Target CPA on lead gen with 15 leads a month: wait for 30 conversions or set it now?

Take a B2B agency getting about 15 form fills a month from Search on a fixed budget. The usual advice is to stay on Maximize Conversions until you hit 30 conversions in 30 days, then move to tCPA. At that volume, that could mean months on a bidding setup you can't steer very well.

I'd set the tCPA earlier, and base it on the cost per qualified lead you can actually afford, not the cost per form fill. Then send CRM outcomes back as offline conversions so the bidder learns what a real lead looks like. The first few weeks will be noisier and volume may dip, but at least you're optimizing toward something sales cares about.

The weak spot is lag. Qualification takes time, so the early data trails behind. Has anyone run this with a sales cycle longer than a month?

Setting it early is right, but I'd start the tCPA looser than the number sales says they can afford. At 15 leads a month, a target sitting right at your ideal cost can starve the auction and leave you with no data at all. Going 20 to 30 percent above your current qualified cost gives the bidder room to work, and you can tighten it once volume settles.

On the lag, check the click-to-conversion window on the imported offline action. If a deal closes after that window, the bidder never sees it, so your "qualified" count quietly undercounts the leads that actually closed.

Marco, I wouldn't start loose if the goal is qualified leads. If the bidder still treats form fills as the primary conversion, a target 20 to 30 percent above your qualified cost just tells it to go buy cheap fills that sales will reject. You'll get plenty of conversions, but mostly the wrong ones. I'd make the imported qualified action the primary conversion from day one so the bidder learns from it, and set the target against that signal. Volume will be thinner for a while, but that data is worth more.

Agreed on the primary action, as long as you're still getting around 10 qualified leads a month. Below that the bidder doesn't have enough to learn from. If you're under that, keep form fills as primary and set the target at your qualified cost times your fill-to-qualified ratio, so it still reflects what sales actually accepts.

The 10-lead threshold holds up, but the fill-to-qualified ratio is what goes stale. Recalculate it monthly, and only use leads old enough that sales has already qualified or rejected them. If you pull the last 60 days, you're mixing in fills sales hasn't judged yet, so the ratio looks worse than it really is. That drags the fill target too low, and the bidder starts buying cheap leads that sales will reject.

It matters because the bidder only sees the number you give it. A stale ratio means a wrong target, and the platform won't flag it. When you do adjust, keep changes to around 20 percent at a time. Bigger jumps can make the bidder relearn from scratch, and at 15 leads a month you can't afford that reset.

Agreed, and the cohort matters as much as the age. If you pull leads by the date sales judged them, the fills and the outcomes come from different months, so the ratio wobbles for no real reason. Group by the original click date instead. That's roughly how Google credits an offline import anyway, so fills from a given week and their qualified or rejected outcomes land in the same bucket, and your 20 percent adjustment rule ends up reacting to a steadier number.

It comes down to your sales cycle length, and that's worth checking before you copy the click-date grouping. If sales usually judges a lead within two or three weeks, a click cohort works fine, because form fills and outcomes settle at about the same time. If judgment takes two months, the newest cohort is still mostly unjudged, so it'll look bad every month. In that case you need a maturity cutoff more than a different grouping.

I'm also not sure Google's default reporting matches the click-date view for imports. Confirm which date your offline conversions are reported under before you rely on the 20 percent rule.

I'm not sure that holds at 15 leads a month. If you only trust leads older than 60 days, you could end up with a handful of judged fills, and two or three rejections can swing the ratio a lot. Then the 20 percent rule is just reacting to noise. I'd keep the full window, label the newest cohort as provisional, and only move the target when the older cohorts point the same way over two cycles. It's slower, but you won't chase a bad month.

Mira's right that at 15 leads a month the ratio is jumpy, and I'd guard against small samples as well as age. Don't let the 20 percent rule fire until you have a minimum count of judged leads, maybe 10 to 15 across the window. Two rejections out of six judged leads looks like a third of your leads are junk, but it could just be one rep who was out for a week. Sales teams also tend to clear backlogs in batches, which makes monthly ratios lurch. Calculate the ratio on judged leads only, and note when a big batch of outcomes lands so you can tell a real shift from a pile-up.