns. With no derived median the rule stays silent rather than judging against a number tuned for somebody else's business — a generic benchmark would flag every client whose leads are worth more than the one it was tuned on.
- 1.5 is paired with the 0.8 used for “comfortably cheap”: under 0.8× the median is a scaling candidate, over 1.5× is genuinely expensive, and the band between them is ordinary variance nobody needs a card about.
- Requires at least one result in the window. A campaign with no results at all is a different finding entirely.
Most often mistaken for
Creative fatiguecreative_fatigue
Check frequency and click rate before touching the creative. Fatigue is frequency climbing past about three with the click rate falling as it does — the same people, seeing it too often, responding less. A cost per lead rising while frequency stays flat is not fatigue, and replacing the creative there restarts the learning phase in exchange for nothing.