Our 2026 DTC Predictions: AI Commerce, AppLovin, and the Death of the Brand Website (Decoded)

Late last year I watched a brand I admire freeze their entire Q1 ad budget because a podcast told them the brand website was dead.

The founder had heard a smart operator say he pays Shopify around A$60k a month for a store he thinks will be a relic inside 20 years. So he reasoned, why pour money into a channel that's about to be replaced? He paused scaling, sat on his hands for two months, and waited for the future to arrive.

It didn't. His competitors kept acquiring. He lost share he's still clawing back.

That's the trap with prediction season. The takes are usually directionally right and the timelines are almost always wrong, and founders blow themselves up acting on a 20-year forecast as if it lands next Tuesday.

So here's my take on the loudest 2026 DTC predictions doing the rounds right now. I run a Meta-led agency, so I'm filtering each one through a single question: does this change what I'd do with a 6 or 7-figure brand's money in the next 12 months, or is it a stage demo that's fun to talk about and useless to act on?

Here's my take on each.

Prediction 1: The brand website becomes a relic

My verdict: real direction, wrong timeline. Plan for it, don't act on it yet.

The argument goes that AI agents will do your shopping for you. You won't browse ridge.com or some homewares store, you'll tell an assistant "I need a new water bottle" and it'll go fetch, compare reviews across the whole internet, and come back with three options. Some people are calling the end state a "Mi store" - a personalised feed built just for you, not a brand's storefront.

I think the direction is right. Conversational, agent-led discovery is coming, and it'll eat a slice of the crude click-and-scroll experience we have now.

But here's the thing. The same operators predicting this also admitted, on the same podcast, that they'd never let an agent just pick their shirt for them. They want to see the thing. They want the buying journey. A big chunk of shopping is a hobby, not a chore, and you don't outsource your hobby. The "send my agent to buy toilet paper" use case is real and dull. The "send my agent to buy a gift, or the jeans I'll obsess over for a week" use case is the one most people will keep doing themselves, and that's most of discretionary commerce.

So what actually happens in 2026? Roughly nothing you can bank on. To put this into perspective, the mobile phone took about 15 years to go from novelty to the thing we do everything on. Agentic commerce is at year zero of that curve. If you torch your acquisition engine now to prepare for a buying experience that isn't built yet, you're the founder I described up top.

What I'd propose: keep your site sharp, keep your feed clean and structured (agents will read product data, so your titles and descriptions matter more, not less), and keep acquiring customers the way that works today. You can't sell to an agent's owner if you went broke waiting for the agent.

Prediction 2: AI agents reshape discovery and "the Mi store" arrives

My verdict: partly real, and the part that's real helps you now.

Strip the sci-fi off the "Mi store" idea and there's a genuinely useful core: AI gets you closer to a one-to-one selling conversation at scale. One operator made the point that if every customer could talk to the founder directly, conversion would be off the charts. E-commerce is just a crude, scalable stand-in for that. AI narrows the gap.

I believe the near-term version of this isn't a new storefront, it's better matching on the channels you already run. Personalised creative. Showing the right angle to the right person. Feeding the ad algorithms cleaner first-party signal so they find your buyers faster.

That's not a 2030 fantasy. That's a 2026 creative and data job, and it's one of the few bits of this whole conversation you can act on Monday morning.

Here's the honest bit though. "Personalised AI everything" is also the most over-promised line in the deck. The image and video tools genuinely jumped a level last year - keeping a product looking like itself across a generated scene used to be impossible and now it mostly isn't. But "the algorithm will read your whole life and build a bespoke store" is still a keynote slide. Treat the tooling as a real lever for making more, better, more relevant ads. Treat the rest as a roadmap, not a launch.

Prediction 3: AppLovin becomes a real third channel

My verdict: real, and the smart money is already testing it.

The prediction I'd weight most heavily for next year is the boring one. Operators are calling it a "massive success" if AppLovin captures somewhere between 5 and 10% of ad wallet share. Not 50%. A single-digit slice that's incremental to Meta and Google.

That's the right frame, and it's why I take it seriously. Nobody serious is saying drop Meta for AppLovin. They're saying there's a third meaningful auction opening up, and the brands who learn it early get the same edge the early Facebook advertisers got - cheaper inventory before everyone else piles in.

I built a chunk of my own thinking on exactly that pattern. The brands that win during a platform shift are the ones willing to learn the new thing while the big, comfortable players think it's beneath them.

What I'd propose for a 6-7 figure brand: don't go all-in, ring-fence a test budget. A reasonable starting frame is carving out maybe 5 to 10% of your paid social spend to learn a new channel properly, the same way you'd never let your whole account ride on one untested concept. So if you're spending ~A$80k a month on Meta, that's a A$4-8k monthly learning budget, not a bet-the-quarter swing. Judge it on incremental new customers, not on a vanity ROAS that double-counts demand you already had. If it can't clear your real acquisition cost on net-new buyers, park it and revisit. If it can, you've found cheaper customers before your competitors did.

The one caveat I'd flag: a new channel always looks worse than Meta on day one, because Meta has years of your data and the new auction has none. Give it a real runway before you judge it. The brands that bailed on TikTok or Pinterest in week two on a bad ROAS line were measuring a learning phase as if it were steady state.

This is the one I'd actually move on in 2026.

Prediction 4: Meta goes hard after new customers

My verdict: real, and it quietly changes how you budget.

Here's one that got less airtime but matters more to your P&L than any agent fantasy. The take is that Meta has heard the feedback that it's become too good at finding people who were going to buy anyway - your existing customers, your 30-day repeat buyers - and is pushing harder toward genuine new-customer acquisition.

Why does this matter? Because you can't reacquire a repeat customer at break-even and call it growth. Growth needs net-new buyers, full stop. If Meta tilts its optimisation toward new customers, that's good for brands that have been quietly inflating their blended ROAS on the backs of people who'd have come back regardless.

I believe this is the prediction most likely to expose weak accounts. If a big slice of your "performance" has secretly been retargeting and brand-search picking up demand you already created, a shift toward new-customer acquisition will make your numbers look worse before they look honest.

There's a reframe in here I really like. One operator described marketing spend as an asset on the balance sheet, not a cost on the P&L - he said he'd far rather have spent A$50 million acquiring customers over the years than zero. The dollars you spend acquiring a customer today are buying you a repeat-purchase stream for years. If you only measure the first order, you'll always underspend and you'll always feel poor doing it.

What this means: separate your new-customer numbers from your blended numbers now, before the platform forces the issue. If you don't know your new-customer ROAS and your new-customer acquisition cost as distinct figures, you're flying on instruments that are about to get recalibrated.

Prediction 5: 2026 is a "reformation year"

My verdict: real, but it's a vibe, not a tactic.

The framing a lot of operators are using is that 2026 is a reset - businesses go under, the AI froth gets repriced, and the brands that stayed profitable through a rough 2025 come out the other side stronger. The comparison being thrown around is the dot-com bust: the bubble popped, the bad businesses died, and the survivors got to be the next decade's winners.

I think the spirit is right and the specifics are unknowable. Will there be a market wobble that dents the top 10% of households who drive most discretionary spend? Maybe. Is the consumer more deal-motivated than they've been in years? Yes, that part's visibly true heading into the back half.

But "reformation year" is a mood, not a media plan. You can't optimise against a macro feeling. The useful version of this prediction is narrow: be profitable, know your numbers, and don't bet the business on things you don't control.

The best line I heard on the whole macro circus was the river-rafting one. You can't be a good guide if you don't know what the river's doing - but you also can't pretend you control the river. Be aware, don't be paralysed, and spend your energy on the three or four levers you actually own. For most DTC brands those levers are the same as ever: your offer, your creative, your acquisition economics, and the value you give the customer for their money.

What I'd actually do with this for 2026

Strip away the keynote theatre and the real list is short and a bit boring.

• Don't pause acquisition to prepare for a future buying experience that doesn't exist yet. The website isn't dead in 2026. Acting like it is, is how you hand share to a competitor.
• Use the AI creative tooling that's genuinely good now to make more relevant ads and feed the algorithms cleaner signal. That's the near-term cash value of the whole "AI commerce" story.
• Ring-fence a real test budget for a third channel like AppLovin. Single-digit wallet share, judged on incremental new customers, is a sober bet worth making early.
• Split your new-customer economics out from your blended numbers before Meta's shift toward new-customer acquisition makes your reporting lie to you.
• Treat ad spend as an asset, not a cost - which mostly means knowing your real payback window so you stop underspending out of fear.

The thread running through all of it: the predictions that are worth acting on in 2026 are the unglamorous, account-level ones. The ones that go viral are the 20-year ones, and 20-year predictions are terrible budget inputs.

If you're heading into 2026 planning and you genuinely can't tell, looking at your own account, which of your "performance" is net-new growth and which is demand you already paid to create once, that's the gap worth closing first. A Signal/Noise Audit walks your account, creative history and unit economics specifically to draw that line - so your 2026 plan is built on what's actually moving the numbers, not on what played well on a podcast.

Which of these predictions would change what you do next year - and which one were you about to act on a decade too early?