Almost every guide to Dynamic Search Ads is a setup tutorial. Open a campaign, choose a dynamic ad group, target the whole site, write two descriptions, done. The advice is accurate and, in July 2026, close to useless on its own.
Google confirmed in April that DSA is being folded into AI Max for Search. Automatic upgrades start in September, and new DSA campaigns cannot be created after that. Learning the setup screen has weeks of shelf life left.
What does not expire is what DSA was always really testing: whether Google, crawling your site cold, can find a page worth sending a buyer to. That question matters more under AI Max, not less.
This piece covers what dynamic search ads do, how they perform beside keyword ad groups, which pages Google actually picks, which controls cut waste, and what to do before the switch. Four datasets from ZenWeb-managed accounts sit underneath it. The video below walks through the setup itself.
Source video: How to Create Dynamic Search Ads in 6 Simple Steps [+ Examples] on YouTube
Quick Answer: Dynamic Search Ads use Google’s crawl of your website instead of a keyword list. Google reads your pages, matches a live search to the page it judges closest, writes the headline itself and sends the click there. You supply only the description lines and the bid.
Three jobs move from you to Google: which searches you appear for, which page receives the click, and the headline itself. You keep the description, the bid, the budget and the exclusions — less control than most advertisers expect.
The trade is coverage. On a plumbing site with 40 service pages, DSA surfaces phrasing your bid list never had: part brand names, misspelled suburbs, oddly specific problems.
Two mechanics matter more than the rest:
The headline generation is what people forget. Google writes it from the page it picked, so a thin page produces a weak ad — which makes DSA performance a direct readout of page quality. If your ad copy discipline lives in your responsive search ads, DSA quietly bypasses it.
Google’s April 2026 announcement confirmed DSA campaigns upgrade automatically to AI Max for Search from September 2026, with new DSA creation ending at the same point. Anything you build now is a migration, not a foundation.
Not sure which pages Google is sending your clicks to?
We audit the landing page and query mix on Malaysian Google Ads accounts before the September migration. See how our Google Ads management works →
Quick Answer: On Malaysian accounts running both, dynamic ad groups buy cheaper clicks but more expensive enquiries. Median CPC sits around 29% below keyword ad groups, while cost per enquiry runs roughly 19% higher. DSA earns its place as a discovery channel, not as a cheaper version of search.
The cheap-click story is real. The enquiry story shows up 90 days later. Both are in the table below.
| Measure | Keyword ad groups | Dynamic ad groups |
|---|---|---|
| Median CPC (RM) | 3.90 | 2.75 |
| Conversion rate | 4.6% | 3.1% |
| Median cost per enquiry (RM) | 118 | 141 |
| Spend on queries with no enquiry in 90 days | 21% | 38% |
| New converting queries found per quarter | 9 | 24 |
Source: ZenWeb client sample, 500+ Malaysian SME accounts, 2024–2026. Licence.
The last row is the one worth budgeting for. Dynamic ad groups surfaced 24 new converting queries a quarter against 9 from keyword ad groups. Those queries then get promoted into keyword ad groups, where they convert at the higher rate — which makes the 38% wasted-spend figure closer to a research cost than a leak.
Dynamic ad groups found roughly three times more new converting queries per quarter than keyword ad groups — and wasted nearly twice the spend doing it.
For context, WordStream’s 2026 Google Ads benchmarks put the average conversion rate across accounts at 8.18%. Malaysian lead-generation accounts sit below that, so compare the two columns above with each other, not with global figures.
Quick Answer: Google sends about a third of dynamic clicks to blog posts on Malaysian sites, and those clicks produce only 7% of enquiries. Service pages take 34% of clicks and return 58% of enquiries. The mismatch, not the bid, is where most DSA budget disappears.
This is the table nobody runs before switching DSA on. It splits every dynamic click by page type, then follows the enquiry.
| Page type Google picked | Share of clicks | Share of enquiries | Cost per enquiry (RM) |
|---|---|---|---|
| Service pages | 34% | 58% | 96 |
| Location or area pages | 12% | 19% | 108 |
| Pricing pages | 6% | 11% | 84 |
| Blog posts | 31% | 7% | 402 |
| Homepage | 13% | 4% | 355 |
| Policy, careers, other | 4% | 1% | 610 |
Source: ZenWeb client tracking, Malaysian SME accounts, 2024–2026. Licence.
Blog posts take almost as many clicks as service pages and return an eighth of the enquiries. That is not Google behaving badly. It is Google reading a site where the blog is deep and heavily interlinked while the service pages are thin.
Sites with strong content programmes leak the most dynamic budget into articles. Sites with no blog rarely have the problem at all — they just get less coverage. The fix is a targeting decision, covered in the next data section. If your service pages are the weak link, that is a separate job: why great ads still get you no leads covers the page side, and seven landing page fixes covers the quick wins.
Quick Answer: DSA fails locally for structural reasons rather than bidding ones. Bilingual pages, WhatsApp-only enquiry paths, single-page service sites and JavaScript-rendered content all give Google either too little to crawl or the wrong thing to crawl. The account looks broken when the website is the problem.
Most troubleshooting guides send you to bids and negatives. On Malaysian SME sites the failures cluster somewhere else entirely.
There is a useful side effect. Because Google chooses the page, a DSA campaign is a blunt audit of how your site reads to a machine. If dynamic ad groups keep landing on your blog, an AI answer engine is likely reaching the same conclusion about where your expertise lives.
That is the argument for running a short dynamic campaign even if you never keep it. Ten days of data tells you which pages Google considers your real answers. Compare that against the pages you want buyers to see; the gap is your site’s structural problem, written down.
Quick Answer: Negative dynamic targets on blog, careers and policy URLs remove the most waste — a median 41% reduction when applied alone. Page-feed targeting and negative keyword lists follow. Bid adjustments barely move the number, which is why most DSA tuning advice underdelivers.
Each lever was measured on its own, on accounts where it had not been applied before. The reductions are not additive, but the ranking holds.
| Control | Waste removed | Median | Days to effect |
|---|---|---|---|
| Negative dynamic targets on blog, careers, policy URLs | 41% | 4 | |
| Shared negative keyword list from search terms | 27% | 3 | |
| Page feed instead of whole-site targeting | 22% | 7 | |
| Excluding the homepage as a target | 9% | 5 | |
| Bid adjustments on dynamic targets | 6% | 14 |
Source: ZenWeb operational data, Malaysian SME campaigns, 2024–2026. Licence.
The order matters. Exclusions first, because they change what the later data means.
Step five repays the whole exercise. Broad match against exact match covers where harvested terms should sit, and branded versus non-branded bidding covers whether to keep your own brand in the mix.
Want this done before the September upgrade?
Our team runs the exclusion and harvest pass on Malaysian accounts as part of ongoing management. Compare what an SEM agency should handle →
Quick Answer: Only as a diagnostic. Running a tightly-excluded dynamic campaign for two or three weeks tells you which pages Google favours and which queries you are missing. Building a long-term DSA structure now is wasted work, because Google upgrades it to AI Max automatically from September 2026.
Three situations, three answers.
The habit worth carrying forward is the harvest loop, because AI Max expands queries in a similar way. Google reports that AI Max campaigns using the full feature set see 7% more conversions or conversion value at a similar cost per acquisition, compared with search term matching alone. That is a real gain, and it still depends on your pages being worth landing on.
Watch your search impression share and ad rank through the transition. Both shift when matching widens, and they are the earliest signal that you are entering different auctions. If you also run Microsoft Advertising, its dynamic ads are unaffected by Google’s timeline — a useful control group.
Quick Answer: Dynamic ad groups peaked at roughly a fifth of Malaysian search spend in 2024 and fall to zero during the September 2026 upgrade. Automated expansion absorbs that share and keeps growing, while keyword ad groups shrink but remain the largest single block through 2027.
The table below models the shift across managed Malaysian accounts. The 2027 column is a projection, not a measurement.
| Structure | 2022 | 2023 | 2024 | 2025 | 2026 | 2027* |
|---|---|---|---|---|---|---|
| Keyword ad groups | 86% | 82% | 77% | 70% | 63% | 55% |
| Dynamic ad groups | 14% | 17% | 19% | 18% | 12% | 0% |
| Automated expansion | 0% | 1% | 4% | 12% | 25% | 45% |
* Modelled projection based on Google’s April 2026 upgrade timeline and ZenWeb account mix, 2022–2026. Licence.
The direction is not in doubt, only the pace. Keyword ad groups do not disappear, and the accounts likely to hold their cost per enquiry through 2027 are the ones keeping a tight keyword core while automation expands around it. That is also how the strongest SEM shortlists get judged, and the structure most white label search teams build for partners.
Quick Answer: Dynamic search ads are worth running for what they reveal, not for what they replace. Exclude the pages you do not sell from, harvest the queries that convert, and carry that discipline into AI Max when the upgrade lands in September 2026.
The advertisers who did well from DSA never treated it as a set-and-forget traffic source. They treated it as a query discovery engine with a page-quality readout attached, and moved every winner into a properly written ad group.
That approach transfers cleanly. AI Max widens matching further and hands back a different set of controls, but the job is unchanged: make sure the pages Google can reach are the ones that answer a buyer’s question, then measure. Your ad assets and your thank you page matter more once you no longer control the headline.
Ready to get your Google Ads account through the AI Max switch?
Book a free 30-minute strategy session — we’ll review your landing pages, your dynamic targets and your query mix, then give you a concrete 90-day plan with realistic cost per lead and pipeline targets.
Yes, in their current form. Google announced in April 2026 that campaigns using dynamic search ads will automatically upgrade to AI Max for Search from September 2026, and that new DSA campaigns can no longer be created in Google Ads, Google Ads Editor or the API after that point. Existing settings and historical data are carried across during the upgrade.
No. Dynamic ad groups match searches using Google’s crawl of your website rather than a keyword list. You still need negative keywords, though. Without a shared negative list, dynamic campaigns pick up job applications, DIY queries and competitor brand searches that no keyword campaign would ever bid on.
Yes, through targets and negative dynamic targets. You can restrict the pool to specific URL patterns, upload a page feed listing only the URLs you want, or exclude sections such as your blog and careers pages. Whole-site targeting is the default and is the main cause of budget landing on non-commercial pages.
They suit businesses with many distinct service or product pages and clear enquiry tracking. A single-page website gives Google only one URL to work with, so every click lands on the homepage. If your site is one long page or your only call to action is an untracked WhatsApp button, fix that before running dynamic campaigns.
Upgrading early keeps the decisions in your hands. Google’s guidance is that voluntary upgrades let you port settings and data on your own terms, while automatic upgrades mirror your legacy setup with all AI Max features enabled. Either way, clean your exclusions and harvest your converting queries first so the migration starts from a tidy account.
Complete the form and our team will contact you to discuss your goals. Let’s grow your business.
Online
