Copywriting: Why Good Copy for the Wrong Keyword Fails
Copywriting is writing text meant to move the reader to action: an inquiry, a signup or a purchase. It is judged by results, not style, and that is where the trouble starts, because a piece can be very good and completely worthless at the same time. We have our own measurement of this: one of our posts got indexed by Google within two days, landed around position seven, and brought in 22 impressions and zero clicks, because the keyword it was written for had no measurable demand in our market. Below we break down what to check before anyone writes the first sentence.
Why is the most expensive mistake made before writing?
Because a piece written for a keyword with no demand cannot be fixed by editing. You can shorten it, lengthen it, rewrite the style, and the result will be the same.
In August 2026 we compared 14 of our posts against Google Search Console data. Together they produced zero clicks and about 23 impressions. When we checked why, it turned out that six of the fourteen topics had no measurable search demand at all. The texts were fine. The queries did not exist.
We measure the same mechanism on every new project. For one industry cluster we checked 37 keywords and 20 of them returned no volume data at all. Writing twenty articles for those keywords would have cost months of work and brought in nothing.
Traffic from content is not the same as customers
There is a worse version of the same problem, and it happens more often: the content ranks, traffic grows, and there are no inquiries. The report looks good, so nobody looks for the cause.
In August 2026 we spoke with the owner of a hair salon whose blog drew decent traffic on hair care topics, the “how to take care of your hair” kind. The traffic was real. But it came for advice the reader would apply at home, not for a service. On top of that, the treatments the salon actually makes money on were not properly described, and the site had nothing tying it to the salon’s city. A person ready to book had nowhere to land, and the person who did land had no intention of buying.
This is not an argument against guides. It is an argument for checking three things before the next piece gets written:
- whether there is somewhere to send that traffic, meaning a described service,
- whether a page exists for the buying query, including the local one if the business sells in a specific city,
- whether you can count how many inquiries came out of it, not just how many visits.
If the answer to any of these is “no,” we start there, not with writing. How we put this together is described on our content marketing page.
Three things to check before writing the first sentence
1. Does the keyword have measured volume? Not “it seems popular,” but a number from a tool. Checking a few dozen keywords costs pennies, and the difference in outcome is binary.
2. What is the intent? Commercial and transactional keywords are won by service pages, not guides, because Google fills those results with offers. A guide written for a commercial keyword loses regardless of quality. If a topic comes with a commercial keyword, reframe it as a question.
3. What is the difficulty? Above 40, a single article will not make the top ten. Take an easier keyword from the same cluster and build the topic up step by step.
A practical note from our own work: check keywords in the exact form people type them. In one of our audits we entered part of the keywords in a shorthand form (in our home market, without the language’s diacritics) and got a false “no demand” for phrases that actually have several hundred searches a month. Spelling variants, plurals and abbreviations can hide real demand the same way.
What separates copy that works from copy that just exists
Search engines and language models filter out content that rewrites what is already in the index. That is why we use a gate every piece has to pass. A text goes out only if it contains at least one of five things:
- our own number from a campaign, benchmark or measurement, with the source named,
- our own screenshot from a dashboard or tool,
- a linked case study the argument rests on,
- a first-hand observation with a specific detail, not a generality,
- a position that is not the consensus, with the reasoning behind it.
The test is one sentence long: could a competitor publish the same text after swapping in their company name? If yes, the text does not go out. Better to have no piece than a generic one, because generic pieces drag down the rating of the whole domain.
Copywriting for Google alone is no longer enough
More and more queries end with an answer generated by a model, with no click on a result. That changes the goal of the text: it has to be quotable, not just findable.
We measure this on our own site. Our monitor asks four AI models real buying questions and checks whether our brand appears in the answer and whether our site is cited as a source. We do not publish the results, because they are our internal reference point, but the method is easy to repeat, and it is what shows whether changes in how we write make any difference.
One concrete editorial rule came out of that data: models cite specific, named entities with a clearly described specialty. The sentence “we are an experienced agency” is unquotable. The sentence “X runs Y for Z, and has served this many clients since this year” is. More on this part of the work on our AI SEO (generative engine optimization) page.
How to write so the text can be lifted out of context
Four things, all of which follow from how models read pages:
- The answer in the first paragraph, because that is where most citations come from.
- Definitions in a fixed pattern: the term, its category, what sets it apart. “Copywriting is writing text that…” instead of a three-paragraph run-up.
- Headings as real questions, because models match them to user prompts.
- Self-contained sections, roughly 150 words, that can be read without the rest of the text.
How much should copywriting cost?
Billing per word or per thousand characters rewards volume, and volume is not what the client pays for. Stretching two facts across 1,200 words makes the text worse and does not improve rankings.
A more sensible model is billing per piece, with keyword research, fact gathering and editing included, because those decide the outcome, and the writing itself is the shortest part of the job. When comparing quotes, ask directly what is included in the price besides writing. The same principle applies to other content services, which we broke down using the example of social media management pricing.
Where to start in your own business
- Check keywords before writing, not after. It is the only stage where you can still turn back at no cost.
- Decide where the specifics for each piece will come from. Without them the piece will be correct and indistinguishable.
- Separate commercial content from guides. The first goes on service pages, the second on the blog.
- Review after a few months. Rewrite or merge the pieces that do not work instead of adding new ones next to them.
adsfox is a marketing agency for service businesses that runs the whole of your marketing: content, SEO and AI search optimization, ad campaigns and your website. More than 350 clients since 2018, Badged Meta Business Partner and Google Partner.
Related adsfox services
Content production for your blog, service pages and campaign materials is handled by our content marketing agency. The technical side of visibility, site architecture and links is part of our SEO agency work, and visibility in AI answers is covered by our AI SEO agency. How to write for specific search keywords is broken down in our post on SEO copywriting.
Book a free marketing consultation: we will check whether your content targets keywords with real demand and tell you straight which topics have no chance.
