Is broad match safe now that smart bidding exists?
Broad match works when the bidding system has clean conversion signal to steer with, and burns budget when it does not. It is less a keyword setting than a permission slip: you are telling Google to go find demand it believes will convert. If your conversion definition is loose — every call counted, every form counted, no values — broad match will faithfully buy more of the wrong thing, efficiently.
Broad match is an input to bidding, not a keyword decision
Under manual bidding, broad match was reckless because nothing filtered the matches. Under smart bidding, the match is only the first step — the system then decides whether to bid at all, and how much, based on what it predicts about that specific auction.
This is why the same setting behaves completely differently in two accounts. The difference is not the keyword. It is the quality of the conversion history the prediction is built on.
The prerequisites, in order
- A conversion that means something. A qualifying lead, not a click-to-call tap or a page view.
- Enough conversion volume. A campaign producing a handful of conversions a month has nothing to learn from.
- Values that differ. If every conversion is worth the same, the system cannot tell a repipe from a washer replacement.
- A maintained negative list. Broad match finds adjacent demand, and some adjacency is a different business entirely.
How it fails in service businesses specifically
The characteristic failure is trade drift. A plumbing keyword on broad match starts matching handyman queries, then appliance repair, then general home repair. Each hop is defensible on its own and the sequence ends somewhere you do not serve.
The second failure is employment traffic. Trade searches sit next to trade job searches in the language model's view of the world, and without hard negatives you will buy applicants. The third is product research — people looking to buy a unit rather than hire someone to install one. All three convert into something, which is why they survive if your conversion definition is loose. Fixing the definition is the real fix, and it is the same work that makes marketing intelligence trustworthy.
A safer way to test it
Put broad match in its own campaign with its own budget and its own negative list. Do not mix it into a campaign with exact and phrase keywords, because you will not be able to attribute the change to anything.
Then judge it on cost per booked job rather than cost per lead. Broad match reliably lowers cost per lead and frequently raises cost per job, and only a connection between the ad platform and your field service system will show you which of those happened.
Topics: broad match · smart bidding · match types · keyword strategy
Have a version of this question about your own business?
The useful answer usually depends on which systems you run and how they're connected. That's a conversation, not a blog post.