"The ads aren't working" usually means one of two things: the account is getting results the business can't see, or it's getting results the business doesn't want. Over five years of taking on Meta accounts, the cause has almost always sat in one of five places. I check them in this order, because the early ones make the later ones look worse than they are.
1. The objective is wrong
Meta gives you what you ask for. If a campaign is set to engagement, it will find people who like and comment, and the cost per like will look wonderful. It will not find people who buy.
A contracting company in Bahrain came to me with an engagement campaign that was producing WhatsApp conversations as a side effect, at $7.04 each. I replaced it with campaigns optimised for conversations directly, and the next ones came in at $5.09, $4.26 and $3.23. Same business, same audience, same platform. The only change at first was telling Meta what the result was. The Al-Rehman case study has the full table.
Check: open the campaign, look at the objective and the optimisation event. If it says traffic, engagement or reach and the business wants sales or leads, that's the first fix.
2. The platform can't see the result
The second most common problem is that the campaign is set to purchases or leads, but the pixel isn't recording them properly, so Meta is optimising on a guess.
Cheezious, a fast food brand whose account I took on in 2023, was further back than that: only awareness and engagement campaigns, no sales campaigns at all, and no pixel recording purchases, so nobody could say what the ads sold. GA4, Tag Manager and the pixel went in first, with purchases tested order by order. Only then did I launch sales campaigns, and I scaled the monthly budget as the return held. The account averaged a 36.88 ROAS, because Meta was learning from real orders. The tracking case study walks through what was set up.
Check: Events Manager. Is the purchase or lead event firing, is it deduplicated between the pixel and the Conversions API, and does the count roughly match what the business saw last week? If not, stop and fix this before anything else, because every other decision rests on it.
3. The creative has gone stale, or never worked
Audiences see the same ad for weeks and stop reacting. The cost per result creeps up and nobody can say why, because nothing was changed.
The clearest example I have is on TikTok rather than Meta, but the lesson is the same. For Rewayat, a restaurant group, the Eid and Mother's Day discount videos held 1.3% and 0.9% of viewers past six seconds. The rooftop launch walk-through held 41.7%, with an average watch time of twelve seconds. The impressions cost about the same. One creative was worth running and the others weren't, and the only way to know was to test them against each other. The Rewayat TikTok case study has the full table.
Check: frequency and the date the creative was last changed. If frequency is high and the ads are a month old, new creative is the cheapest fix available, and I test new ads every fortnight as a rule.
4. The audience is too broad, or the wrong people
Broad targeting works when the pixel has plenty of purchase data to learn from. On a new or badly tracked account it finds whoever is cheapest to reach, which is rarely a buyer.
LandLife, a property developer I worked for in-house from 2022, had broad campaigns producing leads with neither the budget nor the intent to buy. Rebuilding the targeting around property interest and purchasing power, adding lookalikes of past buyers and putting a qualifying question on the form brought the cost per result down 60 to 75 percent and produced 2,422 qualified leads in three months. The case study covers it.
Check: who is actually converting, in the breakdown by age, location and placement. If the campaign is spending on segments that never convert, exclude them. If it's set to the whole country for a business with two branches, draw the radius.
5. Nobody is following up
This one isn't in Ads Manager at all. The ads bring a lead or a WhatsApp message, and nobody replies for a day. By then they've booked with someone else, and the report says the ads didn't work.
For Rewayat Events' catering enquiries, labels on every WhatsApp chat (new, qualified, quoted, booked) showed which campaigns brought people who paid. On LandLife, a CRM with an assignment rule put each lead in front of the right salesperson in minutes. In both cases the reply speed had more effect on the outcome than the next ad change would have.
Check: how long a lead waits for a first reply, and who is responsible. If the answer is "it depends", that's the problem.
The order matters
Fix the objective, then the tracking, then test creative, then tighten the audience, then sort the follow-up. Done in that order, each fix makes the next one easier to judge. Done in the wrong order, you end up changing creative on a campaign that was never asked to sell anything, and blaming the creative when it doesn't.
If you'd like a second pair of eyes on an account, read-only access is enough for me to run through these five and tell you what I'd change first.
I run paid ads on Meta, Google and TikTok across the UK, the Gulf and Pakistan, and build the tracking and pages around them. Get in touch or find me on LinkedIn.