Our Warmest Audience Was Our Most Expensive One

Every B2B playbook says the same thing. Upload your customer and prospect list, target those people, and your ads will work better because they already know you.

We ran that test properly this month for a B2B software client. Same ad, same creative, same lead form, same objective, two audiences, both live at the same time. One was a matched list of 59,000 people who had already received two emails from the company at a 33% open rate. The other was cold attribute targeting: job function and company industry, with the customer list explicitly excluded so the two never overlapped.

The warm list cost $41.64 a click. The cold audience cost $6.27.

That is not a rounding difference. The audience everyone would tell you to start with was six and a half times more expensive per click, and at the point it was switched off it had produced zero leads.

The numbers, both sets

The client is not named. The figures are exact.

Matched list ad set, final: $249.82 spent, 6 clicks, $41.64 per click, 0.49% click-through rate, zero leads. About $75 of its budget was never spent because it was switched off.

Cold attribute targeting, as of the most recent read: $200.51 spent, 2,805 impressions, 32 clicks, $6.27 per click, 1.14% click-through rate.

Same ad. Same landing experience. Same week.

Day one already said it, and we did not listen hard enough

The list set launched first. Its first full read looked like this: $68.09 spent, 374 impressions, 5 clicks. That works out to a $182 CPM and $13.62 a click at a 1.34% click-through rate.

Read that carefully, because it is the part people get wrong.

The ad was working. A 1.34% click-through rate on LinkedIn is three to four times what the platform’s own forecaster predicted for that audience. The list responded to the creative. The copy was fine. The offer was fine.

The cost was not the ad. The cost was the audience. A $182 CPM means we were paying $182 for every thousand times the ad was shown, and no amount of better creative fixes that. Creative moves click-through rate. It does not move what an impression costs.

That distinction is worth internalizing because it changes what you do next. If your click price is high and your click-through rate is bad, rewrite the ad. If your click price is high and your click-through rate is good, you have an audience pricing problem, and rewriting the ad will burn a week and change nothing.

Then it got worse in a specific, predictable way

The list set’s click-through rate did not hold. It went 2.09%, then 0.75%, then 0.49% across the flight. Between two reads late in the run it spent roughly $89 and bought zero additional clicks.

That decay is the mechanic that makes small warm audiences expensive, and it is structural rather than bad luck.

A matched list is a fixed population. Ours was 59,000 people after the US and English filters, out of 64,110 matched at an 85% match rate. That sounds like a lot until you are buying impressions against it every day. Frequency climbs fast. The people most likely to click do so early, and everyone remaining has already scrolled past the ad several times. The auction, meanwhile, is still charging you a premium to reach a narrow, tightly defined set of professionals who are also being bid on by every other advertiser targeting exactly those job titles.

The cold audience had 5.5 million people in it. Delivery could spread out. Frequency stayed low. Nothing had to be re-shown to the same person eleven times to spend the budget.

A small audience is not automatically an efficient one. Past a certain spend rate, a small audience is a guarantee of frequency, and frequency on a fixed list is how you pay full price to annoy people.

The platform told us in advance and we should say so

Before launch, LinkedIn’s own forecaster quoted the matched list at $20 to $45 a click and $181 to $213 per lead. Our earlier plan for the account had assumed $7 to $10 clicks, which was a number carried over from a different objective and a different audience type.

The forecast was right. We landed at $41.64, inside the range it gave us, at the expensive end.

The lesson is not that the tools are psychic. Forecasters are noisy, and this one gave us two different ranges for the same cold audience depending on which screen we read it from. The lesson is that the platform’s own worst-case number is the number you quote to a client before you spend their money. If the forecaster says $181 to $213 for the first lead and your plan says leads will cost $40, one of those is a hope. Quote the forecast, and if you beat it everyone is pleased.

We had also written kill lines into the plan before launch: a click over $12 or a click-through rate under 0.35% after the learning period means pause and rework. Having those written down in advance is what turned an uncomfortable conversation into an easy one. Nobody had to argue about whether $41.64 was bad. The threshold was agreed while everyone was calm.

What we would do differently

Start cold, then use the list for something else. Cold attribute targeting on LinkedIn found people at $6.27 a click. If the goal is volume at a sane price, that is where the budget goes. The customer list is better used as an exclusion, to keep paid budget off people you can already email for free, or as a seed for a lookalike audience, which is a much larger population built from the same signal.

Excluding the list from the cold set was the one thing we got right early. It meant the two sets never bid against each other for the same person and the comparison was clean. On LinkedIn this matters structurally: inside a single ad set, an uploaded list and audience attributes are combined with AND, not OR. You cannot widen an existing list-targeted ad set by adding job functions to it, because that narrows it to people who are on the list and match the attributes. Widening requires a second ad set, which is exactly why this ended up as a real side-by-side test.

Watch frequency, not just cost per click. The click price is the symptom. Rising frequency against a fixed audience is the cause, and it shows up first.

Do not read a single day and call it. Day one on the list looked expensive but explainable. It got materially worse over the following week. The reverse also happens. One day is a sample of one.

The honest gap in this story

We are not going to tell you the cold set produced leads, because that is still in dispute.

LinkedIn’s ad set table shows one lead on the cold audience. The Lead Gen Form surface for that exact form, read minutes apart, shows zero leads and zero test leads, and the download-leads button has produced no file across three separate attempts on three different days.

So there is a counter that says one and no record anywhere that can be opened, named, or handed to the client. A number that only exists on one screen is not a result. Until a record exists, nothing about a lead goes into a client report, and it is not going into this article either.

That distinction is worth its own discussion, and we will write it up separately, because platform counters disagreeing with platform records is a much more common problem than most advertisers realize.

What this does not mean

It does not mean warm audiences are bad. It means warm is not the same as cheap, and on LinkedIn specifically the two are frequently opposites.

It also does not mean LinkedIn is expensive. That was our first instinct when the blended number crossed the threshold, and it was wrong. One ad set was failing and one was working, and averaging them together produced a conclusion about the platform that neither set supported on its own. Do not let a blended number hide a working campaign inside a broken one. Read every ad set separately, every time.

Our paid advertising service page covers how we structure and govern this work, and if you want a read on what your current spend is actually buying, run a free audit.

FAQ

Why would a warm audience cost more per click than a cold one?

Because you are billed by impression and a matched list is a small, fixed population that other advertisers are also bidding hard for. Frequency rises quickly, the people inclined to click do so early, and the remaining impressions get progressively less productive while the price per impression stays high. A large cold audience lets delivery spread out and keeps frequency low.

Should I upload my customer list to LinkedIn at all?

Yes, for two things. Use it as an exclusion so paid budget is not spent reaching people you can already email at no cost, and use it as the seed for a lookalike or predictive audience, which turns the same signal into a population large enough to deliver against efficiently. Targeting the list directly can work for a short, high-value, tightly scoped campaign. It is a poor choice for sustained volume.

What is a normal cost per click on LinkedIn?

It varies enormously by audience, objective, and industry, which is why the useful benchmark is the platform’s own forecast for your specific setup rather than a published average. Ours ranged from $6.27 on a broad cold audience to $41.64 on a narrow warm list in the same week, on the same ad. That spread is the point.

Can I add job titles to an ad set that already targets an uploaded list?

You can, but it will narrow rather than widen. LinkedIn combines an uploaded audience and audience attributes with AND, so the result is only people who are on your list and match those attributes. To run both audiences you need two ad sets, which has the useful side effect of letting you compare them honestly.

How do I know when to switch an ad set off?

Decide before you launch. Write down the click price and click-through rate that mean stop, agree them with whoever is paying, and then read the numbers against those thresholds rather than against your feelings about the campaign. A kill line agreed in advance turns a difficult conversation into an arithmetic one.

Test it, do not assume it

The warm list was the audience everybody would have told us to start with. It was the most expensive thing we bought, and the cheap audience was the one nobody would have recommended.

You will not find that out from a best practices article, including this one. You find it out by running both at once, keeping them separate, and reading each set on its own numbers.

Want a second opinion on a campaign that is not working? Book a 15-minute call. No deck, no fluff. Strategy first. Tactics second. The work works.

Reliable PR & Marketing is a strategy-first marketing agency in Bakersfield, California. We run integrated SEO, PR, web, and content for founder-led companies across Kern County and nationwide. Strategy first. Execution always.