Paid ads

Google Ads headline generator for ecommerce and retail

By Charles Summers · Updated · Free, no signup

Short answer

On a retail query your text ad is rarely the first thing anyone sees. Shopping units and Performance Max product tiles sit above it rendering image, price, merchant name and review stars straight from your feed, so a headline repeating the product name and the price is buying a second impression of information the grid already delivered. The 30-character slots earn their keep on what a feed cannot render: return window, delivery cut-off, size or fit range, stock depth, and the reason to buy from you rather than the four merchants beside you. Anything about a promotion belongs in a scheduled asset, because typed headline text has no end date.

Use the google ads headline generator

Why ecommerce need a different approach

Retail is the one vertical where the ad you wrote is competing with a machine-generated version of itself. The product feed is already on the results page, styled as a grid, carrying the photograph and the price, and it updates itself when either changes. Your responsive search ad sits underneath that, in text, in a container the width of a phone. Writing it as though the grid were not there produces an ad that agrees with the tiles above it and adds nothing to the decision.

The second thing retail does differently is time. A promotion typed into a headline is a permanent object: nobody scheduled it, nothing expires it, and it will still be serving in February unless a person remembers to open that ad group and delete it. Multiply that by however many ad groups a catalogue generates and the January clean-up becomes the single most common source of stale offers, disapprovals and clicks paid for on a discount the visitor cannot actually get.

It runs entirely in your browser. Nothing you type is sent to a server, no account is required, and there is no usage limit, because there is no cost per run to control.

The grid already said the price, so stop paying to say it again

Open a phone and search a category term you sell into. The first unit is almost always product tiles: photograph, title, price, merchant, sometimes review stars and a delivery line, all pulled from Merchant Center rather than written by anyone. Scroll past it and there is your text ad, repeating the product noun and quite often the same price. You have paid to occupy two positions and used both to communicate one fact.

What the feed physically cannot show is the interesting part of most retail decisions. It cannot say the return window is sixty days, that the range runs to a 4XL, that stock is in a UK warehouse rather than shipping from abroad, that the item is made to a specification the cheaper listing above it is not, or that a human answers the phone if the size is wrong. Those are the reasons someone buys from a merchant rather than from the cheapest tile, and every one of them fits inside thirty characters.

This changes which of the six asset categories deserves the most slots. Keyword-match assets still matter, because a query match earns bolding and confirms the visitor landed somewhere relevant, but two is usually enough on a category term. The objection-handling category is where retail wins: returns, delivery timing, sizing risk and the fear of buying from a merchant nobody has heard of are the specific hesitations that stop a cart, and they are hesitations the grid above you leaves entirely unanswered.

Query type should also change the weighting. On a brand term you are defending a name the searcher already trusts, so the job is to occupy the space and confirm the offer, not to persuade. On a model-number query the searcher has decided what to buy and is choosing where, which makes delivery, price certainty and returns the whole argument. On a broad category query nobody has decided anything yet, and that is the only case where a benefit headline about the product itself is doing real work.

A discount typed into a headline never expires

Promotion assets and countdown customizers exist because Google understood that retail copy is time-bound. Both carry scheduling: a start date, an end date, and in the countdown case an automatic wind-down that switches to your default text when the clock runs out. A headline that says twenty per cent off carries none of that. It is a string. It serves until a person deletes it, and people forget.

The consequence is not just an embarrassing ad. Currently documented policy treats an offer the visitor cannot obtain when they arrive as misrepresentation, which is a disapproval category rather than a performance problem, and price mismatches between ad text and landing page fall in the same territory. The feed side of your account resolves this automatically because the price is read from the feed. The text side does not, which is precisely why the two drift apart during a sale and stay apart after it.

The operational fix is to decide, once, which fields are allowed to contain a number. A defensible rule is that percentages and prices live in promotion assets and ad customizers fed from a spreadsheet, that headlines carry only claims that stay true all year, and that any exception is written into the campaign naming so the end-of-sale sweep can find it. That rule costs nothing to adopt and removes an entire class of recurring January work.

If you do use ad customizers to inject a live price into a headline, count the longest value the feed can return rather than the one showing today. A customizer resolving to a four-digit price plus a currency symbol is four characters longer than a two-digit one, and when the substitution overflows the field Google falls back to the default text you set. A default nobody thought about is how a carefully priced headline quietly becomes a generic one at exactly the moment traffic peaks.

  • Scheduled, so it ends itself: promotion assets, countdown customizers, seasonal ad customizer rows with date columns.
  • Fed, so it stays accurate: price, stock status and shipping estimates pulled through the feed or a business data source.
  • Typed, so it needs a human: anything you wrote directly into a headline slot. Keep this set free of numbers that can go out of date.
  • Sweep list: before peak, export every asset containing a digit or a percent sign and confirm each one has an owner and an end date.

Peak trading rewards preparation and punishes editing

Retail demand is not evenly distributed and search costs move with it. In the weeks around a major sales event, competition for the same queries rises, auction prices rise with it, and the margin on any given click compresses. That is exactly the moment when an account has the least tolerance for a learning reset, a disapproved asset or a headline that stopped being true. Everything you want live in November should be written and approved in October.

The volume problem compounds it. A catalogue of any size generates ad groups faster than a person can maintain them, so the realistic goal is not fifteen bespoke headlines per group. It is a set of durable assets shared across the account, plus a small number of group-specific lines that name the actual product. Chasing bespoke copy everywhere produces the worst outcome available: hundreds of ad groups where nobody remembers what is in them.

Judge the whole thing against contribution rather than cost per click. A cheap click on a low-margin accessory can be a worse trade than an expensive one on a high-margin item, and blended return on ad spend across a mixed catalogue hides which of the two you are actually buying. If you sell across a wide price range, segment the reporting by margin band before you decide any headline is working, because the average is describing a customer nobody is.

Numbers worth knowing

MetricTypicalWhat it means
Where a text ad sits on a retail queryusually beneath the product gridShopping and Performance Max tiles render image, price, merchant and stars from Merchant Center. Layout varies by query, device and competition, and Google reshuffles retail results regularly, so check your own top terms on a phone rather than assuming.
Cost per click, retail non-brandtypically the lowest of the major verticals, across a very wide rangeA category term for a fifteen pound accessory and one for a three thousand pound sofa are both retail. Judge any figure against gross margin per order rather than an industry average, because the average is a blend of two different businesses.
Promotion assets and countdown customizerscarry start and end datesThey stop serving on their own. A discount typed into a headline has no expiry and continues until a human edits that ad group, which is the usual source of post-sale disapprovals.
Price or offer stated in ad textmust be obtainable on the landing pageCurrently documented policy treats an unavailable offer as misrepresentation rather than as weak copy. Feed-driven surfaces reprice themselves; hardcoded headline text does not.

Mistakes that quietly cost you results

Repeating the product name and price the Shopping tile already showed
You have bought two positions on one query and used both for the same fact. Spend the text slots on the things a feed cannot render: return window, delivery cut-off, size range, warranty, who you actually are.
Typing the sale percentage straight into a headline
Headline text has no end date, so it serves until someone deletes it and becomes an unavailable offer the day the sale closes. Put percentages in promotion assets or customizers that carry their own scheduling.
Writing the same headline set for brand, category and model-number queries
These are three different decisions. Brand traffic needs occupation and reassurance, model-number traffic is choosing a merchant on delivery and returns, and only broad category traffic is still deciding what to buy at all.
Optimising the whole account to a blended return on ad spend
A mixed catalogue hides which margin band the return came from, so the number describes an average customer who does not exist. Segment by margin before deciding any asset set is performing.

What does the output look like?

This is the exact output the tool produces from the example inputs. It is generated by the same code that runs when you click the button, so what you see here is what you get.

ASSET SET: Merino wool base layers Keyword: merino base layer · Audience: winter hikers · Offer: Free returns for 60 days Campaign type: performance-max HEADLINES (15), limit 30 characters, spaces included [keyword-match] 1. Merino Base Layer for Winter Hikers 35/30 FAIL trimmed to fit: Merino Base Layer for Winter (28/30) 2. Merino Base Layer 17/30 PASS 3. Compare Merino Base Layer Tools 31/30 FAIL trimmed to fit: Compare Merino Base Layer (25/30) [benefit] 4. Cut Merino Base Layer Admin 27/30 PASS 5. One Place for Every Job 23/30 PASS 6. Stop Chasing Updates 20/30 PASS [offer] 7. Get Free Returns for 60 23/30 PASS 8. See This Month’s Deal 21/30 PASS [proof] 9. [X]+ Teams Switched Over 24/30 PASS 10. Rated [X.X]/5 by Customers 26/30 PASS [objection-handling] 11. Your Data Stays Yours 21/30 PASS 12. Set Up in an Afternoon 22/30 PASS [call-to-action] 13. Compare Plans Today 19/30 PASS 14. Try It This Week 16/30 PASS 15. Get a Quote 11/30 PASS DESCRIPTIONS (4), limit 90 characters 1. Run merino base layer without the spreadsheet sprawl. Free returns for 60, then decide. 87/90 PASS 2. Compare plans, see the real price, and start when you are ready. Free returns for 60 on every plan. 99/90 FAIL trimmed to fit: Compare plans, see the real price, and start when you are ready. Free returns for 60 (84/90) 3. Set up in an afternoon, not a quarter. Bring your existing work across. Free returns for 60. 92/90 FAIL trimmed to fit: Set up in an afternoon, not a quarter. Bring your existing work across. Free returns (84/90) 4. Winter hikers switch for one reason: nothing falls through the gaps. Free returns for 60. 89/90 PASS LONG HEADLINES (Performance Max only), limit 90 characters 1. Merino Wool Base Layers for Winter Hikers: Free Returns for 60, then keep the plan that fits. 93/90 FAIL trimmed to fit: Merino Wool Base Layers for Winter Hikers: Free Returns for 60, then keep the plan that (87/90) 2. Everything for Merino Base Layer in one place, set up in an afternoon. Free Returns for 60. 91/90 FAIL trimmed to fit: Everything for Merino Base Layer in one place, set up in an afternoon. Free Returns (83/90) 3. [X] Winter Hikers run their projects on Merino Wool Base. Free Returns for 60 to see why. 89/90 PASS Asset groups also expect a short description on a tighter limit than 90, plus images and a logo. Check the current field limits in the asset group editor before you paste, because they differ from the RSA set above. MEASUREMENT SUMMARY Headlines within limit: 13/15 Descriptions within limit: 2/4 Assets making a checkable claim: 9/15. Every one needs evidence you would be comfortable showing a reviewer. Assets containing an [X] placeholder: 2. Substitute a real figure, then recount, because a four-digit number is three characters longer than the placeholder. Your keyword is 17 characters, so a headline that contains it in full has 13 characters left for everything else. That is enough room for a qualifier, but not for a qualifier and a call to action in the same line. VARIETY CHECK (pairwise token overlap, stop words removed) Candidate lines discarded during generation for repeating an earlier asset: 1. Closest surviving pair: headlines 1 and 2 at 60% shared vocabulary. Nothing in the delivered set crosses the 65% threshold, so each slot carries an argument the others do not. That is the condition Ad Strength is grading, and the condition the combination engine needs in order to learn anything. PINNING Leave everything unpinned unless a line is legally or contractually required to appear in a fixed position. Pinning one headline to position 1 removes that position from testing entirely; pinning all three positions turns a responsive search ad into a static ad and usually downgrades Ad Strength. If you have to pin, pin two or three assets to the same position so the system still has a choice inside the constraint. Limits used here are the values Google currently documents (30 characters for a headline, 90 for a description). Field limits and available asset types change, and are halved for double-width languages, so confirm in the editor before a bulk upload.

Frequently asked questions

If Performance Max already covers my product queries, is a text ad still worth writing?

It is, but for a different job. The feed-driven surfaces handle the visitor who has decided what to buy and is choosing a merchant on price and picture. Text assets reach the queries that have no clean product match: research phrasings, comparison wording, sizing and fit questions, and your own brand name. Those need words rather than tiles, and they are also where you control the message rather than letting the feed compose it.

Should the price appear in a retail headline at all?

Only when your price is the argument, and even then prefer a customizer over typed text. If you are consistently cheaper on a specific range, saying so beats making the visitor compare, but it commits you to editing that headline every time the price moves. If you are not the cheapest, the slot is better spent on the reason someone would pay more, which is usually returns, availability, speed or specification.

How do I stop a promotional headline running after the sale has ended?

Do not put it in a headline. Promotion assets and countdown customizers accept a start and an end date and stop on their own, so the expiry is part of the asset rather than a task on someone list. If you must type a promotion into a headline, record it in the campaign name or a shared spreadsheet during the same session you create it, because the sweep at the end of the sale can only find what someone wrote down.

What should objection-handling headlines say for an online store?

Name the specific reason a cart gets abandoned in your category. For clothing it is sizing and returns; for furniture it is delivery timing and whether it fits through a door; for electronics it is warranty and whether the seller is authorised. Generic reassurance like satisfaction guaranteed is doing none of that work, and every competitor tile carries the same implicit promise anyway.

How many bespoke headlines does a large catalogue actually need?

Fewer than most accounts try to maintain. A durable set of shared assets covering returns, delivery, range and trust, plus two or three lines per group naming the actual product and its key attribute, is maintainable and testable. Bespoke copy in every ad group sounds thorough and reliably ends as hundreds of groups nobody has read since launch, which is worse than a smaller set that is actually kept accurate.

Related free tools

Some links on this site are affiliate links, which means Hacking Demand may earn a commission if you buy through them at no extra cost to you. This does not influence which tools are listed. The tools on this page are free and have no affiliate relationship of any kind.