Algolia's sixth annual ecommerce search report lands with a useful headline: despite macro uncertainty, AI investment in retail search is resilient, and search remains the single most-cited digital priority — above personalization, above recommendations, above everything else retailers claim to be doing.

When brands say 'search is our top priority' in 2026, they no longer mean keyword matching and autocomplete. They mean semantic understanding, AI-powered ranking that has become a merchandising function, and the product data infrastructure to surface 'structured navy blazer for a spring wedding' without the customer typing those exact words. The operational implication: search optimization has migrated off the marketing team's desk and onto the merchandising team's. Most brands haven't reorganized accordingly. Sir John Crabstone is digging into the related silence today — the gap between what the industry is doing with AI and what it's willing to say publicly.

The report also flags a meaningful pivot in B2B ecommerce: organizations are shifting from AI expansion to AI optimization — the 'throw AI at everything' phase is ending, and 'make the AI we already have work' has begun. That's healthier. It also requires skills most fashion operators are still acquiring.

Two quiet announcements from the Google Ads Developer Blog that fashion brands should have marked in their calendars: starting April 1, Google Ads API stops accepting new Customer Match adopters — no new list-based targeting against email databases, CRMs, or loyalty program members. Then on April 30, Lookalike user lists are being deprecated — the tool that let brands build audiences resembling their best customers.

Lose both in the same month and fashion DTC brands lose two of the sharpest precision tools in their Google paid-media stack. Customer Match was how you retargeted your email list without a cookie. Lookalike was how you found new customers who behaved like your existing ones. Neither is disappearing for existing users — but new campaigns and new advertisers won't have access.

The direction is unmistakable: Google is migrating advertisers toward AI-automated audience building — Performance Max, Demand Gen — where the machine selects audiences based on conversion signals rather than explicit list inputs. Brands with rich first-party data and clean conversion tracking will navigate this well. Brands running list-dependent retargeting without a first-party data strategy are about to feel a sharp edge. Read it alongside our piece on Google Search becoming a retail interface — these API changes are part of the same architecture shift, not separate decisions.

BoF's executive memo on multibrand retail makes the structural argument plainly: wholesale-dependent brands built their distribution logic for a world where department stores were discovery infrastructure. That world is done. Multibrand retail is now fragmenting — into pure-play online, niche boutiques, brand-controlled DTC, and increasingly into AI-mediated discovery channels that don't resemble 'retail' in any traditional sense.

The uncomfortable question hanging over all of this: if discovery migrates to AI agents, to Google's AI search layer, to TikTok's recommendation engine — who is the 'multibrand retailer' in that model? Possibly the AI itself. Possibly a shopping agent becomes the new Selfridges, assembling a curated consideration set without a buyer or a physical floor. Sir John Crabstone has the full argument today on what this means for brand accountability when the agent gets it wrong. The customer relationship question is the real thing to watch.

BoF profiles Toolio's pitch for AI-assisted retail planning. Let me say the quiet part: most fashion brands are still doing buy planning in Excel. Not metaphorically. Literal spreadsheets, manually reconciled, by buyers who've been running the same process for fifteen years. Toolio's argument — that AI can close the gap between merchandising data and buy decisions — is the correct argument. Whether Toolio wins the category is a separate question; there are at least a dozen tools making similar promises right now.

Admiral Vale goes considerably deeper today on the planning room problem — the argument being that this function is where AI wins or loses retail, and most brands haven't started. Toolio is exhibit A. What neither the BoF profile nor most vendor pitches address is the precondition: AI-assisted planning on dirty data is just faster bad decisions. The tool is available. The data discipline usually isn't.

Chaussea's acquisition run — Kickers, San Marina, Pataugas — is correctly framed by FashionUnited as vertical integration, not opportunistic buying. Own the brand, own the manufacturing relationship, own the retail floor: control the margin stack at every level.

Connect this to what SMCP is doing with resale (Admiral Vale has the full analysis today — the argument is about margin recapture, not sustainability theater), and to Ba&sh's recovery story below, and a pattern emerges: French fashion is in an aggressive consolidation wave right now. Ba&sh crossed €300M. SMCP is reclaiming second-hand margin. Chaussea is buying its supply chain. Whether this is defensive (surviving the squeeze) or offensive (building scale for the next decade) is probably both — but the trajectory is clear. VC has reclassified fashion as technology. The French are responding by reclassifying themselves as logistics companies.

Ba&sh is back above €300M after two financially punishing years, and the recovery drivers are instructive: cut underperforming wholesale accounts, tighten product curation, lean into store experience. What the coverage underplays is the AI-assisted rationalization that made this possible — fewer SKUs, smarter markdown timing, cleaner data feeding better buy decisions.

The lesson for mid-market premium brands: AI-as-simplification outperforms AI-as-expansion. The brands recovering fastest right now are using AI to do less, better — not more, faster. Sir John Crabstone's Chief AI Officer analysis today lands on something related: the successful implementations aren't building AI empires inside fashion houses. They're using AI to remove noise from processes that were already broken before the model arrived.

Richard Vanoli is now MD of Debenhams. I've watched enough zombie brand revivals to recognize the script: new executive, strategic review, selective relaunch, then either a quiet exit or a genuine transformation. Debenhams has been in post-collapse IP-holding mode long enough that the interesting question isn't who's running it — it's what they're actually running.

In a world where John Lewis is already grappling with its identity as an AI discovery node, Debenhams faces the same problem with fewer assets. The name recognition is real. The trust is vestigial. Without stores to anchor the experience, you're betting on nostalgia and an email list — which is not nothing, but it is not a brand strategy for 2026.

Pepco Group — variety discount retail across Central and Eastern Europe — is printing positive H1 numbers. Nobody writes thinkpieces about value retail because it's not glamorous, but every time the macro turns uncomfortable, the value segment performs. Pepco is to European discount fashion what Primark is to the UK high street: relentless, margin-aware, and immune to the hype cycles the rest of the industry chases.

The underreported angle: discount retailers are structurally well-positioned for AI-driven optimization. Their value proposition is simple, their inventory is high-velocity, their customers are price-elastic. AI markdown optimization and demand forecasting pay off fastest in this model. If you want to see where operational AI is actually working in fashion retail, look at value, not luxury. The use cases are boring. The results are not.

Asia's largest fashion and lifestyle trade fair closed its Spring 2026 edition in good shape. CHIC matters as a calibration point: the handshakes, showrooms, and sample rails of the trade floor are where wholesale relationships form — and where you can read what buyers actually want before AI-assisted trend systems tell you what they're supposed to want.

That said, the post-show work is migrating to software at pace. Wholesale buying is becoming an AI interface, and events like CHIC are the last part of the process to hold their human format. The show floor isn't going anywhere. The follow-up already has. Admiral Vale's piece today on Southeast Asia's ecommerce surge is the companion read — video and live commerce, not search, is the infrastructure story in this region, and the trade fair format is adapting faster there than Western coverage acknowledges.

Al-Futtaim Group — the UAE conglomerate running IKEA, ACE Hardware, and a substantial fashion and beauty retail portfolio across the GCC — is deploying Algolia for search performance across its ecommerce sites. A useful data point on MENA retail tech adoption, which consistently moves faster than Western coverage suggests.

The GCC is building digital retail from a relatively recent baseline: less legacy tech debt, stronger willingness to deploy modern stack, a customer base that is genuinely mobile-native. If you want to understand where AI-powered search performs in a high-revenue, high-SKU environment, the MENA market is an underreported testing ground. We noted the region's supply chain exposure last week — the Al-Futtaim move is a reminder that the same market is investing aggressively in digital infrastructure at the same time. Operational resilience and digital buildout are happening in parallel, not sequence.

BoF is running the 'AI comes to the physical store' angle — smart fitting rooms, AI customer service, inventory visibility. Solid entry-level framing. Admiral Vale goes considerably deeper today in the piece on the store becoming an AI operating system. Parallax Pincer's analysis of the Shark Beauty and LOOKFANTASTIC AI catwalk integration is also out today — which is the live-commerce end of the same argument.

The BoF piece treats the store as a deployment surface for AI tools. The structural argument is sharper than that: physical stores generate behavioral data — dwell time, navigation patterns, product interaction, conversion by location — that no ecommerce session can replicate. Brands that instrument their stores well will have an asymmetric advantage in AI training data. The store isn't just distribution. It's data collection infrastructure. That's a fundamentally different investment thesis than 'add a smart mirror.'

Read past the celebrity angle in this FashionUnited profile. Yes, Kim and Khloé. But the operational model underneath is what actually matters: find an underserved customer with a genuine fit problem, build tight product feedback loops, and use influencer reach as a distribution mechanism — not as brand strategy.

In the AI era, the Grede model is quietly powerful because it generates proprietary fit and preference data that compounds over time. Skims' size-inclusivity feedback. Good American's extended-size fit data. These aren't marketing assets — they're training datasets. The brand with the deepest product-fit feedback loop wins the AI personalization game because they have what everyone else is trying to manufacture from scratch. Revolve is running a different version of this thesis — AI-generated product at volume — but the Grede approach starts with customer insight rather than generative output. One of these ages considerably better.

The fine jewelry market is growing because consumers are buying things that hold value rather than things that express the moment. This isn't a luxury trend — it's a behavioral signal running across price points. When purchasing power feels uncertain, customers migrate toward durable goods with real resale potential.

This is the same logic that makes SMCP's resale play smart business, and Admiral Vale's piece today makes the margin argument for it. The fine jewelry data is the demand-side confirmation: consumers are already pricing in resale value at the moment of purchase. Brands that ignore this are leaving both the first sale and the second sale on someone else's table.