I just wrapped a 3-month Meta Ads experiment for Merged Media.
$16,240 invested. Over 100 creatives uploaded. Hundreds of hours testing what works in 2026.
Here's what I learned: Meta Ads have changed so drastically in the past year that if you're still running campaigns the way you did in 2023, you're burning money.
Let me walk you through what actually worked, what bombed spectacularly, and the new reality of Facebook advertising that most agencies still haven't figured out.
Remember when we used to spend hours dissecting customer avatars?
Figuring out if your ideal customer's favorite food is sushi or pizza? Whether they read at night or in the morning? If they prefer working out at 5 AM or 5 PM?
I spent years doing this. Building detailed personas. Creating separate ad sets for different demographic segments. Testing interests like "entrepreneurship" versus "small business ownership."
All of that? Gone.
The algorithm has it figured out for us now. And it does it better than we ever could.
Here's how I ran this experiment.
Any time I had a fresh content idea, I pulled out my iPhone and recorded a quick video. No fancy production. No script. Just raw, authentic content shot wherever I happened to be.
Airport lounge. Hotel room. My office. The rugby field. Didn't matter.
I'd throw the video at Opus Clip for captions, push it into my campaign, and let the algorithm do its thing.
Out of 100+ videos I uploaded, two of them generated 90% of the leads.
Let that sink in. Two videos out of over a hundred.
That's not a failure. That's how the algorithm works now. It tests everything, figures out what resonates, and doubles down on winners.
Let me tell you about one of my biggest flops.
Last month I was on a layover in Zurich. Had a few hours to kill. Walked up to the observation deck overlooking the flight line and had an idea.
I recorded a video that started with 3 seconds of a plane taking off - engines roaring, wheels leaving the tarmac, that dramatic moment right before liftoff.
Then I flipped the camera to myself and asked: "Does your business have a flight plan for 2026?"
In theory? Brilliant ad. Great hook. Strong metaphor. Visually compelling.
The algorithm loved it. Thousands of impressions. High retention rate. People were watching the whole thing.
The result? Not a single lead.
Here's what I think happened: The algo was showing it to people who liked airplanes. Or people who were waiting for something exciting to happen in the video. They stayed for the plane takeoff (first 3 seconds), which told Meta "this is engaging content."
But engagement doesn't equal conversion.
They watched for the wrong reason. The hook grabbed attention, but it didn't qualify the right audience. So Meta kept showing it to people who liked planes, not people who needed marketing help.
The lesson: The hook matters, but a clear hook with a direct path to an end result matters 10 times more.
You can have the perfect attention-grabber. But if it doesn't immediately signal what you're offering and who it's for, you're just entertaining people for free while Meta charges you for the privilege.
So what about those two videos that generated 90% of the leads?
Video 1: Me walking down the street talking about how I started my business and the journey with my first client. No desk. No setup. Just me walking and talking about the early days - going door to door, landing that first client.
Video 2: Me walking down the street again, but this time calling out how dodgy SEO agencies are stealing business owners' money with outdated 2020 tactics.
Both different messages. Both shot the same way. Both absolute winners.
Why did these work when the Zurich airport video bombed?
Authentic setting. Walking down the street feels real. It doesn't scream "ad."
Clear value proposition. Within 3 seconds, you knew if this was for you or not.
Qualified hook. People who stayed weren't staying for visual spectacle. They had the exact problem I was talking about.
Direct relevance. No metaphors to decode. No creative concepts. Just straight talk about real business problems.
The Zurich video? It entertained people. The walking videos? They solved problems.
Meta launched the Andromeda update and fundamentally changed how ad delivery works.
The shift: Meta's ad delivery is now AI-first. The algorithm scans every element of your ad - image, video, text, engagement patterns - to determine who it resonates with.
This has massive implications for how you should be building campaigns.
Broad targeting now outperforms detailed targeting and Lookalike Audiences.
I know. This feels wrong. We spent years perfecting our LAL audiences and detailed interest targeting.
But here's the reality: Lookalike Audiences are now largely redundant. Meta's AI finds those people within broad audiences anyway. You're just limiting the algorithm's ability to discover who actually responds to your ads.
Stop over-targeting. Give the algorithm room to work.
Here's the setup that's generating results for us:
One broad ad set. That's it.
Advantage+ audience targeting for CA/US/UK (or whatever regions you serve).
No age restrictions. No gender filters. No detailed interests. Just broad geographic targeting and let Meta handle the rest.
Don't run multiple ad sets with different targeting variations. This is what everyone still does, and it's killing their performance.
Why? Because you're cluttering the signal. Meta needs clean data to optimize. When you have 5 different ad sets all competing for similar audiences, the algorithm can't figure out what's actually working.
Fewer ad sets = cleaner data = better optimization.
I know this feels counterintuitive. We're marketers. We want control. We want to segment and test and optimize manually.
But the algorithm is legitimately smarter than you at this now. Let it do its job.
Stop obsessing over targeting. Start focusing on creative volume.
Here's the play: Upload as many different creatives as you can produce authentically. Different hooks. Different angles. Different formats.
The algorithm tests them all and figures out what resonates.
My 100+ videos? Most got minimal spend. Meta tested them, decided they weren't winners, and moved the budget to the ones that were crushing it.
That's exactly what you want.
Your job isn't to pick the winners. Your job is to create enough options that the algorithm has winners to choose from.
Think of it like batting average. Even the best hitters in baseball fail 70% of the time. But they keep swinging.
Same with Meta Ads creative. Most of what you create won't be a home run. But when you create enough, the algorithm finds the ones that connect and scales them automatically.
The Zurich airport video? Swing and a miss. Cost me $300 in ad spend and generated zero leads.
But those two walking videos? Home runs. And Meta found them, scaled them, and generated 90% of my leads.
Feed the beast. Let it choose the winners. Get out of the way.
Here's something that surprised me: the low-production iPhone videos outperformed anything polished.
People don't want to see ads anymore. They want to see real people talking about real problems.
When I pulled out my iPhone and recorded myself talking directly to camera - messy desk in the background, natural lighting, no script - those converted better than anything we shot with professional equipment.
Why?
Because they didn't look like ads. They looked like content. Like someone sharing genuine insight, not trying to sell you something.
The more your ad looks like organic content, the better it performs. Meta's users have banner blindness for anything that screams "professional advertisement."
But a genuine-looking video from someone solving a problem they care about? They'll stop and watch.
Meta Ads in 2026 is a completely different game than it was even a year ago.
The algorithm is smarter than you. Let it do its job by giving it:
Your job isn't to outsmart Meta's AI. Your job is to create content that connects with real people and let the algorithm find those people for you.
The businesses still trying to manually control every aspect of their targeting are fighting yesterday's war with yesterday's weapons.
The businesses winning? They're feeding great creative to an incredibly smart algorithm and getting out of the way.
Want to see how this could work for your business? Merged Media helps service-based businesses implement modern Meta Ads strategies that actually generate leads. Book a strategy call here.
Co Founder of Merged Media & Co Owner of Chatello.ai
He is the co-founder of Merged Media, an award-winning digital marketing agency specializing in AI-powered strategies. A sought-after speaker and educator, Jason teaches "Leading Change in Digital Tourism" at George Brown College and has presented at major events like Affiliate World and AdWorld. From his unique journey as a Japanese rockstar to building a successful agency, Jason combines creativity and innovation to help businesses worldwide scale their growth and achieve transformative marketing results.