Why my Facebook ads are not working

Key takeaways

• Boosting a post buys visibility. It does not test whether anyone wants what you sell.
• A broad audience produces averages, and averages cannot be attributed to anything you can change.
• An ad without a specific claim gives strangers nothing to agree or disagree with.
• One day of spend against a few dozen people is a coin flip, not a result.
• Where people stop tells you whether the ads are broken or the offer is broken.
• A campaign with no decision written down before launch is spending, not testing.

The complaint covers two completely different situations, and the first job is to work out which one you are in. Either the ads run and produce nothing at all — no clicks, no messages, no calls — or they produce clicks and the clicks produce nothing. Those have different causes and different repairs, and treating them as one problem is how a budget disappears eighty dollars at a time over three months.

Very little of what follows is about Facebook. The platform is a delivery system. It takes whatever promise you hand it, shows that promise to whoever you described, and does it fast enough to make a bad decision expensive. When a campaign fails, the failure is nearly always in the promise, the audience, the page, or the definition of success. All four belong to you.

The two failures that look identical from the dashboard

A campaign that produced nothing and a campaign that produced the wrong thing both show up as a number you do not like. The difference is where people stopped. Everything useful in ad diagnosis comes from locating that stopping point rather than staring at the total.

A plumber running "Quality plumbing, family owned since 2011" who gets four hundred impressions and one click has a promise problem: nobody saw a reason to react. A plumber running "We fix the same leak twice for free" who gets sixty clicks that bounce off a homepage full of stock photography has a delivery problem. Both say the ads are not working. Only one has an ad problem.

Boosting a post is not a test

The boost button is the most expensive convenience in small business marketing. It takes something you already published and buys it more views, optimized by default for the cheapest possible reaction — so Meta finds the people most likely to tap a like button, because likes are abundant and inquiries are scarce. You get a post that looks popular and a business that is unchanged.

Boosting is not useless. If you want more of an existing audience to see something good, it does that. But it cannot answer the question you are asking, which is whether a stranger feels a problem strongly enough to act. It has no clean audience, no defined objective, usually no distinct destination, and no threshold — four missing parts, all of them the parts that make a result readable.

What a real campaign has that a boost does not

• A chosen objective. You tell the platform what action counts, and it goes looking for people who take that action rather than people who tap things.
• A defined audience. One group, described specifically enough that a failure tells you something about that group.
• Several angles of the same claim. Three or four creatives testing different ways into the same problem, so a flat result is about the problem rather than one unlucky sentence.
• A destination built for the promise. A page whose first line is the same sentence as the ad, not a homepage that lists everything you do.

If you have never run anything other than boosts, the mechanics of the real version are in how to run your first Meta ads campaign. It is less complicated than the interface makes it look, and the difference in what you learn is enormous.

The audience is too broad to fail cleanly

When results are bad, the instinct is to widen the net. More people means more chances, and the platform even encourages it. This is backwards, and it is the single most common structural mistake in small business advertising.

A broad audience produces an average, and averages are unactionable. Target everyone within forty miles between twenty-five and sixty-five, get a middling cost per click and no inquiries, and you have learned nothing you can change: the number is a blend of one group that might have wanted this and eight that never would. Narrow it until you could name three real people who fit, and failure becomes informative — that group, given that claim, did not react.

Meta will faithfully deliver an unclear promise to the wrong people, on time and under budget.

What narrow looks like in practice

• A landscaping business. Not "homeowners nearby." Homeowners in three named suburbs with houses above a certain value who have moved in the last year, because new owners buy garden work.
• A bookkeeping practice. Not "small businesses." Restaurant and salon owners in one metro area, because their books are messy in a specific, recognizable way.
• A Shopify store selling dog gear. Not "dog owners." Owners of large breeds who pull on the leash, which is a problem with words attached to it.
• A freelance designer. Not "businesses that need branding." Independent food producers about to get shelf space, who have a deadline and a reason to care this month.

The narrowing decision is really a channel decision in disguise, because some audiences are simply not reachable well on Meta. How to choose a marketing channel for a small business covers when the answer is a different room entirely.

There is no claim in the ad

Read your ad back and ask what a stranger could disagree with. If the answer is nothing, there is no claim, and without a claim there is nothing to respond to. "Professional cleaning services for your home or office" is not a claim. It is a category. Nobody has ever felt anything about a category.

A claim names a person, names a problem in the words that person uses, and states an outcome specific enough to be checked. "Your cleaner keeps missing the same three things. Ours photograph the kitchen before they leave." That is disagreeable. Somebody reading it either recognizes the situation immediately or does not, and both reactions are useful to you.

Most ads have no claim because the owner writes from the far side of their own expertise. You have thought about this work for years, so you compress it into a category name. Your buyer has a specific bad afternoon and a workaround they are slightly embarrassed about. How to write ad copy for a small business is about closing that gap, and how to write a problem statement from a conversation is about finding the sentence in the first place.

The objective is buying the wrong action

Meta optimizes hard for whatever you told it to optimize for, and it is very good at it. That is the trap. Choose engagement and you will get engagement — comments from people three states away, likes from accounts that will never buy, and a campaign that looks alive while producing nothing. The platform did exactly what you asked.

The objective should match the action that would change your mind. If a reply in the inbox is what convinces you, optimize for messages. Every objective further from a real commercial action makes the numbers cheaper and less meaningful at once, which is the worst combination for someone trying to decide something.

One day of spend is not a sample

Meta distributes unevenly while it learns, so the first day or two tells you almost nothing about the days that follow. Forty impressions and no clicks feels like a verdict. It is closer to flipping a coin twice and concluding the coin is broken.

Decide the window before launching, run to the end of it, and read the whole thing at once. Four to seven days is usually enough for direction at small budgets. The temptation on day two runs in both directions and both are bad: killing something that had not been delivered yet, or pouring more money into something that got lucky early. How to read ad test results covers what volume a readable answer actually requires.

Rough sample sizes to keep in your head

• Under a few hundred impressions. Nothing. Do not adjust anything on the basis of it. You are watching delivery start.
• A few thousand impressions, one narrow audience. Directional. Enough to tell an ignored claim from one that gets a reaction.
• A hundred or more clicks on one page. Enough to see where people stop on the page, which is often the real answer.
• Fewer than ten conversions. Not a conversion rate. It is a handful of events you are about to over-interpret.

The creative describes you instead of them

Open the ads you are running and count the sentences about your business — how long you have operated, how much you care, how experienced the team is. Then count the sentences about the situation the reader is in right now. In most failing campaigns the ratio runs four to one in the wrong direction.

This is not a style preference. A stranger scrolling has no reason to care about your history, because they do not yet know they have a reason to care about you at all. The only thing that earns their attention is recognition — the small jolt of seeing their own situation described more precisely than they would have described it themselves. Credentials matter later, on the page, once recognition has already happened.

The fastest fix costs nothing. Go where your buyers already complain: a local Facebook group, a subreddit, the review section of whatever they currently tolerate, your own text messages from customers. Copy the phrasing they use before anybody has sold them a framework. Put that phrasing in the first line. How to get customers from Reddit is largely an exercise in learning that language, and it happens to produce customers as a side effect.

The landing page contradicts the ad

This is the most common cause of the specific complaint "I get clicks but nothing happens." The ad promises one thing, the click lands on a homepage that promises eleven things, and the visitor spends two seconds trying to work out whether they are in the right place before leaving. The ad worked. The transition destroyed it.

The rule is boring and almost always ignored: the first line the visitor reads on the page should be recognizably the same sentence as the ad they clicked. Same problem, same words, same promise. If the ad said "same leak twice, fixed free," the page says that, immediately, before anything about the company. Continuity is not a design flourish; it is the visitor confirming they were not tricked.

Everything downstream of that first line — proof, friction, price framing, and how the page reads on a phone — is covered in traffic but no sales, which picks up exactly where this post stops.

No decision was written down before launch

Ask most owners what result would have made them stop, and they cannot answer. That is the tell that the campaign was never a test. Without a threshold set in advance, every outcome reads as whatever you already believed on the morning you launched. A weak result becomes "it needs more time." A lucky day becomes "it is working," and the budget quietly grows.

Writing the threshold down is a five minute job and it changes the entire character of the spend. Something like: "If four hundred dollars across five days against roofers within thirty miles produces fewer than eight replies, the claim is wrong and I will rewrite the problem rather than the headline." Now the money buys a decision instead of an experience. What counts as an acceptable cost per reply for your kind of business is worked through in what is a good cost per lead for a small business.

Readable test

An advertisement of one specific claim to one narrow audience, run for a fixed window at a budget large enough to reach a few thousand of the right people, with a success threshold written down before launch. Anything missing one of those four parts produces spending rather than evidence.

Telling a broken ad from a broken offer

This is the distinction that decides what you do next, and it costs nothing to make. Broken ads fail before comprehension. Broken offers fail after it. The stopping point separates them cleanly, which is why locating it matters more than any individual metric on the dashboard.

Signs the ads are the problem

• Impressions with almost no clicks. People saw it and felt nothing. The claim or the audience is wrong.
• Wildly different results across similar creatives. The delivery is still unsettled, or one creative accidentally described a different problem.
• Clicks that leave instantly. A mismatch between the ad and the page, not a rejection of the offer. They never read the offer.

Signs the offer is the problem

• Full reads with no action. They understood it and declined. That is a judgement about the thing itself.
• Enquiries that stall at price. The problem was never sized, so the number reads as pure expense rather than a comparison.
• Interest with no urgency. People say to contact them next quarter. The problem is real and not felt today, which is a market timing answer.

If you work through this and the honest conclusion is that nobody is reacting anywhere, that is a demand question rather than an advertising one, and why is no one buying my product takes it apart in the order that costs least.

Rebuilding the test in one week

The repair is not complicated, and it is cheaper than the drifting spend it replaces. One claim, one audience, one page, one window, one decision.

1. Write the claim as one sentence. Who it is for, what problem it removes, what happens next. If it does not fit in a sentence, the ad will not fit it either.
2. Narrow the audience until you can picture three people in it. A trade and a radius, or a life event and a zip code. Never an age range and a city.
3. Produce three or four angles on the same claim. Different ways into the same problem — the cost of it, the embarrassment of it, the time it eats, the thing that goes wrong when it is ignored.
4. Build one page whose first line repeats the ad. One promise, one action, one form field if you can manage it. No navigation bar full of escape routes.
5. Choose the objective that matches the action you care about. Messages or leads for services, purchases for stores with working tracking, never engagement.
6. Write the threshold down before you launch. The number of replies, at what spend, by what date, and what you will change if it is missed.
7. Run to the end of the window and read the drop-off point. Not the total. The point where people stopped is the sentence the test is actually speaking.

What a small budget honestly buys

It helps to have realistic expectations about what different amounts of money can settle, because most disappointment with paid ads is a mismatch between the size of the question and the size of the budget.

Sizing the budget to the decision you want to make, rather than to what feels affordable in a bad month, is the subject of how much should a small business spend on ads.

When Meta is the wrong room

Some businesses genuinely should not be advertising on Facebook or Instagram first, and no amount of repair changes that. If your buyer only looks for you at the moment something breaks, search intent beats interruption. A locksmith does not need to be discovered while somebody scrolls; they need to be there at eleven at night when a person is standing outside their own front door.

Equally, if your buyers gather in a forum, a professional network, or a handful of local groups, turning up in those rooms consistently may outperform paid social. That work is slow and cumulative, which is why how to post on social media consistently for a small business is more about routines than about content, and why how to market a local business online spends most of its time on the unglamorous channels.

Related reading

• Traffic but no sales — what to fix once the clicks are arriving and stopping
• Why is no one buying my product — the four causes of silence, in the order that costs least
• How to run your first Meta ads campaign — the mechanics, without the boost button
• What is a good cost per lead for a small business — how to set the threshold before you launch
• An ad is not a company — why a good result is permission to continue, not proof of a business