Free ad copy generator for a stronger first draft.

A mobile bike repair service runs two vans across a handful of city postcodes. The mechanic comes to your home or your office, a standard tune up is a flat rate, and most jobs are finished in under an hour on the pavement outside. The audience is commuters who ride four or five days a week. The single problem: getting a bike to a shop and back costs an evening or a Saturday.

Written as a responsive search ad, that becomes a set like this. Headlines: "Bike Repair At Your Door", "Fixed While You Work", "Flat Rate Commuter Tune Up", "Mon to Sat, Two Hour Slots". Description: "We service your bike outside your home or office. Most tune ups done in under an hour, no shop trip."

Each headline sits inside the 30 character limit, and each one can appear in any order beside the others, which is the constraint people forget: headlines are assembled, so no line may depend on the line before it. Every one of them holds a complete idea. The description does the job the headlines cannot, which is to remove the objection outright: nobody is being asked to leave a bike anywhere or take time off work.

The same angle in a social feed is a different piece of writing, because the first line carries everything before the more link and there is no search query telling you what the reader wants. It opens on the situation instead: "Your bike has needed a service since March." Then the mechanism, then the offer, then the action. Feed copy can afford three short lines where search copy cannot afford three words.

The weaker version is the one that gets written by default: "Bike Repairs You Can Trust | Quality Service | Best Prices | Call Us Today". It fails in four separate ways. Trust and quality are claims every competitor makes in the same words, so they carry no information and cannot be judged from a results page. Best prices asserts a comparison the reader has not made and the advertiser cannot support. Call us today asks for the action before giving a reason for it. Most damaging, nothing in the entire ad mentions the one thing that makes this business different, which is that nobody has to go anywhere. The mechanic could hand that ad to any bike shop in the country and it would fit them equally well, which is the clearest sign that copy is doing no work.

Do not write ad copy before you know what happens after the click. If there is no landing page, or the page is a home page with six equally weighted options, the ad has nowhere to send anyone and the spend goes on introducing people to a dead end. Build the page around the single promise first, then write the ad from the page rather than the other way round. The sentence at the top of the page and the strongest line in the ad should be close to identical.

Do not use it to paper over an offer nobody wants. Advertising accelerates whatever is already happening: if the offer does not convert when a warm referral lands on it, paid traffic will produce the same outcome faster and with an invoice attached. Fix the offer, the price or the proof first. That is unglamorous work, but writing sharper headlines around a proposition people are declining for good reasons is the most expensive way to learn nothing.

Hold off while the tracking is broken or absent. Copy is only worth iterating on if you can tell which version produced enquiries rather than which produced clicks, and those two rankings frequently disagree. Without conversion tracking, or at the very least a phone number and a form you can attribute, you are choosing headlines on the click through rate, which rewards curiosity and punishes qualification. An honest ad that repels the wrong readers looks worse on that measure and better on the invoice.

Finally, check the rules before the words. Regulated sectors, health claims, financial promotions, price comparisons and anything aimed at under eighteens all carry restrictions on what an ad may say, and platforms enforce their own policies on top. A rejected ad is a wasted afternoon; an unsupportable claim that runs is a bigger problem. Draft inside the rules rather than writing freely and hoping to edit your way back.

Treat every placement as a different container. Google search gives you headlines of up to 30 characters and descriptions of up to 90, with several of each shown in a combination the system chooses, which means order is not guaranteed and no line may complete another. Write each headline so it survives being read first or last.

Feed placements are the opposite problem. There is no character cap that matters, only a fold: the copy is cut after a line or two on mobile with a more link, and the cut point moves with device and language. Everything load bearing has to sit above it, and that includes the offer, not just the hook. Display sizes cut differently again, since the constraint is the box rather than the count. The practical method is to write the shortest usable version first, the one that would work as a 30 character headline, and expand outwards from it. Copy written long and then trimmed loses its point in the trim; copy written short and then extended keeps it wherever the cut lands.

Three parties judge it and they do not agree. The platform judges relevance between the query, the ad and the page, and it will quietly reduce how often a mismatched ad shows. A human reader judges it in about a second, scanning for whether it answers the thing they just typed. Your accounts judge it on cost per enquiry, which is the only one of the three that pays anybody.

The trap is optimising for the reader's attention alone. A headline built on curiosity can lift the click through rate and raise the cost per enquiry at the same time, because it pulls in people who were never going to buy. Read the numbers as a pair. If clicks rise and enquiries do not, the copy is attracting the wrong reader or promising more than the page delivers. If clicks fall and enquiries hold, you have written a more qualifying ad, which is usually a good outcome even though the dashboard makes it look like a decline. Judge a variation on enough conversions to mean something, not on a two day difference in click rate.

These three get written by the same person in the same afternoon and they are not the same craft. An ad is read by someone who has no relationship with you and did not ask to see it, so it has to earn attention and qualify the reader at once, and it competes sideways against other results on the same screen.

A landing page headline is read by someone who has already clicked, so its job is confirmation rather than persuasion: it repeats the promise so the reader knows they are in the right place, and being nearly identical to the ad is a feature rather than laziness. An email subject line goes to someone who chose to hear from you, so it can be quieter, more specific and more familiar, and it is judged against the rest of an inbox rather than against competitors. Copy that would make a good subject line usually makes a weak ad, because it assumes context the cold reader does not have. Copy that makes a good ad often reads as shouting in an inbox.

A single ad is not testable, and twelve at once is not attributable. Build the set around one variable. Keep the audience, the offer and the landing page fixed, and vary only the angle in the headline: one on the problem, one on the mechanism, one on the objection, one on the specific outcome. Three to five variations per group lets a difference show while keeping the cause readable.

Give them a run that covers whole weeks, because behaviour on a Tuesday morning and a Sunday evening are not the same, and stopping early on a promising day is how people convince themselves of things that are not there. When one angle wins, do not simply keep it. Write the next set from it, treating the winner as the new baseline and varying the next element down: the description, then the call to action, then the framing of the offer. Keep a plain record of what you tried and what happened, since the compounding value of this work is a document of what your audience responds to, and that outlives any single campaign.