LinkedIn: target job titles or upload a named account list for a small B2B market?

If you're selling a sales-led B2B SaaS product to finance teams, LinkedIn's job title, seniority, company size and industry targeting gets you an audience that's big and only roughly right. The other route is uploading a named list of around 300 target companies and showing ads only to people at those firms. My instinct is that the account list wins when the total market is small. You're paying to reach the dozen people who can actually sign, not "CFOs at 200-500 person companies" across the whole country.

Where I'm less sure is when the list gets too thin. Match rates on uploads can be poor, and then a few hundred people end up seeing the same ad every day, so frequency climbs fast. Has anyone run both against the same pipeline? Did the account list produce more meetings, or just a cheaper click?

Check the audience size first. Three hundred companies isn't a few hundred people. Each firm has a finance team plus everyone around it, so the matched audience is often several thousand profiles. Frequency only climbs fast if you narrow it, so keep the account list but layer job function and seniority on top. That way you're reaching finance leads at those firms instead of everyone who works there.

On matching, upload web domains where you have them, not just company names. Names match inconsistently because of legal suffixes, "Group" versus "Holdings," and old trading names, and domains cut most of that noise.

For the meetings question, compare pipeline per account rather than cost per click. A cheap click from the wrong seniority is still a wrong click.

Agreed on domains, and pipeline per account is the right metric. My worry is that it'll flatter the ads for a reason that has nothing to do with them. If those 300 companies are already in outbound, some of that pipeline is your SDRs' work, and the ads get credit for meetings that were coming anyway. A holdout would show it: keep a slice of the same firms out of the audience and see whether they book fewer meetings. Are these accounts already in a sequence?

Holdout is the right test, but whether you can read it depends on how many meetings those firms actually produce. In a small market, 300 accounts might only generate a handful of meetings a month. Hold out 50 firms and you could end up comparing two meetings against one, which tells you nothing and lets you believe whatever you hoped.

So get the baseline before you split. Pull last quarter's meetings per account from this list, then size the holdout so the test group would plausibly produce 15 to 20 meetings over the test window. If the numbers don't get you there, run it longer or accept that it can only detect a large effect.

And yes, separate out any firm already in an active sequence. Otherwise the SDR activity swamps both groups and the holdout stops measuring the ads.

I'd push back on pulling the active-sequence firms out. That's exactly where the ads and the SDRs overlap, so it's the question you most need answered. Drop those firms and the test only shows what ads do on accounts nobody is working, which is a much narrower question. Randomize within each group instead, sequenced and not sequenced, so both arms carry the same SDR load. Then you can read the ad effect inside each stratum, and the SDR activity mostly cancels out instead of swamping the result.