Most Malaysian business owners fill in a responsive search ad the same way they fill in a form. Fifteen boxes appear, so they write fifteen things. Half say the same idea in different words, and the ad still runs.
It runs, but it does not sell. Because Google reports the ad as one unit, the owner never sees that three headlines are carrying the account while the other twelve take up slots.
This guide is about the writing, not the setup: what each headline should do, when pinning helps, how many assets are enough, and how long to wait before judging any of it. Four datasets from ZenWeb-managed Malaysian accounts sit behind it.
First, Google’s own tutorial covers how the format assembles combinations.
Source video: Google Ads Tutorials: Responsive search ads best practices on the Google Ads YouTube channel
Quick Answer: Responsive search ads let you supply up to 15 headlines and 4 descriptions. Google assembles them into different combinations per search, then learns which combinations perform. You are not writing one ad. You are writing a pool of parts that must make sense in any order.
Google’s responsive search ads documentation sets the boundaries. Headlines cap at 30 characters, descriptions at 90, display paths at 15, and you need at least 3 headlines and 2 descriptions to save the ad. Most ads then show one to three headlines at a time.
That last point trips people up. Three consequences follow:
Treat the ad as a parts bin where every part is independently sellable. That is also why it works alongside your sitelinks, callouts and other ad assets rather than replacing them: the ad body carries the offer, the assets carry the detail.
Not sure whether your ads or your keywords are the problem?
We read asset-level performance before touching bids on every account we take over. See how our Google Ads management works →
Quick Answer: Across ZenWeb-managed Malaysian accounts, performance improves sharply from 3 headlines up to about 10, then flattens. Filling all 15 with thin variations does not beat 10 strong ones. On small budgets it is often worse, because the extra assets never gather enough data to be judged.
The standard advice is to fill every box. It hides the reason it works: the benefit comes from distinct ideas, not a full form. Reworded duplicates add slots without adding coverage.
| Distinct headlines supplied | Conversion rate index (3–5 = 100) | Median CPL index | Assets with usable data at 30 days |
|---|---|---|---|
| 3–5 | 100 | 100 | 92% |
| 6–8 | 121 | 88 | 80% |
| 9–11 | 138 | 81 | 64% |
| 12–14 | 141 | 80 | 51% |
| 15 (all distinct) | 143 | 79 | 44% |
| 15 (padded with duplicates) | 112 | 95 | 39% |
Source: ZenWeb client sample, 500+ Malaysian SME accounts, 2024–2026. Licence.
The last row is the important one. Accounts that reached 15 by rewording what they had performed worse than accounts that stopped at 8 honest ideas. The final column explains why: on a fixed budget, the more assets you add, the smaller the share that collects enough impressions to be judged. Our guide to writing Google Ads copy that converts covers how to find different angles instead of synonyms.
Quick Answer: Give every headline one job and do not repeat a job more than twice. Six jobs cover almost every Malaysian SME ad: match the search, name the offer, prove you can do it, remove the risk, state the price signal, and tell them what happens next.
Most RSA advice lists headline “types” and stops there. Jobs is the more useful framing, because a job can only be filled so many times before the next one is wasted. Six jobs, roughly two headlines each:
Two headlines can carry your brand name. That matters more than owners expect, and our comparison of branded versus non-branded keywords explains why the two search types behave differently. Beyond two, brand repetition eats slots that could be doing real work.
Quick Answer: Pin when a message legally or commercially must appear every time. Pin two or three candidates to that position rather than one, so Google still has a choice. Pinning every position turns a responsive ad back into a fixed ad and gives up the format’s only advantage.
The common line online is that pinning is bad because it lowers Ad Strength. That is a half-truth. Google’s own guidance says pinning is not recommended for most advertisers and can affect Ad Strength. The same page also says that if text must appear in every ad, it has to be pinned to Headline 1, Headline 2, or Description 1.
For Malaysian advertisers that “must” is common. Halal certification, licence numbers, panel clinic status and promotion terms belong in every impression, not in 60% of them.
| Pinning approach | Typical Ad Strength | Median CTR index | Median CPL index |
|---|---|---|---|
| No pinning | Good to Excellent | 100 | 100 |
| One headline pinned to position 1 | Average to Good | 104 | 96 |
| Two to three pinned to position 1 | Good | 107 | 92 |
| Every position pinned | Poor | 88 | 118 |
Source: ZenWeb client sample, 500+ Malaysian SME accounts, 2024–2026.
Pinning two or three candidates to one slot is the setting almost nobody uses and it is the one that holds up. You keep control of the message while Google keeps a decision to make. The same logic applies to match types: control helps until it starves the system of choices, which our comparison of broad match against exact match works through in detail.
Quick Answer: Ad Strength measures how varied your inputs are, not how well your ad sells. Google reports real gains when advertisers move from Poor to Excellent, but the rating cannot see your conversion rate. Use it as a completeness check, then judge the ad on cost per lead.
Google is specific about the upside. Advertisers who improve Ad Strength from “Poor” to “Excellent” see 15% more clicks and conversions on average, per Google internal data from August 2025. The same page reports a 6.6% conversion lift from adding a second RSA to an ad group, and 3.7% from a third.
Those numbers are worth acting on. The trap is treating the rating as the goal:
The metric that rewards better writing is Quality Score, and its ad relevance component, because it is measured against real search behaviour. If you lean on Google’s automated suggestions to fill gaps, our review of whether AI-generated Google Ads assets convert is worth reading first.
Ads rated Excellent but still not selling?
That gap is usually the offer or the page, not the rating. See why good ads still produce no leads →
Quick Answer: Keyword-match headlines win the most impressions but convert around average. Price-signal and risk-removal headlines win fewer impressions and convert far above average. Generic brand and slogan headlines take slots and return the least.
Since Google began reporting clicks, impressions and conversions at asset level, you can see which of the six jobs is carrying the ad. Google is careful to call asset-level ratios directional, since assets always serve in combination, so read the figures below as a pattern across many accounts rather than a verdict on any single headline.
| Headline job | Share of served impressions | Conversion rate index (all assets = 100) | Verdict |
|---|---|---|---|
| Match the search | 31% | 103 | Essential, but not the seller |
| Remove the risk | 12% | 148 | Most under-used slot |
| Signal the price | 9% | 139 | Filters out wrong clicks |
| Prove you can do it | 14% | 121 | Strong with specific numbers |
| Say what happens next | 16% | 112 | Best when it names WhatsApp |
| Brand name only | 11% | 86 | Cap at two |
| Slogan or generic praise | 7% | 71 | Cut it |
Source: ZenWeb client sample, 500+ Malaysian SME accounts, 2024–2026. Licence.
Read the first two columns together. Risk-removal headlines take only 12% of impressions but convert nearly half again above average, so most accounts under-supply the job that sells hardest. Check your search terms report before rewriting: if the queries reaching your ad are exploratory, risk-removal matters more still. Longer sales cycles shift the ratio again, as our guide to running search ads for B2B covers.
Quick Answer: Give a new responsive search ad four to six weeks before you rewrite it. On typical Malaysian SME budgets, fewer than half the assets have enough impressions to judge at two weeks. Editing early resets learning and hides which headline was working.
Impatience is the most expensive habit in a small account. Every edit to headlines or descriptions restarts the assembly learning, so an owner who tweaks weekly never lets a version finish proving itself.
| Week after launch | Assets with judgeable data | Conversion rate index (week 12 = 100) | Safe action |
|---|---|---|---|
| Week 1 | 9% | 64 | Check for disapprovals only |
| Week 2 | 24% | 78 | Add negatives, leave copy alone |
| Week 4 | 51% | 91 | First honest read of asset data |
| Week 6 | 72% | 97 | Replace the clear losers |
| Week 8 | 84% | 99 | Launch a second RSA to test against |
| Week 12 | 93% | 100 | Full rewrite if still underperforming |
Source: ZenWeb client sample, 500+ Malaysian SME accounts, 2024–2026. Budget-normalised to RM 3,000 monthly search spend. Licence.
Smaller budgets stretch this timeline, not shorten it. At RM 1,000 a month, week 4 looks like week 2 here. If you want the account watched without watching it yourself, Google Ads scripts can flag disapprovals and spend anomalies daily while the copy is left to settle.
Quick Answer: Write for a bilingual searcher who wants a price and a WhatsApp number. Mix English and Bahasa Malaysia terms the way your customers type them, name the area you serve, and put the contact method in the ad rather than leaving it as a surprise.
Copy translated straight from American PPC templates underperforms here, for reasons that have nothing to do with language quality:
Two platform notes follow. If your service is dispatched to a customer’s address, Google Local Services Ads may sit above your search ads and change what the RSA needs to say. If you are weighing a second search channel, our look at whether Microsoft Advertising is worth it in Malaysia covers how far the same copy travels.
Quick Answer: A responsive search ad can only sell the click. If the landing page does not repeat the exact promise the headline made, the visitor re-reads, doubts, and leaves. The ad and the page are one message split across two screens.
This is where most RSA rewrites quietly fail. The ad improves, click-through rises, cost per lead does not move, and the owner concludes the copy did not work.
The mismatch is usually specific. If the headline promised “Free Site Inspection” and the page opens with company history, the promise broke before the visitor scrolled. Three checks settle it:
Our breakdown of the seven landing page fixes that lift Google Ads leads covers the page side in order of impact. Conversion rate optimisation services explain what structured testing looks like once the obvious fixes are done.
Quick Answer: Write 10 to 12 headlines that each do a different job, pin only what compliance demands, get Ad Strength to Good and stop, then leave the ad alone for four weeks. Judge it on cost per lead, and fix the landing page before you rewrite the copy a second time.
Responsive search ads reward clear thinking more than clever writing. The format punishes vagueness, because vague copy produces interchangeable headlines, and interchangeable headlines give Google nothing to choose between.
Three habits separate accounts that improve from accounts that churn. Assign every headline a job before saving. Read asset-level data instead of ad-level totals. Give each version four quiet weeks before deciding anything.
To hand the work over instead, it sits with a SEM specialist, and our guide to choosing a paid search agency that grows revenue sets out what to ask. Where search ads and SEO run together, our view on a search marketing agency under one roof explains the trade-offs. See the full range of what we do at ZenWeb.
Up to 15 headlines and 4 descriptions, with at least 3 headlines and 2 descriptions needed to save the ad. Headlines cap at 30 characters, descriptions at 90. In practice, 10 to 12 genuinely different headlines outperform 15 that repeat each other.
Pinning one asset to a position cuts the combinations Google can test and usually lowers Ad Strength. Pinning two or three assets to the same position keeps the message controlled while leaving Google a choice, and in our client data that setup beats both no pinning and full pinning.
Aim for Good, then stop. Ad Strength scores how varied your inputs are, not whether people buy. Google reports 15% more clicks and conversions when advertisers move from Poor to Excellent, but padding with filler to reach Excellent can lower conversion rate on small budgets.
Four weeks on a typical Malaysian SME budget, longer if you spend under RM 1,500 a month. Before week 4, roughly half the assets have too few impressions to judge. Editing headlines restarts learning, so frequent tweaks keep the ad in its weakest phase.
Almost always a mismatch between the ad and the landing page. The headline earning the clicks makes a promise the page does not repeat: a different offer, a different price, or a form where the ad implied WhatsApp. Fix the page before rewriting the ad again.
Ready to find out which headlines are actually selling?
Book a free 30-minute strategy session — we’ll review your asset-level performance, your landing pages, and your competitors, then give you a concrete 90-day plan with realistic CPL and pipeline targets.
Complete the form and our team will contact you to discuss your goals. Let’s grow your business.
Online
