Paid Search Ad Copy Best Practices: A Writing Process, Templates, Testing, and Metrics

Quick Answer: The best paid search ad copy comes from a repeatable process, not inspiration. Start with the conversion you want and the intent behind the search, build a small set of message angles (benefit, offer, proof, objection, action), turn them into a bank of headlines and descriptions that fit Google's limits, decide carefully what to pin, and match the copy to the landing page. Then test message angles instead of single words, judge each test on conversions per impression and cost per conversion (not CTR alone), and check that the traffic behind the result is real. Since June 30, 2022, you can no longer create or edit expanded text ads; responsive search ads (RSAs) are the standard format for new text ads in Google Ads Search campaigns, so you write assets that Google combines, not one fixed ad. This guide gives you the workflow, fill-in templates, a testing method, and the metrics to evaluate results.

This guide focuses on process, templates, experiments, and metrics. It does not repeat the example teardowns in our PPC examples guide.

What is paid search ad copy?

Paid search ad copy is the text of ads that appear when people search on Google and other search engines. In Google Ads, you provide up to 15 headlines (up to 30 characters each) and up to 4 descriptions (up to 90 characters each), and the system assembles combinations that it expects to fit each search. The pieces you write are called assets.

That changes the job of a copywriter in three ways:

1. You write modular lines, not one ad. Every headline must make sense next to many others.
2. You test angles, not exact wording. Google decides combinations, so the useful question is "which message theme performs?"
3. You have less control over what shows together. Pinning gives control but limits testing.

What is the best process for writing search ad copy?

Use these eight steps for every ad group or theme.

Step 1: Define the outcome and the conversion

Write down the single action you want (call, quote request, purchase, trial) and how you will measure it. Copy for a phone call differs from copy for a form fill, and a vague conversion produces vague copy. Decide the target cost per conversion before you write, so you know what "good" means.

Step 2: Read the search intent

Look at the search terms report and group queries by intent:

• Problem-aware: "why is my [thing] doing [problem]"
• Solution-seeking: "[service] near me," "best [product]"
• Comparison: "[brand] vs [brand]," "[product] alternatives"
• Ready to act: "buy [product]," "[service] quote," "book [service]"
• Navigational or branded: a company or product name

Write separate copy for different intents. An ad for a ready-to-buy searcher should not read like an ad for a researcher.

Step 3: Collect the customer's language

Use real words from reviews, support tickets, sales calls, call recordings, form messages, and on-site search. The phrases customers use for their problem often outperform internal jargon. Also check what competitors say in their ads and where you can honestly differ (Google's Ads Transparency Center and your own searches on a clean device help).

Step 4: Build message angles

An angle is a reason to click. Common ones:

Choose three to five angles for the ad group. Select no more than 15 headlines from the asset bank below; the suggested ranges are alternatives, not a requirement to add them all.

Step 5: Draft the asset bank

Fill the headline slots by purpose, so each one has a job:

For the four descriptions, give each a different role:

1. Benefit plus action.
2. Proof plus offer.
3. Objection plus reassurance.
4. Detail or availability (service area, hours, shipping, terms).

Step 6: Edit

• Check limits: 30 characters per headline, 90 per description.
• Make each headline stand alone and read correctly next to any other.
• Remove near-duplicates. Several headlines that say the same thing waste slots and restrict variety.
• Check for repeated claims (for example, "free" in every line).
• Cut filler such as "click here" and "best in town."
• Verify every claim against reality and policy.

Step 7: Decide what to pin

Pin only what must appear in a fixed position, such as a legal disclaimer, a required brand name, or a key promotion. Leave the rest flexible (see the pinning section below).

Step 8: Match the landing page and run a pre-launch check

The page should repeat the main promise from the ad in its headline and make the action obvious on mobile. Then run the checklist at the end of this guide before launch.

A worked example: from angles to assets

This is an illustrative template (a made-up home energy audit service). It shows the process, not a real campaign. Replace the details with true statements about your business.

Intent: ready to act ("home energy audit near me") Conversion: booked audit Angles chosen: relevance, benefit, convenience, proof, action

Notice that no headline repeats another's job without adding something new, and each can sit beside any other. (Character counts for all lines above were checked against the limits.)

What are the best templates for headlines and descriptions?

Use these as starting formulas. Fill the brackets with specific, true details, and keep each within the character limits.

Headline templates

Description templates

Sitelink and callout templates

• Sitelinks: pair a short link title with two short description lines that point to specific pages (pricing, reviews, service areas, a product category, a booking page).
• Callouts: short proof or perk phrases, such as a guarantee, a delivery option, or hours.
• Call and location assets: include them where phone calls or visits matter, and make sure the phone number and address match the landing page.

Style rules that usually help

• Use the searcher's words.
• Start with what matters most.
• Prefer specific numbers and facts to adjectives.
• Write in plain, active language.
• Keep one idea per line.
• Sentence-case text is easy to read, but test it for your market instead of treating it as a rule.

What do ad policies require?

Check Google Ads policies for your industry, because rules change and some categories are restricted. As a general guide:

• Claims need support. Superlatives such as "best" or "#1" and statements about results need evidence you can show.
• Avoid gimmicks. Excessive capitalization, repeated punctuation, and symbols used in unusual ways can be disapproved.
• Keep the ad and page consistent. Offers, prices, and terms in the ad must match the landing page.
• Be careful with trademarks, competitor names, and regulated or sensitive topics.
• Do not promise what you cannot deliver, such as "free" with hidden conditions.

How should you use pinning and ad strength?

Pinning locks an asset into a position. Whether pinned text always appears depends on the position and format. For required wording, prefer Headline 1 or Description 1 and review Google's current format rules. Pinning reduces the combinations Google can test.

Practical guidance:

• Pin as few lines as possible. Pinning several assets to the same position gives Google more flexibility than pinning one line.
• Lower positions (such as Headline 3 or Description 2) are not guaranteed to show, so critical text should not depend on them.
• Review position and combination reporting to see the effect of your pins.

Ad strength is a setup indicator of how varied and relevant your assets are. Google has said that stronger ad strength is associated with more clicks and conversions on average, but independent analyses have found little or no link to actual cost per conversion in specific accounts. Use it as a coverage checklist (enough unique assets, relevant keywords), not as a performance target. Do not add weak headlines just to turn the indicator green.

How should you test search ad copy?

What to test

Test message angles, not minor wording. In rough order of impact:

1. Offer (free trial vs discount vs free shipping)
2. Primary benefit (save money vs save time vs reduce risk)
3. Proof (credential vs rating vs guarantee)
4. Audience call-out (who it is for)
5. Call to action (get a quote vs book now vs compare)
6. Landing page (if copy and page are tested together, say so)
7. Smaller items such as punctuation and word order, only when you have lots of data

Design the test

1. Write a hypothesis. "If we lead with [angle B] instead of [angle A], then [metric] will [increase], because [reason about the searcher]."
2. Choose the unit. An ad group or a set of ad groups with the same intent.
3. Change one thing. Use a controlled Google Ads experiment where supported, with one message angle in each arm and comparable asset quality and counts. Two RSAs in one ad group do not guarantee random or equal exposure and are insufficient on their own for a causal A/B test. Keep bids, budget, keywords, and landing pages the same during the test.
4. Do not stop early. Give the system time to learn, and avoid pausing a new ad after a few days.
5. Set the sample size in advance. A common approximation for a conversion-rate test at roughly 95% confidence and 80% power is n per variant ≈ 16 × p × (1 − p) ÷ δ², where p is the baseline conversion rate and δ is the smallest absolute improvement you care about. For example, if your baseline is 5% and you want to detect a lift to 6% (δ = 0.01), you need about 7,600 clicks per variant.
6. Account for time. Run for whole weeks, and avoid periods with unusual events or promotions unless you are testing them.
7. Use Google Ads experiments for larger changes, such as landing pages or bidding strategies, where you want a clean split of traffic. Be aware that RSA combination testing happens inside each ad, so an experiment is not the same as a pure A/B copy test.

What if you do not have enough volume?

• Test bigger differences in angle, since small differences need large samples.
• Use a higher-volume proxy metric (such as clicks to a pricing page) only as a secondary signal, and keep the business conversion as the primary.
• Combine similar ad groups to a theme-level test.
• Use before-and-after comparisons carefully, because seasonality and competitors can mimic a copy effect.
• Accept that a test may stay inconclusive, and keep the angle that is clearer, cheaper, and more compliant.

How to read the results

Use asset-level and combination reports to see which assets get impressions and how they perform, but remember:

• Asset reporting needs enough volume; use the metrics currently available in your account.
• An asset may show less simply because it is paired with strong or weak partners.
• Compare themes (for example, all offer headlines against all proof headlines) instead of single lines.

Which metrics should you use to evaluate ad copy?

Copy changes both the click rate and the quality of the click, so judge it with a metric that combines them.

Primary metrics (decide the winner)

Why CTR alone misleads: a more attractive headline can raise CTR and still lose.

Ad B gets fewer clicks but more customers per impression. If the cost per click is also similar, Ad B wins on cost per conversion. (The numbers are illustrative.)

Diagnostic metrics (explain why)

Guardrail metrics (stop bad "wins")

How can bots and bad traffic distort copy tests?

A headline that happens to appeal to bots, accidental taps, or low-quality sources can look like a winner on CTR. Fake traffic can also dilute true differences or land more in one variant. Before you declare a winner:

1. Compare clicks with sessions and engagement by variant.
2. Break results down by device, location, hour, and source, and look for spikes or odd patterns that affect only one variant.
3. Review search terms and placements for that variant.
4. Look at lead quality, not just lead count.
5. Check invalid click reports where available, and your own click-level data.
6. Apply any traffic filters to all variants and define them before you look at results.

If a "winning" ad has a high CTR but weak engagement, odd geographies, or no real customers, do not scale it.

How often should you refresh and document your copy?

1. Keep a test log. For each test, record the hypothesis, the angles, dates, budget, sample size, metrics, result, and decision.
2. Review monthly for high-volume ad groups and quarterly for the rest.
3. Retire assets that rarely show or consistently perform poorly after they have had enough impressions, and replace them with new angles.
4. Refresh for seasonality, offers, and competitors, and remove expired promotions promptly.
5. Reuse learnings. If a proof angle wins in one ad group, test it in similar groups.

Test log template

Common mistakes in paid search copywriting

• Writing one ad and treating it as finished.
• Filling slots with near-duplicate headlines.
• Over-pinning, which removes the benefit of testing.
• Chasing the ad strength indicator instead of performance.
• Testing tiny word changes before testing offers and angles.
• Judging winners by CTR alone.
• Declaring results too early or without enough conversions.
• Changing bids, budgets, and copy at the same time.
• Ignoring the landing page, so the promise breaks after the click.
• Making claims that policy or reality cannot support.
• Not checking traffic quality before scaling a winner.
• Failing to log tests, so lessons are lost.

Paid search ad copy checklist

• Conversion and target cost per conversion defined
• Search intent grouped and copy written per intent
• Customer language collected from reviews, calls, and sales notes
• Three to five message angles chosen
• 15 distinct headlines and 4 distinct descriptions drafted, each with a job
• Character limits checked (30 for headlines, 90 for descriptions)
• Every claim verified and policy-checked
• Pinning limited to required text only
• Landing page matches the promise and works on mobile
• At least two RSAs per ad group with different angles
• Test hypothesis, sample size, and primary metric set in advance
• Winners judged on conversions per impression and cost per conversion
• Lead quality and invalid traffic checked before scaling
• Test log updated, and refresh schedule set

FAQ

What are paid search ad copy best practices?

Match the searcher's intent, state a specific benefit or offer, add proof, give one clear action, send people to a matching landing page, and test message angles using conversions and cost per conversion instead of CTR alone.

How many headlines and descriptions should a responsive search ad have?

Google allows up to 15 headlines (30 characters each) and up to 4 descriptions (90 characters each). Filling the slots with distinct, useful lines gives the system more combinations to test, but quality matters more than count.

Are responsive search ads the only option?

Responsive search ads are the standard Search text ad format. Since June 30, 2022, expanded text ads can no longer be created or edited; this does not mean all other Search ad formats are RSAs.

Should I pin headlines?

Only when specific text must appear, such as legal wording or a required brand line. Pin as few assets as possible, because pinning reduces the combinations Google can test.

Is a high ad strength rating important?

It is a helpful setup check for asset variety, not a performance guarantee. Aim for good coverage, but judge ads by conversions, cost, and quality.

How long should I run an ad copy test?

Until you reach the sample size you planned, and long enough to cover full weekly cycles. Avoid judging a test after a few days or a handful of conversions.

How much data do I need to call a winner?

It depends on your baseline conversion rate and the size of the difference you want to detect. For an independent two-arm conversion-rate test, a rough approximation for clicks per arm is n ≈ 16 × p × (1 − p) ÷ δ², with p and δ expressed as proportions. This is not a conversion count or a substitute for an experiment-specific power calculation. Smaller differences need far more data.

What should I test first?

Offers and main benefits, then proof and calls to action. Test small wording changes only when you have high volume.

Which metric is best for judging copy?

Conversions per impression (CTR × conversion rate) and cost per conversion, with lead quality and invalid traffic as guardrails. CTR alone can reward clicks that never become customers.

How do I write headlines that fit together?

Write each headline so it makes sense on its own and next to any other. Give each a distinct job (relevance, benefit, offer, proof, action), and avoid repeating the same claim.

Can bots make an ad copy test look better than it is?

Yes. Invalid clicks and accidental clicks can inflate CTR and skew conversion data for one variant. Compare clicks with engagement and sessions, segment results, and check invalid traffic before you scale a winner.

How often should I update ad copy?

Review high-volume ad groups monthly and others quarterly. Update sooner when offers, seasons, competitors, or policies change, and remove expired promotions right away.

Next Steps

Strong paid search copy is built, not guessed: define the conversion, read the intent, choose a few message angles, write distinct assets that each have a job, pin only what you must, and match the landing page. Then test angles with enough data, judge winners on conversions per impression and cost per conversion, and confirm that the traffic and leads are real before scaling. Keep a log so every test makes the next one better.

Related Articles

• PPC Advertising Examples: Google Ads Copy, Landing Pages, and Results (Teardown)
• High Intent Keywords: How to Find Keywords That Convert
• What Is CTR in Ads? Why a High CTR Doesn't Always Mean More Conversions

Investigate Your Paid Traffic

JuicyTraffic helps advertisers investigate landing-page click quality by campaign, source and placement. Compare per-click evidence with platform reports and qualified leads or sales before scaling.

References

• Google Ads responsive search ads