Your CPL looks great — until you open the pipeline report

1 What I look at (before I ever look at CPL)

• The search terms report, first. For B2B software that costs real money, one-word or adjacent-category search terms are a red flag on their face. "Tax software" for a practice-management tool is a consumer doing their taxes, 99 times out of 100. The keyword can look hyper-specific and the traffic still be B2C.
• Brand hiding inside non-brand. Your own brand name converts at roughly a fifth of the cost of everything else. When brand searches leak into a non-brand campaign, they drag the whole CPL down and make a mediocre campaign look like a star — and you learn nothing about whether your real targeting works.
• A 60-second CRM spot-check. If the "conversions" are mostly Gmail and Yahoo addresses, there's your answer. Businesses buying real software tend not to sign up with a personal email.
• The attribution lens — before I condemn anything. The same spend can look dead on first-touch and roughly double on multi-touch. I confirm it's really the leads that are bad, not a reporting view making a working channel look worse than it is.
• Then the funnel math. Leads a month vs. opportunities a month vs. what a deal is worth. Cheap conversions against a high ACV isn't a bargain — it's the tell.

2 What I usually find (one real account)

Almost always: the account is optimizing beautifully — for the cheapest possible conversion. And the cheapest conversion is almost never your buyer.

Under the hood, two things were doing it:

• About half the "conversions" in the bottom-funnel campaign were the company's own brand name (converting at ~1/5 the cost) — flattering the CPL and hiding how the real non-brand targeting performed.
• There was no down-funnel signal going back to Google. Every form fill counted the same, so the algorithm had no idea which leads became pipeline. It was faithfully optimizing toward "a form got filled," over and over.

3 What I changed

• Pulled brand out of the non-brand campaign and split it into its own, so the reporting finally told the truth about what non-brand actually costs and converts.
• Fed a real down-funnel signal back to the platform — the "mid-market or bigger" selection at the form, then SQLs and won deals via offline conversion tracking — so the machine could learn what a buyer looks like instead of what a form-fill looks like.
• Killed the premature automation. It was running automated bidding on ~14 conversions a month — nowhere near enough data — so it was chasing noise. Reverted until there was enough signal to trust.
• Added a couple of honest quality signals (pricing and integrations views, real session depth) to optimize toward while the pipeline data built up.

4 What happens next — and why it fixes your exact problem

What that changes for you: the next time you open the pipeline report, the leads in it are the ones the ad account was built to produce. And when the board asks "what's paid doing for pipeline," you have an answer that survives the question — because you're finally measuring the thing they care about, not a vanity number.

One contrarian note worth keeping: PPC tends to bring in bigger deals. It can run a slightly lower close rate than your other channels and still win on deal size. Judge it on raw lead count and you'll underrate the one channel that's actually feeding pipeline.