A cost per lead is only high or low against the account’s own measured history — an industry benchmark averages businesses with nothing in common, so it calls one client’s leads a disaster and another’s a bargain while being most confident exactly where it is most wrong.
Put a gym membership and a plot of land on the same roster and ask what a good cost per lead is. There is no number that is not badly wrong for one of them. The gym sells a few thousand rupees a month to somebody who decides in an afternoon. The plot sells for lakhs to somebody who will visit the site twice and talk to their family for a month.
A lead worth two hundred rupees to one is worth two thousand to the other, and that is before you get to season, city, competition or how good the follow-up call is. Any single figure held against both is not a benchmark. It is an average of two things that were never comparable.
What a published benchmark actually measures
Benchmark reports are built by averaging the accounts a vendor happens to have data on. That sample is not the market. It is whoever bought that vendor’s software, which skews towards one company size, one set of countries and one kind of buyer.
The averaging then does the real damage. Cost per lead within any real category is not clustered around its mean — it is spread across an order of magnitude, with a long tail. An average across that distribution describes almost nobody in it, which is why every agency that has ever read one recognises their own accounts on neither side of the number.
An industry average is the cost per lead of a business that does not exist, quoted to businesses that do.
The only comparison that means anything
The account’s own history. What did this client, in this city, selling this thing, actually pay per lead across their recent campaigns? That number carries every variable a benchmark strips out, because it was measured inside them.
It needs enough history to be a median rather than an anecdote. One campaign is not a baseline — a median drawn from a single campaign is that campaign’s own cost, which would always judge itself as perfectly average. Three is the floor at which the number stops being one campaign’s luck.
That has an uncomfortable consequence worth stating plainly: for a brand new client, nothing can honestly tell you whether their cost per lead is good. Not a benchmark, not a tool, not experience. You are three campaigns away from being able to answer, and any system that answers sooner is guessing at you confidently.
Two lines, and a gap between them
Even against the right median, one line is not enough. Half of any account’s campaigns sit above its median — that is what a median is — so a rule that fires above the median flags half of everything, every week, forever. Within a month nobody reads it.
So you need a band. In practice around 0.8 times the median at the bottom and 1.5 times at the top: comfortably cheaper is worth scaling, comfortably dearer is worth investigating, and everything between is the ordinary week-to-week movement nobody should be sent a message about.
- Below 0.8×: cheaper than this account normally manages. Worth looking at — and worth checking the leads are real before you scale it.
- Between 0.8× and 1.5×: noise. Nothing happened.
- Above 1.5×: genuinely expensive for this client, by their own measure. Now ask what moved.
Then ask which half moved
Cost per lead is two things multiplied: what it costs to reach people, and how many of the people reached converted. They have completely different causes and completely different fixes, and the single number hides which one moved.
Reach getting dearer is usually an auction story — more competition, a narrower audience, a worse time of year. Conversion falling is usually a landing page, an offer, or a creative that stopped working. Treating the second as the first produces the classic mistake: cutting budget on a campaign whose form was broken.