Google Ads Search partners: keep them for volume or cut them for cleaner clicks?

Search partners are still on by default for a lot of Search campaigns, so the keep-or-kill question keeps coming up. The case for keeping them is volume. Partners add impressions and some cheap clicks, which matters on smaller accounts where Google.com inventory is thin. The case against is that partner clicks land on sites you didn't choose, and the search terms report tells you very little about where they came from.

I think it comes down to whether the numbers hold up. If conversion rate and cost per conversion on the partner network are close to Google.com over a few weeks, keep them. If partners eat a big share of spend and conversions lag, switch them off and compare cost per conversion before and after.

What I'm least sure about is how to read the split when a campaign only produces a handful of conversions a month. Has anyone found a sample size where they actually trust it?

Honestly, conversion volume matters more here than the partner ratio. With only a handful of conversions a month, the split won't tell you much, because one partner conversion more or less can swing cost per conversion a lot and you'd mostly be reading noise. At that volume I'd skip the split and run the before-and-after test you described, but over six to eight weeks and against the same weeks last period, so seasonality doesn't muddy it. If the account does a few hundred conversions a month, the split becomes readable within a couple of weeks.

On Microsoft Ads specifically, the partner slice is usually a much smaller share of total volume than on Google. That weakens the case for keeping partners on volume alone, so check what share of your clicks and conversions actually comes from them before you spend time on the split.

I'm not sure a before-and-after test works at that volume. Say a plumbing company gets six conversions a month. Eight weeks gives you about a dozen in total, and one cold snap or a slow week of calls will move cost per conversion more than switching partners off ever will. You'd end up with a number you can't separate from seasonality.

For local services I'd judge partners on lead quality instead. Pull the call durations and the junk form fills from the last month, because partner clicks often bring the wrong kind of contact even when the cost per lead looks fine.

Agreed on lead quality, and I'd settle the cutoff before you look at any numbers. At six conversions a month, whatever you see afterward is going to read like a story. Pick a rule now, something like: if partner leads that pass a real qualification check (a call over 60 seconds, or a form that gets a reply) run more than a third below Google for two months in a row, switch partners off. Otherwise leave them on. A preset rule keeps you from reading noise as a trend, which is the main risk at that volume.

I'd keep the preset rule but change the threshold. At six conversions a month, "a third below Google" can come down to one or two calls. If partners bring in three qualified leads and Google brings in four, a single call that runs 40 seconds instead of 70 flips the result, and you'd be switching off a channel on coin-flip data. Two bad months in a row sounds strict, but at this volume it happens by chance all the time.

I'd base the rule on counts instead. Switch partners off only if they produce zero qualified leads over the test window while Google produces at least two, or if cost per qualified lead runs more than double Google's. Those triggers are blunt enough that one slow week can't talk you out of them. The window also has to be long enough to hold a real count, which at this volume probably means three months, not two.