Google Search Partners: keep them on for lead gen or switch them off?

Search Partners is on by default in most Search campaigns, and the usual case for keeping it is cheap extra volume. My concern is what it does to bidding. Smart Bidding optimises toward a single target, so if partner clicks convert at a fraction of the rate of Google's own results, the tCPA gets dragged toward the weaker traffic and the blended number hides it.

Take a plumber getting 40 leads a month from Google and 10 from partners at twice the cost per lead. The total looks acceptable, but the partner share is the reason the account costs more than it should. Segmenting by network in the Segments view shows this quickly. I'd only keep partners if the split shows lead quality holding up, not just cost per lead. Has anyone run that split and kept them on?

I don't think the tCPA point holds. Smart Bidding doesn't average partner clicks into one target and drag the whole account down with them. It predicts conversion rate for each auction and bids accordingly, so weak partner clicks mostly just get bid down.

The real problem is visibility. Partner search terms are harder to review, and a form fill can look fine in Google's data while the lead turns out to be junk once someone actually calls it. So judge partners on qualified leads after the call, not on the network split in Segments.

Mostly agree, but "weak clicks just get bid down" leaves something out. The bid is set per click, yet the model learns from one shared pool of conversions. If partner form fills are junk, they teach it that a cheap-looking form is a good outcome, and that lesson can spill over into Google's own traffic.

The fix is to import qualified leads as offline conversions, so the model learns from the sales call and not the form. Once that's in place, the network split matters a lot less, because the signal is cleaner whichever network the click came from.

Mostly with you, but I'm not sure the spillover holds. The model conditions on auction signals like device, location and network, so a junk partner form fill doesn't teach it that cheap forms are good everywhere. The bigger catch is the import itself. Offline conversions match back to the click by GCLID, so if the form doesn't write the gclid into a hidden field, the sales outcome never joins to anything. Test that field end to end with a real submission, and expect the offline data to lag by a few days.

Fair point on the GCLID join. A broken join fails quietly, so test it before you trust anything else. Once the data flows, judge partners on cost per qualified lead, not cost per form fill, and only once each network has around 30 qualified leads in the period. Below that, one good or bad week decides the answer and you'll keep flipping the setting.

If partner cost per qualified lead lands within roughly 20 to 30 percent of Google's own results, keep them on and leave them alone. If it's well above that, either switch them off or move them into their own campaign with a small budget, so you can keep watching them without them eating spend in your main account. The network split only tells you where the money went. The sales outcome is what tells you whether it was worth spending.

Agreed on the 30 qualified leads floor. I'd add one rule for the setup itself: if partners can't be split out from Google's own traffic, the offline import just averages the two together. Honestly, I'd put partners in their own campaign from day one, so you can read the sales outcome per network without digging.

Splitting partners into their own campaign makes sense, and the offline import is a big reason why. Google's offline conversion import reports against the campaign and ad group the click came from, so if partners sit in a mixed campaign, the sales outcome gets credited to whatever blend of traffic was there and you can't untangle it later. Separate campaigns keep the qualified-lead count honest per network. The catch is that a separate partner campaign has a smaller pool, so it takes longer to hit the 30-lead floor, and you'll need a longer window before the numbers mean anything.

I don't think the attribution point holds. The offline conversion is tied to the original click through the GCLID, and that click still carries its network, so in a mixed campaign Segments by network should still separate qualified leads from partners, as long as the join works.

The real case for a separate partner campaign is budget control and bidding. Partners can't pull spend away from Google's own results, and Smart Bidding can learn each pool on its own.

I'd test the network split on the first batch of imported leads before rebuilding the structure around a limitation that may not exist.