Broad match with Smart Bidding on a new account: start broad or phrase?

Been going back and forth on this for a new account doing maybe 30 conversions a month. The argument for starting broad with Smart Bidding is that Google needs query variety to learn who converts, and phrase or exact starves it of signal. The argument against is that broad on a small budget burns the first few weeks on queries that look nothing like your offer, and the learning phase never gets clean data anyway. At low volume I think phrase match with a tight negative list teaches the bidder more per dollar, because the query mix stays readable. Broad can come later, once you have a few months of search terms to build negatives from. Has anyone seen broad win early on a low-volume account? If so, I'd like to know what the conversion count was when they switched.

I can't give you a switch number from experience, and I'd be wary of any number that comes without context. It mostly depends on how clean the conversion action is. If it's a qualified lead or a sale, broad can surface queries you'd never have thought of, and the bidder has something real to learn from. If it's a soft action like a form start or a page view, broad will chase whatever gets people to that event, so phrase match with a tight negative list keeps it closer to the offer. Thirty a month is close to the usual guidance either way, which is why I'd look at the conversion definition first.

Agreed on the definition. If a soft action like a form start sits in the same goal Smart Bidding optimizes for, broad match will buy cheap starts no matter what match type you pick. Moving it to a secondary goal can matter more than the keyword choice. Is the form start in your primary goal right now?

Agreed, and the quickest check is the Conversions column versus All conversions. Conversions only counts your primary goals, so if All conversions is much higher, the gap tells you how many of those 30 are soft starts. Once you demote the form start, the bidder stops chasing it, but the signal it learned from will shift. So don't judge broad against phrase on a month where the goal mix changed, because you'd be comparing two different optimization targets.

Change one variable per test. If you're comparing broad against phrase, keep the conversion goals fixed. And if you need to demote the form start, make that its own change and let it settle first. The cleanest way to compare is a Google Ads experiment with a 50/50 split, since both arms share the same goal and the same dates. A month-over-month comparison mostly measures the goal change, not the match type.

50/50 is the right split, but at 30 conversions a month each arm only gets about 15, which is too few to tell broad from phrase with any confidence. One cheap week can make an arm look like a winner and then flatten out. Plan for at least two full conversion cycles, and longer if your click-to-sale lag runs a week or more, because the last few weeks of conversions haven't landed yet.

Agreed on the lag. Also look at budget-lost impression share in each arm. Broad usually enters more auctions, so if one arm is capped by budget, you're really comparing spend levels rather than match types.

Budget-lost share is the right thing to watch, and I'd make it a hard rule before launch. If either arm loses more than roughly 10 to 15% of impressions to budget, fix the shared budget or shrink the test before you read any results. Otherwise the cheaper arm can look efficient just because it entered fewer auctions.

Write the stop rule down too, something like two full conversion cycles plus a cost-per-conversion gap you'd actually act on. That way you don't end up calling it after one strong week.