Your Ads Are Fine. What They Point at Isn’t.

There is a sentence that turns up over and over in the forums where small sellers talk about ads: “I’m seeing absolutely zero sales.” Then someone asks about the click-through rate, someone says try a different audience, someone says the platform is broken this month. Everyone is talking about the ad. Almost nobody asks what happened after the click.

Ads do not create demand. They buy traffic for a machine you already own. If that machine cannot turn a stranger into a buyer today, for free, spending money only means more strangers watching it fail to do that, faster, at £1 a go. Here are three things, all free to fix, all worth doing before a single pound goes out.

One page that keeps one promise

Your ad says one thing. Then the click lands somewhere that says something else, or nine things at once: a homepage listing six services, an about page, a blog and a contact form, or a shop grid of forty products when the ad showed one candle. The person came for one thing and arrived at a menu.

The check: read your ad, then read the first screen of the page it points at, out loud, as if you have never heard of yourself. If a stranger would have to scroll to find out they were in the right place, they will not scroll. The fix is usually deletion, not building. One page, one promise, one thing to do next.

An offer a stranger can say yes to without asking a question

Three things decide this: the price, what is included, and who carries the risk if it goes wrong. If a buyer has to email you to find out any of the three, your ad is buying enquiries that you answer by hand, mostly to people who were never going to buy. The check: write down every question a buyer has to guess the answer to. That list is what the offer really costs.

One outcome you can actually see

Not a dashboard. Not eleven metrics. One event that fires when the thing you care about actually happens: the enquiry counted as an enquiry, or the purchase recorded once, with the right amount on it. Then a theme gets updated, or a form plugin changes, and the number quietly stops moving while the money keeps going out.

The check: go through your own buying or enquiry process as a stranger, on your phone, with your card or your own email address. Then look at whether the number moved. This is the least glamorous of the three and the one I would do first, because without it you cannot tell a bad ad from a broken page.

What it looks like when the three are true

I know one account where the ads do work. In one week it spent £207.96 and brought back 123 leads. That is £1.69 each. There is nothing clever in those ads. The page makes one promise, the thing on offer is obvious and free, and the event fires the moment somebody signs up.

Your one job this week

Pick the third one. Go through your own enquiry or checkout as a stranger, and confirm the number moved. It takes about ten minutes and it costs nothing. If the number does not move, you have found something worth more than any ad you were about to run.

The week in brief

• Google started rolling AI visibility reporting into Search Console for all sites in the week of 7 September.
• Reddit opened its Max automated campaigns to everyone on 8 September. Test it with £50 you can afford to lose.
• Google Analytics zeroed out 1 September for a lot of accounts. It is the second measurement fault in a month, so screenshot your monthly numbers on the first of each month.
• Amazon extended Price History from 90 days to 365 in the week of 8 September.
• Klaviyo said on 9 September that more than 260 of its tools can be reached directly by AI systems. Decide who is allowed to launch a campaign that way.
• On 10 September Amazon Ads and OpenAI opened a pilot letting some US advertisers extend campaigns into ChatGPT.
• YouTube will allow personalised alcohol ads from 30 October where local law permits, reported from a weekly recap, so check the policy itself first.