From The Trenches Part 6: The Shein-ization Of Paid Social

By /

In “Reshuffle”, Sangeet Paul Choudhary describes how the use of AI in Shein’s coordination of design and production operations turned it into the fastest of fast fashion. Here’s a quick summary of how Shein’s model differs from the traditional fashion industry: it’s fast, cheap and furious in bringing popular styles to market, based on real-time data on which styles drive clicks and sales.

Traditional Fashion IndustryShein
Planning horizonOne season aheadWeekly
Design ownershipHighly specializedGig-based, freelance
CadenceOnce per season (2x/year)Constant (batch-level)
Feedback signals fromCreative / designer community, buyersAlgorithm, social media

Generative AI: The unlock that transformed creative testing

So how does the story of Shein relate to the world of Paid Social? Like high fashion, video ads before the advent of Generative AI were slow and expensive to produce. Of course, the due diligence of creative briefing and revisions, going through multiple stakeholders, has its benefits. It’s how we nail brand voice, get consistency in how we show up across channels, and results in high production quality.

However, that approach isn’t what platforms love. Especially as Meta has derived 98% of its revenue from ads in Q1-Q3 2025 (source: Q3 earnings presentation) and has announced its rapid investment in data centre infrastructure as well as its intent to improve ad microtargeting for users of all its apps, extending ad placements into Threads and WhatsApp Status as well.

What does that mean? Just like how Shein thrives on making nearly infinite styles in micro-batches, so that it can flex into demand and mass-personalization at scale, Meta wants to have nearly infinite ad variants to micro-target individual users with the combination of products, aesthetics, and creative design elements that are most likely to drive clicks and interaction from that consumer.

Traditional creative processes drive quality. What Meta wants is quantity and speed. Can you see the parallels between traditional fashion and Shein?

Before Gen AIWith Gen AI
Creative refreshAligns to design sprintsAd-hoc, near-instantaneous
Creative iterationIntentional test roadmapReal-time platform signals
MeasurementA|B testsPredictive clustering / deep learning algos

End-to-end AI ad automation: A partial turnaround

On 1st May, Mark Zuckerberg made a quote at the Stratechery podcast that fundamentally threatened the value proposition of agencies and Paid Social campaign managers:

“… make it so that any business that basically wants to achieve some business outcome can just come to us, not have to produce any content, not have to know anything about their customers. Can just say, ‘Here’s the business outcome that I want, here’s what I’m willing to pay, I’m going to connect you to my bank account, I will pay you for as many business outcomes as you can achieve…

…Over the last 5 to 10 years, we’ve basically gotten to the point where we effectively discourage businesses from trying to limit the targeting… basically, we believe at this point that we are just better at finding the people who are going to resonate with your product than you are…

… But there’s still the creative piece, which is basically businesses come to us and they have a sense of what their message is or what their video is or their image… we just make it for them… in general, we’re going to get to a point where you’re a business, you come to us, you tell us what your objective is, you connect to your bank account, you don’t need any creative, you don’t need any targeting demographic, you don’t need any measurement, except to be able to read the results we spit out.”

Can you see how that wipes out the perceived skill premium of any Paid Social professional? Targeting, creative strategy, bidding strategy, and measurement are the four main technical pillars that require an experienced, data-savvy professional to drive Paid Social ROAS (Return on Ad Spend). And with this announcement, the ad community wasted no time calling out everything that was problematic about a complete “black box” model. (See linked article from The Verge, with its comments, in the References section).

Of course, as my personal experience and the comments point out, full ad automation has been portended for years but never come close to taking over the majority of Google or Meta ad spend because without levers, it’s very hard to achieve sustainable ROAS. Fully automated bidding might help small businesses to run ads with minimal resources, but for any organization with dedicated Paid Social headcount, testing into the optimal campaign settings and creative, and ensuring brand voice is preserved, are all paramount.

In the Q3 call on 29th October, Zuckerberg hasn’t pushed the matter farther. In fact, he quantified the annual run rate of end-to-end automated ad placement spend (revenue to Meta) at $60 billion, which is only 32% of Meta’s trailing 4 quarters’ ad revenue, far from a complete transition.

What’s above and below the algorithm now?

I feared that the rollout of fully automated ads might send Paid Social campaign managers entirely below-the-algorithm. Thankfully, the situation painted by Zuckerberg two quarters later isn’t as absolutist as that. But it remains that the value which campaign management teams can drive through in-platform optimization is subject to the vagaries of the levers that Meta and Google give to us.

Therefore, the only guaranteed way for an in-house marketer to have leverage over outcomes is through capital allocation strategy. Platforms may grant or deny levers as they develop their AI features, but the power of the advertiser is in deciding where and how much to spend.

As you can imagine, the “Shein-ization” of Paid Social has reshaped our jobs substantially. Before, creative taxonomy and the ability to measure statistical significance, as well as an intentional A|B test roadmap, were the key to designing winning creatives. Now, with Gen AI tools for video creation built right into the Meta Ads platform, the platform is leaning towards rewarding volume over intentionality, and everybody needs to be prepared to play along. Like Shein, Meta’s underlying principle is that production at volume and speed generate more gains than careful testing and iteration.

What does this mean for campaign management and analytics professionals? What’s within our control are measurement technology, quantifying incremental P&L impact accurately, and strategically flexing the channel mix. At the operational level, we’ll still need strong Specialists to scale creative variants using in-platform AI while ring-fencing risks to brand voice and brand safety. We’ll need to branch out from analysing last-touch ROAS funnels to triangulating that against Marketing Mix Modelling and incrementality test results.

In short, fluency in multiple channels and multiple methodologies has become the sweet spot where we can drive value above the algorithm.  Performance marketers and marketing analysts aren’t going away, but we need to consider where the value is going in our ecosystem and chase it.

And in Part 7 next week, I’m going to talk about the day I woke up at 5 am for a webinar at exec ed… to learn that the A|B testing approach I had done for ten years might be going the way of the dodo… and what’s in store next. Stay tuned!

References:

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Scroll to Top