B2B with a 90-day sales cycle: optimize LinkedIn to demo requests or closed-won?

Picture a B2B software company running LinkedIn with demo requests as the conversion. Demos are cheap, so the platform's cost per lead looks great. Then sales reports that half of them are students or people from competitors, and the pipeline per dollar is weak. The platform number and the CRM number are telling two different stories.

Moving the goal all the way to closed-won sounds right, but it depends on volume. At something like 15 to 20 opportunities a month, the campaign doesn't have enough events to learn from and will sit in learning for ages. So I'd keep demo requests as the optimization event and send qualified-opportunity status back as an offline signal, then judge spend monthly against that instead of weekly against the platform's CPL.

How do you handle this when the sales cycle runs longer than the attribution window? Do you trust the platform's number at all, or rebuild it from CRM data?

I'd keep demo requests as the optimization event too, but check the SQL rate by job title and company size before you trust any CPL. Say a payroll software company gets 120 demo requests a month and 25 become SQLs. If that rate swings a lot by seniority or company size, you've got a targeting problem you can fix in the campaign, and you'll see it long before the 90-day closed-won number shows up. The offline feed helps here too. If CRM stage changes go back with the real timestamps, you can see which campaigns produce opportunities even while the deals are still open.

Splitting 25 SQLs by seniority and company size doesn't hold up. Each bucket ends up with three or four deals, so one rep's qualification habits or a single big account will swing the rate more than your targeting does. Pool it over a quarter before you act on any segment.

Fair, the buckets are too thin to act on monthly. The rate only tells you something if SQL means the same thing across reps, though, so before pooling, look at how many demos got marked SQL within a week of the call. If it's most of them, you're measuring rep habits, not targeting.