"LinkedIn CPLs are too high, we're moving the budget." We hear some version of this most months, and the number behind it has usually been calculated against a definition of "lead" that nobody has ever written down, though sometimes moving the budget is the right call anyway.
What are you actually counting?
In a lot of the accounts we inherit, a lead is anyone who completed a form. That includes students, job seekers, competitors having a look around, and people who wanted the whitepaper and had no further interest in speaking to anyone.
Count all of that and Meta beats LinkedIn on cost per lead every time. We've seen the raw gap run to 10x.
Volume is the mechanism. Left loosely targeted, Facebook will hand you somewhere between ten and a hundred times the lead count LinkedIn produces, at a qualification rate low enough that most of it never reaches a sales conversation. Tightening the targeting raises that qualification rate and cuts the volume at the same time.
Even after all that noise, Meta sometimes still comes out ahead on cost per qualified inquiry. Not always, and not predictably enough to assume - but often enough that writing the channel off on principle costs people money. We compared the two properly here. You can't tell which platform wins until the definition is tight enough to make the comparison mean anything.
A cost per lead built on a definition that loose is a measure of what your forms cost, and people conflate that with what their pipeline costs when they decide a channel has stopped working.
Get sales to agree what's real
The fix is boring and organisational, which is probably why it gets skipped so reliably.
Sit down with sales and agree in writing what counts as an MQL and what counts as an SQL: which job titles, which company sizes, what has to be true before anyone calls it qualified. Then hold the campaigns to that definition rather than the platform's, which is generous by design. This matters more than it sounds, because sales and marketing quietly disagreeing about lead quality is how paid social programmes die - marketing reports volume, sales complains about junk, nobody ever reconciles the two, and eventually someone cuts the budget without anyone having established what was actually working.
In practice that document is usually a page, sometimes less.
The best-performing programmes we've run had weekly sales feedback on lead quality going straight back into targeting, and that closed loop is the single biggest difference we've seen between accounts that compound and accounts that plateau.
Judge on cost per qualified opportunity
Once the definition is tight, the number that matters is cost per qualified lead, and behind it pipeline. CPC, CTR and CPM are diagnostic inputs that tell you why that number moved.
Run cost per qualified opportunity and LinkedIn's premium often makes sense. The clicks cost multiples of Facebook's and the leads are routinely worth many multiples more. Where that ratio doesn't hold for your business, spend the money elsewhere, but you can only know which situation you're in once the definitions are honest.
Getting to the number takes measurement that follows a lead past the form fill and into the CRM. That's attribution and reporting work, and it is what settles the CPL argument, though it is less interesting than launching a campaign and usually the last thing anyone gets round to.
Expect the numbers to get worse first
Tightening the definition usually makes things look worse before they look better. Lead volume drops. The honest cost per lead goes up, sometimes considerably, and someone senior will ask what happened.
After that you start optimising toward the leads sales actually wants, and the platform learns from better signals. Cost per qualified opportunity then begins to fall. We've watched lead targets get better every quarter while total spend came down, on accounts that did nothing more exotic than this.
The accounts that struggle are usually the ones where three people in the business would give three different answers to what a lead is, and the price of the clicks has very little to do with it.
