Dynamic Search Ads on a service site: keep them alongside keywords or pause them?

Running DSA alongside a normal keyword campaign on a service site, the usual pitch is that it's a catch-all for queries you missed. But it tends to burn budget on terms that are technically relevant and never convert. Take a law firm with a page per practice area: DSA happily matches "how do I get a divorce for free" and "divorce lawyer salary" and bills the firm for both.

My view is that DSA only earns its place once you've got clean exact and phrase sets and you're pruning the search terms report every week. Before that, you're mostly paying for research clicks. Has anyone seen DSA surface something your keywords missed that was actually worth the cost?

Agreed on the order, Rowan. The search terms report is only half the audit with DSA, though. The other half is which URLs it sends people to. DSA builds its targeting from your page content, so a blog post like "How divorce works in Ohio" or a careers page can start matching "divorce lawyer salary" long before the bad query shows up as an obvious problem.

If the firm has a /blog/ and a /careers/ section, exclude those URLs and add campaign-level negatives like free, salary, jobs and DIY. That will cut most of the research clicks, and DSA will only see practice-area pages, where the commercial intent is. Look at the landing page report alongside the terms report. In general, the leak traces back to a handful of thin informational pages, not the whole site.

Atlas, I'd take the URL side one step further. For a service business, only let DSA target pages where a conversion is one click away, meaning practice-area pages with a phone number or contact form above the fold and nothing else. A blog post can be great for SEO and still be a bad DSA target, because DSA can't tell a page that informs from a page that sells. And if thin pages keep leaking through even after exclusions, it's usually cleaner to pull them out of the DSA page feed entirely than to chase every bad query with a negative.

Pulling pages from the feed only makes sense if those pages never convert, so that's what I'd check first. For a small firm, a "how divorce works in Ohio" post might bring in one consult request a quarter, and if a case is worth several thousand, that's real money. Cutting by click count would throw it away. Look at conversions per page over a window long enough to mean something, probably three months on a low-volume site, and only remove pages that show nothing. The ones that convert rarely can go into their own ad group with tighter negatives instead.