Does Social Media Marketing Work? You're Measuring the Wrong Thing

“All data is motivated, meaning the people producing it have a financial incentive.” — Scott Adams

Social media is not failing your business. You’re measuring the wrong thing … and the entire industry trained you to measure it wrong, because the wrong metrics are the ones that justify their fees.

I’ve spent fifteen years watching businesses pour time and money into social media and come out confused. I sold the first company I built. I’m building the second. A few hundred thousand dollars in advertising across every major platform. Agencies, in-house teams, doing it myself at 2 AM. I now run a company that creates AI-generated social media content. So I’ve seen this from every angle: as a buyer, as a skeptic, and as a seller.

Here’s what I’ve figured out: the businesses that feel like social media isn’t working are almost always chasing the metrics that platforms, agencies, and benchmark reports told them to chase. Followers. Engagement rates. Ad conversions. Reach. And those metrics, for most businesses, are either dead, fraudulent, or irrelevant to how customers make buying decisions.

The businesses that win with social media aren’t measuring any of that. They’re asking a simpler question: when someone checks us out, do we look credible and alive?

That’s a winnable game. And it’s a different game entirely than the one the industry sold you.

Why everyone feels like social media stopped working.

The promise was simple. Show up consistently. Post good content. Build an audience. The customers will follow.

That promise was always more complicated than it sounded, and by now it’s mostly fiction for the majority of businesses. Not because social media isn’t valuable. It is. But because organic reach on most platforms has collapsed to the point where “posting consistently” doesn’t mean what it used to.

If you believe the stats, organic social reach on Facebook is currently sitting around 1.65%. On Instagram, 3.50%. That means your typical post reaches 1 to 2 people out of every 100 who follow you. You built an audience of 2,000 followers and your post reaches 33 of them. You worked for years building that audience and the platform quietly turned a dial that made it nearly worthless for organic distribution.

This isn’t an accident. It’s business. The platforms need advertising revenue. The way you create demand for advertising is to make organic reach scarce. Facebook’s organic reach has been declining for over a decade in a slow, deliberate squeeze. Post for free and reach almost no one. Pay to reach the people who already follow you. That’s the model.

So the businesses following the “post consistently” advice are doing everything right and getting almost nothing back, not because they’re doing it wrong, but because the game changed and nobody told them.

Then there’s the paid side. The promise there was even simpler: tell us who your customers are, we’ll show them your ad, you’ll get clicks and conversions. Except the clicks aren’t always what they appear to be, and the platforms reporting those clicks are the same ones collecting your money.

I ran a Facebook ad targeting real estate agents.

I built a landing page specifically for real estate agents. It speaks directly to agents, shows them AI-generated social media posts for real estate, and offers a free signup. No credit card. No commitment. Just sign up and see your posts.

I set up a Meta ad campaign. Targeted real estate agents. The targeting was tight: job title, industry, interests. Meta’s tools let you get very specific about who sees your ad. That’s the whole pitch: “We know exactly who your customers are, and we’ll show your ad to them.”

293 landing page views. Zero signups.

Not one person. Out of 293 supposed real estate agents who supposedly clicked through to a page built for them, offering them something free, not a single one signed up.

I sat with that number for a while.

If those were real estate agents—people who need social media help, who are actively working, who clicked on an ad about a social media tool—there’s no scenario where zero out of 293 convert on a free offer. Even a terrible landing page converts 1-2% on a free signup. That’s just probability. You’d expect at least one or two signups by accident.

Which means either my landing page is so repulsive that it repels humans on contact, or those 293 “clicks” weren’t what Meta told me they were.

I’ve seen this before. At the first company I built, we easily spent a few hundred thousand dollars on advertising over the years. Across platforms. Different campaigns, different audiences, different creatives. The return on all of it was effectively zero. The only channel that consistently brought customers was search engines. People who were looking for what we sold, found us, and bought.

Fifteen years later, same story. Different company, different product, different market. Advertising produces nothing. Search produces customers.

The social media benchmark lie.

The reason most businesses don’t question this earlier is that the benchmark data looks convincing. Reports from Rival IQ, Hootsuite, Sprout Social—credible names, professional design, precise numbers. Financial Services on Instagram getting 3.70% engagement. Healthcare at 2.85%. It looks like a map you can navigate by.

I wrote a whole post about social media benchmarks a few weeks ago. In it, I showed how four different “credible” sources report wildly different engagement rates for the same industry on the same platform. Financial Services on Instagram: 0.26% from Rival IQ, 3.80% from Hootsuite. One number is 26 times larger than the other.

Since then, I’ve been thinking more about why this data exists and who benefits from it.

Rival IQ sells competitive analysis software. Hootsuite sells social media management tools. Sprout Social sells social media analytics. Every company publishing benchmark data has a financial incentive to make social media look measurable, optimizable, and worth your money.

Here’s what Rival IQ’s methodology page says: they select 150 companies at random per industry from a database of 200,000+, with requirements like being active on specified platforms and having certain follower thresholds. Hootsuite has been publicly criticized for using only 100 profiles per segment.

Think about that. These benchmarks exclude accounts that are too small, too new, or not active enough. They’re not measuring “what social media is like.” They’re measuring “what social media is like for accounts that meet our inclusion criteria,” which filters out most of the small businesses reading the report and trying to compare themselves.

A dentist in Phoenix with 400 Instagram followers is reading a benchmark report built from Nike, Sephora, and whoever else cleared the follower threshold. And the report doesn’t make this clear. It just says “Healthcare engagement rate: 3.70%” like that number means something to a dentist posting photos of clean teeth to 400 followers.

The marketers who share these reports know this. They know the methodology is shaky. They know two sources measuring the same thing produce numbers that are 26x apart. But they put the data in their pitch decks anyway, because their jobs depend on social media looking like it works. If the benchmarks showed that organic social is dead for most small businesses, the budgets would disappear. The agencies would close. The tools would lose subscribers.

So the reports keep getting published. The marketers keep sharing them. The businesses keep spending. And the hamster wheel keeps spinning.

The social media platforms grade their own homework.

Before I could even question whether Meta’s metrics were real, I had to spend a week just getting them to accept my money.

Apaya is a newer company. New Facebook page, new Instagram page, new Business Manager account. I connected everything, entered my credit card. Rejected. Entered another one. Rejected. Tried a third card. Tried connecting a bank account directly. Each time, Meta came back saying the payment method wasn’t valid. I deleted them, re-entered them, contacted their support team multiple times. It took about a week just to get a payment method validated on a platform that supposedly wants my advertising business.

Then once that was cleared, I built a campaign. Four ads, broad targeting, small business owners. Turned it on. Four days later: zero impressions. Back to support. Their explanation: because my account was new, Meta didn’t trust me yet. I had to run an awareness campaign first. No targeting. No specific audience. Just run ads to anyone, pay them, and prove I was a legitimate advertiser before I could actually target the people I wanted to reach.

There was no promise of anything useful coming from it. And nothing useful came from it. I handed over money for a campaign I didn’t want to run, targeting nobody in particular, just to earn the right to eventually run the campaign I actually wanted. A trust tax with no refund and no receipt.

After I’d paid that, after I’d proven myself, after I’d finally gotten to run a targeted campaign — 293 landing page views. Zero signups.

This is the on-ramp to Meta advertising for a small business in 2026. And after all of that, you’re still trusting the company that just made you jump through those hoops to honestly tell you whether the money was worth it.

When you run a Facebook ad, Meta tells you how many people saw it, how many clicked, and how well your targeting worked. Meta also collects the money. And Meta also decides what counts as a “click” and what counts as a “view.”

There’s no independent auditor sitting between you and Meta verifying that 293 real estate agents clicked your ad. You’re trusting the company that profits from your ad spend to honestly report whether the ad spend was worth it.

That might be fine if Meta had a clean record. They don’t.

In 2016, Facebook admitted it had overstated average video viewing time for roughly two years by excluding views under three seconds from a key metric. That made video look more engaging than it was. Advertisers who shifted budget toward video based on those inflated numbers got burned. By 2019, Facebook agreed to a $40 million settlement over the inflated metrics.

In 2024, a U.S. Court of Appeals allowed a major class action to proceed, alleging Meta inflated “potential reach” numbers—the figures used to sell ads—by counting duplicate accounts and bots rather than unique people. The alleged inflation: up to 400%. In January 2025, the U.S. Supreme Court declined to shut down the case. It’s still going.

In 2025, the Financial Times reported testimony from a former Meta product manager alleging that Meta overstated performance for “Shops Ads” by 17-19% by using gross rather than net sales in its reporting.

Separately, Reuters reviewed internal Meta documents in 2025 describing a large-scale ecosystem of scam and prohibited ads on Meta properties. Reuters reported that Meta internally projected roughly 10% of its 2024 revenue (about $16 billion) would come from ads tied to scams or banned goods. Internal materials estimated users were shown roughly 15 billion “higher risk” scam ads per day.

This is the platform telling me it showed my ad to 293 real estate agents.

Meanwhile, on X.

On March 8, 2026, Elon Musk, the owner of X, posted a poll to his feed:

Have you ever bought anything based on an ad on this platform?

At 1,137,043 votes: 88% said no.

The CEO of the platform, publicly amplifying a poll showing that ads on his own platform don’t work. Over a million people responded. Nearly nine out of ten said they’ve never bought anything from an ad on X.

This is a self-selected, unscientific poll. It doesn’t prove anything in a peer-reviewed sense. But it’s the owner of the platform showing you the result. He’s not hiding it. If anything, he’s leaning into it. Over a million people bothered to vote, and the number is 88%.

The click fraud nobody talks about.

Lunio, a click-fraud detection company, analyzed 2.7 billion ad clicks across 78,000 campaigns in 2026. Their finding: 8.51% of all ad clicks are fraudulent. Not suspicious. Not “low quality.” Fraudulent.

By platform: TikTok 24.20%, LinkedIn 19.88%, X 12.79%, Bing 10.32%, Meta 8.20%, Google 7.57%.

So on Meta specifically, roughly 1 in 12 clicks is fake. On my 293-click campaign, that’s about 24 clicks that were probably never real humans. But the other 269 supposedly were. And none of them signed up for a free product built specifically for them.

HUMAN Security, one of the major bot-detection firms, frames the problem in a way that hits home: the gap between a “click” and a “human loading your landing page” is getting wider. Bots now simulate engagement well enough to trigger platform “conversion” pixels, making it look like your funnel is working when it’s not.

The Imperva Bad Bot Report for 2025 puts it starkly: 51% of all internet traffic is now automated. 49% is human. Bad bots account for 37%.

Half the internet isn’t people.

The same pattern, everywhere.

When the people producing the measurement profit from the measurement, the same thing happens everywhere.

The Open Science Collaboration replicated 100 psychology studies. The originals reported statistically significant results 97% of the time. The replications? Only 36%.

A paper in the New England Journal of Medicine looked at antidepressant trials. Based on the published literature, 94% of trials appeared positive. Based on FDA’s internal assessment, which included the studies that were never published, 51% were positive. The published record was almost twice as optimistic as reality, because studies that don’t find what the funder wants don’t get published.

Merck’s Vioxx was introduced, marketed with confidence, then withdrawn after it was found to elevate cardiovascular risk. The post-market reality didn’t match the marketed certainty.

I’m not saying science is fake. I’m saying that when the people producing the measurement profit from the measurement, the measurement bends. That’s true in pharmaceuticals. It’s true in psychology. And it’s true in digital marketing, where dashboard metrics, benchmark reports, and platform-reported reach numbers all live inside the same incentive geometry. Just with faster feedback loops and fewer regulators.

So what actually works? You’re measuring the wrong game.

Here’s the thing nobody in the industry will tell you, because it doesn’t sell agency retainers or ad packages:

For most businesses, social media’s job is not to find you new customers. Its job is to not lose you customers you already almost had.

Think about how people make buying decisions today. Someone needs an HVAC contractor. They ask a neighbor, they Google it, they look at two or three websites, and somewhere in there they glance at a Facebook page or Instagram profile. Not to be sold to. Not to watch a video. Just to answer one quiet question: are these people real, and are they still in business?

An active, coherent social media presence answers that question. An abandoned one—last post from 14 months ago, inconsistent branding, stock photos that don’t match the website—raises a flag. Not a loud flag. Just enough doubt to make them click to the next result.

Nobody calls an HVAC company because of a great Instagram Reel. But people have quietly decided not to call because the profile looked dead.

That’s the game. And it’s completely different from the game of chasing followers, engagement rates, and ad conversions. It’s not about reach. It’s about credibility. It’s not about going viral. It’s about looking like you know what you’re doing and that you’re still around.

The dentist in Phoenix with 400 followers who posts consistently—showing patient transformations, answering common questions, sharing the team—that dentist is winning this game. Not because 400 people saw every post. Because when a potential patient checks them out before booking, the profile looks alive and trustworthy. That’s worth something real. It’s just not measurable in the metrics the industry sells.

Google is killing the search channel that used to fill the gap.

Up until now, the standard advice for businesses who felt let down by social media was: lean into search. Invest in SEO. Write content that ranks. Let people find you when they’re actively looking for what you sell.

That advice was correct. Search has been the most reliable customer acquisition channel I’ve seen across two companies and fifteen years. People who are looking for what you sell, find you, and buy. The intent is right. The timing is right. It works.

Except now Google is doing its best to make sure it stops working too.

SparkToro and Datos published a study in 2024: 58.5% of U.S. Google searches result in zero clicks. For every 1,000 searches, only 360 clicks reach the open web. The rest stay on Google. People get their answer from a featured snippet, an AI Overview, or a Knowledge Panel, and never visit a website.

I can see this in my own data. Our benchmarks post has 35,000 impressions in Google over the last 28 days. 47 clicks. That’s a 0.13% click-through rate. Google is showing our content to 35,000 people and sending us 47 of them. The other 34,953 get what they need without ever visiting our site.

Now add AI Overviews. Pew Research Center studied click behavior in 2025 and found that when users encounter an AI summary, they click a traditional link 8% of the time — versus 15% without one. A 2025 analysis by Authoritas, reported by The Guardian, found clickthroughs could drop up to 80% when AI Overviews appear.

seoClarity’s 2026 trend report puts numbers on the damage: desktop CTR for position 3 dropped from 4.88% to 2.47%. Position 4 dropped from 2.79% to 1.05%. The positions that used to send meaningful traffic are getting squeezed by AI summaries that sit above everything.

So the reliable fallback is getting squeezed. Search still works — it’s still the best customer acquisition channel I know — but the traffic it sends is declining as AI systems answer more queries without ever sending the user anywhere.

Which makes the credibility layer of social media more important, not less. When someone finds you through a half-answered AI overview and decides to look you up directly, what they find on your profiles matters more than it did when search traffic was abundant.

What winning actually looks like for most businesses.

Let me be specific about what the right game looks like, because “look credible” is vague.

Winning on social media for a service business, a local business, a small SaaS, or a professional service firm looks like this:

You post consistently. Not daily. Consistently. Enough that someone checking your profile in any given month sees recent activity. A post from two weeks ago is fine. A post from fourteen months ago is not.

The content looks like you. Not stock photos of handshakes and generic “tips for success” graphics. Content that reflects your actual work, your actual team, your actual customers and results. A contractor posting photos of completed jobs. A dentist sharing before-and-afters. A SaaS founder explaining a real problem their product solves. It doesn’t have to be polished. It has to be real.

The voice is consistent. Your social media sounds like your website, which sounds like how you talk when you pick up the phone. Not like four different interns wrote it in four different years. Consistency signals that there’s a coherent business behind the profiles.

You’re present on the platforms your customers check. Not all of them. The ones that matter for your audience. A B2B company probably needs LinkedIn more than TikTok. A restaurant needs Instagram more than LinkedIn. A contractor needs Facebook because that’s where homeowners in their market are. You don’t have to be everywhere. You have to be where the check happens.

That’s it. That’s the game. Not viral. Not 10,000 followers. Not a 3.70% engagement rate benchmarked against Nike. Just: present, consistent, real, and credible.

The slop industrial complex.

There’s a problem with the “just post consistently” advice in 2026, though. The content filling social media feeds is increasingly AI-generated and obviously fake. Words flying around the screen with music. Generic graphics with inspirational quotes. Faceless videos with robotic narration. The bar for content quality has dropped so low that standing out now mostly means being real rather than being polished.

I run a company that creates AI-generated social media content. So I have to be honest about this.

We built AI video generation into Apaya. Kinetic text videos. I think they’re fine for what they are. Our customers like them. But I’m under no illusion that anyone is watching a kinetic text video about an HVAC company and becoming a loyal follower. That’s not the point. The point is that it stops the scroll and makes the profile looks active, the content looks competent, and the business looks alive.

The loop the industry has created is: AI tools create content. Businesses post it. Platforms report “impressions.” The impressions get measured by the same platforms selling the advertising. The metrics go into reports. The reports justify budgets. The budgets fund more AI content. More content gets created. More “impressions” get reported.

At some point a human needs to see it and feel something about your business. The AI content buys you that moment by keeping the lights on. The real content—the job photo, the customer story, the team shot—is what makes that moment matter.

People sign up for Apaya, post AI content, get zero new followers, and come to us confused. “I’ve been posting every day for a month and nothing happened.” I understand the confusion. The industry told them that consistent posting would build an audience. It probably won’t, not in any meaningful timeframe, not without paid promotion or a genuinely remarkable product. What it will do is make their profiles look credible to the people who are already considering them. That’s the real value. It’s just not what the pitch deck promised.

What Apaya does for your social media profiles.

I want to be clear about this, because the difference matters more than it might seem from the outside.

A new wave of AI social media tools has gotten genuinely good at reading your website and generating captions from it. The writing is often solid. They pull your services, your location, your tone, and they produce posts that sound relevant to your business. That’s real progress from the template-library tools that gave every dentist the same five caption formats.

But most of them fill the image slots with memes. Stock meme formats, trending audio, reaction GIFs dropped into a business context. I find it repulsive for a professional service business. Apparently a lot of people love it, those tools have plenty of happy customers, and I’m not going to pretend otherwise.

If that’s your thing, there are tools built for it. Apaya is not one of them.

What Apaya does is take the same idea—read your website, learn your business, generate relevant content—and build it around the actual assets of your business. Your photos. Your copy. The images already on your site. What you do and who you serve. The posts look like your business because they’re built from your business. Not a meme with your logo dropped on it. Not a viral audio clip with your service category attached.

That distinction matters for the credibility game we talked about earlier. Someone checking your HVAC company’s Instagram before they call doesn’t want to see a meme. They want to see a finished job, a real team, a post that tells them you know what you’re doing. Generic content with trending formats keeps the lights on but doesn’t close the trust gap. Content that looks like you does.

And it takes thirty minutes a week instead of twenty hours. You review the posts, swap out anything that doesn’t feel right, approve what works, and it goes. You’re not outsourcing your voice. You’re spending thirty minutes a week making sure your voice stays present instead of twenty hours a week trying to maintain it manually.

While I was writing this post, a social media agency signed up for Apaya. They claim to manage social media profiles for over 600 contractors. They charge $199/month per client. They wanted to see how we do it. The company selling social media services to 600 businesses was looking at an AI tool for answers. Even the people whose job this is are looking for a better way.

Where does this leave us?

The social media industry sold most businesses the wrong game. Chase followers. Chase engagement. Chase reach. And when the numbers came back disappointing—which they almost always did—the answer was always more: more content, more frequency, more ad spend, more tools.

The right game was always simpler. Look credible. Stay present. Sound like yourself. Be findable when someone’s already decided they might want what you have.

That game is winnable. It doesn’t require going viral. It doesn’t require a $5,000/month agency or twenty hours a week of your own time. It requires showing up consistently with content that actually looks like your business.

The platforms will keep selling the wrong metrics. The benchmark reports will keep publishing numbers that exclude everyone they don’t want to count. The click fraud will keep making ad performance look better than it is. None of that changes.

But the credibility check … the moment when someone who found you some other way pulls up your Facebook page or Instagram profile to see if you’re real … that moment is yours to win or lose. It costs almost nothing to win it. And you probably don’t even know when you’re losing it.

That’s what we built Apaya for. Not to make social media into a growth engine. To make the credibility layer automatic, so it stops being a thing that falls off your to-do list when life gets busy. Thirty minutes a week. Posts that sound like you. Profiles that look alive.

The rest of the game, the ads, the viral moments, the follower counts, you can keep chasing it. I just wouldn’t hold my breath.

---

Sources

• Lunio — Invalid Traffic Rates by Ad Platform (2026)
• Variety — Facebook $40M Settlement Over Inflated Video Metrics (2019)
• Reuters — Meta Must Face Advertiser Class Action (2024)
• Reuters — U.S. Supreme Court Rebuffs Meta (2025)
• Financial Times — Meta Shops Ads Performance Testimony (2025)
• Reuters — Meta Earning Fortune From Fraudulent Ads (2025)
• HUMAN Security — Click Fraud Research
• Imperva/Thales — 2025 Bad Bot Report
• SparkToro/Datos — 2024 Zero-Click Search Study
• Pew Research Center — AI Summaries and Click Behavior (2025)
• The Guardian — AI Summaries Causing Devastating Drop in Audiences (2025)
• seoClarity — AI Search Trend Report (2026)
• Rival IQ — 2025 Social Media Industry Benchmark Report
• Socialinsider — Social Media Reach (2025)
• Sprout Social — Organic Reach
• Wired — Facebook Reach Engagement Wrong (2016)
• Open Science Collaboration — Replication of 100 Psychology Studies (2015)
• NEJM — Selective Publication of Antidepressant Trials (2008)
• PMC — Vioxx and Drug Safety (2005)
• XCancel — Elon Musk Poll Mirror