What's actually working in Meta ads for ecommerce in 2026?

I've been running Meta ads for ecommerce brands for over a decade. In that time the platform has rebuilt itself more times than I can count. Every rebuild, the same thing happens: a wave of brands keep running the old playbook, wonder why performance is sliding, and blame the platform. Meanwhile a smaller group adjusts to how the platform actually works now, and pulls ahead.

We're in one of those moments again. The brands acquiring profitably on Meta in 2026 are doing a specific set of things, and most of them are the opposite of what worked three years ago. Here's what I'm actually seeing across the accounts I run and the hundreds of brands I've diagnosed.

The platform makes the targeting decisions now. Your job is to feed it.

For years, the craft of Meta advertising was audience targeting. You found the right interest stacks, the right lookalikes, the right exclusions, and you pointed your ads at them. The media buyer who knew the audiences won.

That era is over. Meta's machine learning is now better at finding your customer than any manual targeting you can build. The platform wants broad targeting and it wants to make the audience decisions itself. When you over-segment, you're fighting the algorithm and starving it of the data it needs to optimise.

The job has shifted. You're no longer choosing the audience. You're feeding the algorithm enough creative variety and clean signal that it can find the audience for you. The brands still spending their energy on audience construction are optimising the wrong thing. The brands winning are spending that energy on creative volume and signal quality instead.

This is the single biggest mindset shift, and most brands haven't made it.

Creative volume is the new media buying

If targeting is mostly automated, the lever you actually control is creative. And the brands winning on Meta in 2026 are producing creative at a volume that would have seemed absurd in 2022.

I'm talking 15 to 30 distinct concepts a month, not variations of one concept, but genuinely different angles, formats, hooks, and messages. The algorithm needs that variety to do its job. It tests, finds the winners, and scales them. If you only give it four concepts a month, you've capped how well it can perform before the auction even starts.

Most brands can't produce at this volume because their creative model wasn't built for it. They have a quarterly hero shoot and a designer making a handful of ad variations. That model is the constraint. The fix isn't more budget on ads, it's restructuring how creative gets made: faster cycles, lower-fidelity concepts, more angles, less precious polish.

When I take on an ecommerce brand and performance has plateaued, the creative production model is the first place I look. It's almost always under-built for what the platform now demands.

Direct response creative beats brand creative on cold traffic

Here's a pattern I see constantly. A brand invests in a beautiful, editorial, brand-led campaign. It looks expensive. It looks like the brand. And it underperforms on cold traffic, badly.

The creative winning cold acquisition in 2026 looks more like 2018 direct response than 2023 brand content. Hooks that land in the first two seconds. Problem-led messaging. Product demonstrations. Real customer testimonials. Comparison angles. Creative that does a job, not creative that wins design awards.

The polished brand campaigns still have a place, mostly in retargeting and warm audiences, where the buyer already knows you. But on cold traffic, where you're interrupting someone who has never heard of you, direct response wins. Every time.

If your cold campaigns are underperforming, look at your top creative. If it looks like a brand campaign, that's your answer.

Signal quality is doing quiet damage to most accounts

This is the one most brands don't see, because it's invisible at the campaign level. Meta's optimisation is only as good as the data flowing back into it. If your conversion tracking is incomplete, your events are misfiring, or your data connection is weak, the algorithm is optimising on bad information.

I run a full check on signal quality in every diagnostic. Conversions API setup, event match quality, deduplication, the integrity of the data flowing back to the platform. More often than not, I find gaps. The brand has been running ads for months on a tracking setup that's quietly feeding the algorithm partial data, and then wondering why performance won't scale.

You can have great creative and the right account structure and still underperform if the signal is broken. It's the foundation, and it's the least glamorous part of the work, which is exactly why it gets neglected.

Account structure has gotten simpler, and that's correct

A few years ago, sophisticated Meta accounts had dozens of ad sets, intricate segmentation, elaborate testing structures. Sophistication looked like complexity.

In 2026, sophistication looks like simplicity. Consolidated campaigns. Fewer, broader ad sets. Letting the algorithm pool data and optimise across a larger pot rather than slicing budget into fragments. When you over-fragment the account, every ad set is starved of the conversion volume it needs to exit the learning phase and optimise properly.

The brands winning have simple, consolidated account structures and put their energy into creative and signal. The brands struggling often have beautiful, complex account architecture that's actively working against them.

Ce que cela signifie pour votre marque

If your Meta performance has plateaued or started sliding, the instinct is usually to look for a new tactic, a new audience, a new campaign type. That's almost never where the problem is.

The real questions are:

• Are you still spending energy on manual targeting the algorithm wants to handle itself?

• Is your creative production volume anywhere near what the platform now needs?

• Is your cold creative actually direct response, or has it drifted into brand mode?

• Is your signal clean, or is the algorithm optimising on partial data?

• Is your account structure simple enough to let the algorithm pool and optimise?

Most brands that come to me with a Meta problem are doing two or three of these wrong. The fix is rarely exotic. It's getting the fundamentals right for how the platform works in 2026, not how it worked in 2022.

Meta is not harder than it used to be. It's different. The brands that adjust to the difference win. The brands that keep running the old playbook keep blaming the platform. The choice is genuinely that simple.

If you want a senior diagnostic of where your Meta performance is actually constrained, you can book a discovery call with me here. 15 minutes, no pitch, no pressure.