GA4 missed 66% of one retailer's holiday orders, Meta replaced its ad auction with LLMs, and EU customs seized a record 196M counterfeit items.
GA4 could be hiding two-thirds of your holiday sales data — and Meta just rebuilt every ad auction with LLM technology. Two urgent reads before Black Friday.
In this episode:
The Digital Commerce Daily is a concise briefing for ecommerce, retail media and digital commerce professionals, hosted by Marco & Klara. New episode every day.
Enjoying the show? A quick rating or review on Apple Podcasts or Spotify helps other marketers find us.
Also from Marco & Klara: The FMCG Marketing Daily · The Health Marketing Daily
Your 8-minute briefing on the platforms, players, and profits driving digital commerce today. Hosted by Marco and Klara.
Marco: Three stories worth clearing your schedule for this morning.
Klara: We've got a GA4 tracking problem that could be quietly wrecking your holiday planning, a fundamental rebuild of how Meta ranks every single ad, and a record-breaking wave of counterfeit goods hitting EU customs.
Marco: Big implications across all three — for brands, for marketplaces, for anyone running paid media.
Klara: Let's go.
Marco: An audit has found that Google's main analytics tool is missing up to two-thirds of some retailers' actual orders — and Black Friday is weeks away.
Klara: Let's back up for a second. GA4 is Google Analytics 4 — the tool most online retailers use to track what's happening on their website. Which products people look at, which ads are working, how many orders come in. It's basically the scoreboard for your whole business. Most retailers treat it as the truth.
Marco: The problem is, it might not be. An audit across eight ecommerce brands found that one retailer's GA4 was missing 66% of their holiday orders. Not a rounding error. Not a small gap. Two-thirds of their sales — gone from the data.
Klara: And this wasn't a one-off. The audit suggests it's a systemic issue — meaning it could be happening quietly across many stores, not just this one. The expert behind the audit is now urging retailers to run checks before peak season starts.
Marco: Here's why this is so dangerous. If your analytics are only recording one in three orders, then every decision you made last holiday season — which ads to spend more on, which channels to cut, which products to push — was based on completely wrong numbers. You were flying blind and didn't know it.
Klara: I'd add some nuance there, though. Some businesses do cross-reference GA4 with their actual order management systems or their payment data. If you're doing that, you'd have caught this. The real risk is for anyone treating GA4 as their only source of truth.
Marco: Which is a lot of people. So before you set your Black Friday budgets — go and check your GA4 data against your actual order numbers. Right now. Not after peak season.
Klara: The time to find a hole in your roof is before it rains.
Marco: Now, let's move on to our next story.
Marco: Meta has completely rebuilt the engine that decides which ads you see — and it's now powered by the same technology behind ChatGPT.
Klara: So let's put that in context. Every time someone scrolls through Instagram or Facebook, there's an auction happening in the background. Hundreds of ads are competing for that one slot, and a ranking system decides who wins. Until now, that system worked on fairly rigid rules — basic signals like how often an ad got clicked before, or how much the advertiser was willing to pay.
Marco: What Meta has done is replace those rigid rules with something far more sophisticated. Their new ranking system uses large language models — the same type of AI that powers modern chatbots — to score over a billion ads every single second. That's not a feature tweak. That's rebuilding the whole foundation. They've also upgraded two other systems underneath this: Andromeda, which retrieves the right ads to consider, and GEM, which predicts how users will engage with them.
Klara: And this helps explain something. Meta's revenue grew 26% last quarter, which is a lot for a company that size. Now we know part of the reason — a smarter auction means better matches between ads and people, which means advertisers get better results, which means they spend more.
Marco: For ecommerce brands specifically, here's what changes. A language-model-based system can actually read and understand your ad — the copy, the product context, the imagery description. That's something a rule-based system couldn't do. In theory, that rewards well-structured, high-quality creative and penalises anything lazy or generic.
Klara: Though I'd be cautious about assuming quality automatically wins. Relevance and budget still matter enormously. A well-funded mediocre ad can still beat a brilliant underfunded one.
Marco: Fair, but it shifts the playing field a bit. Especially heading into peak season — if you haven't looked at your Meta creative quality recently, this is the moment.
Klara: Right, let's keep moving — our third story is up next.
Marco: EU customs agencies seized a record 196 million counterfeit items last year — and that number has real consequences for anyone running or selling on a marketplace in Europe.
Klara: To understand why this matters, you need to know what's been happening in the background. For years, vast quantities of fake goods — counterfeit clothing, electronics, luxury items — have flowed into Europe through cross-border parcels, often from China. Customs agencies have been fighting this, but the scale was always overwhelming.
Marco: Last year, that fight produced a record result. 196 million suspected counterfeit items intercepted. That's not a marginal improvement — that's a signal that enforcement is being taken seriously at a level we haven't seen before. And it lands right alongside everything else the EU is doing: cracking down on Temu, closing the loophole that let cheap parcels arrive duty-free, tightening scrutiny on platforms that use third-party sellers.
Klara: What connects all of this is the DSA — the Digital Services Act. That's the EU's rulebook for online platforms, and it puts direct legal responsibility on marketplaces to police what's being sold through them. This seizure data is exactly the kind of evidence regulators will use when they go after platforms that aren't doing enough.
Marco: For legitimate brands selling into Europe, this is actually a good development. Every counterfeit item seized is one that didn't undercut a real product. Regulatory pressure on fakes raises the cost and risk for bad actors — but only when platforms enforce their own rules properly.
Klara: And that's where I'd add caution. Record seizures at the border are one thing. But plenty of counterfeits still make it through. The burden on marketplace operators to verify who's selling what is enormous, and many simply aren't set up for it. Customs doing record work doesn't let platforms off the hook — if anything, it raises the bar for what regulators will now expect from them.
Marco: The era of looking the other way is closing. If you operate a marketplace, or sell on one, the compliance question isn't something you can defer anymore.
Marco: The single biggest idea from today — the data you trust to run your business might be lying to you, the ads you run are being judged by systems you can't see, and the products competing with yours might be fake.
Klara: So audit one thing this week — your tracking, your ad performance, or your brand protection. Pick one. Then actually look at the numbers.