Retail and FMCG in 2026: channels multiply, margins do not
Six channels, four versions of the same customer, and a reseller network paid from a spreadsheet. What is changing in commerce platforms, affiliate structures and personalization, judged by whether the customer came back.
June 30, 2026 · 5 min read · Product Consultant · Sector landscape 2026
Where the sector is
A mid-sized FMCG brand now sells through general trade, modern trade, quick commerce, marketplaces, its own direct channel and a growing set of affiliates and resellers. Each channel has its own data, its own pricing pressure and its own returns. The brand’s view of its customer is assembled afterwards, if at all, from exports.
In India the picture has an extra layer. General trade, the network of small independent stores served through distributors, still carries most of the volume for many categories, and the brand’s visibility into it ends at the distributor’s invoice. Everything modern about the channel mix sits on top of a base the brand cannot see into.
Retail has the same problem from the other side. A fashion retailer runs stores, a site, an app and marketplace listings, and the same customer appears in all four as a different record. Personalization is promised in every vendor deck and delivered in the email subject line.
Underneath the channel complexity sits a hard economic fact. Acquiring a customer costs more every year and keeping one is where the margin is. The companies doing well in the sector are not the ones with the most channels. They are the ones that know which customers, which partners and which products actually produce return, and act on it faster than the quarter allows.
What is changing
Direct channels are becoming platforms, not storefronts
The first generation of direct-to-consumer was a hosted storefront and a payment gateway. Brands that have outgrown that are now building commerce platforms that hold the things a storefront does not: affiliate and reseller structures with commission logic, group buying, multi-brand inventory, and promotions that run across channels rather than in one. This is product engineering rather than configuration, and the timelines look different. Platforms of this type can support market share growth of up to 37% in target segments and reduce retention cost by up to 26%. The engagement we ran for a mid-scale retail and FMCG brand was built on exactly this pattern.
Affiliate and distributor structures are being formalized
Resellers, influencers, micro-distributors and society-level sellers used to be managed in a messaging group and a spreadsheet of who is owed what. The shift is toward treating them as a governed network: onboarding, a commission engine with rules that can be audited, payout automation, and performance visibility per partner. The mechanics are close to the partner platforms we build for energy and construction, which is not a coincidence.
Personalization is being judged on repeat purchase
The measure that matters is not click-through on a recommendation. It is whether the customer came back. Retailers are shifting personalization work from the homepage to the full journey: discovery, basket, post-purchase, and the timing of the next contact. Platforms of this type can lift engagement by up to 34% and repeat purchases by up to 19%. In our work with a global fashion retailer, most of the effort behind those ranges went into data unification, not model sophistication.
Returns and inventory data are becoming a decision input
Returns were a cost center with a report. They are becoming a signal: which products, from which channel, for which stated reason, and what it implies for sizing, description, or the partner who sold it. The same applies to inventory across channels, where the brand often discovers it has stock in one place and stockouts in another only when a customer complains.
What breaks in practice
Identity. The same customer is four records, and no amount of personalization works until they are one. Identity resolution is unglamorous and comes first.
Commission disputes. A reseller who believes they were underpaid stops selling. The commission engine has to show its working: which sale, which rule, which rate, which payout. If the logic lives in a spreadsheet the disputes never end.
Channel conflict in the data. The marketplace will not give you customer-level data. The distributor will not give you sell-out. Every model has holes, and the honest version of the dashboard shows where they are.
Promotion sprawl. Campaigns configured per channel with no central view produce a customer who is offered three different prices for the same product in the same week. A promotions engine that governs across channels is the fix, and it is a harder build than it looks.
Inventory that is true in one place. The storefront says in stock, the warehouse system says otherwise, and the marketplace listing was updated yesterday. Cross-channel inventory is a synchronization problem with commercial consequences, and the brands that get it right treat stock as one ledger with channel views, not as a number copied between systems on a schedule.
Where we would start
For an FMCG brand, the affiliate and reseller layer. It is where the money leaks, it is where partners feel the brand directly, and building it well produces sell-out data the brand has never had. Start with onboarding and commission settlement for one partner type, with the audit view built in.
For a retailer, identity resolution across channels, then one personalization flow with a repeat-purchase measure attached. Not a recommendation engine on the homepage. One journey, tracked properly.
In both cases the first release should reconcile to finance. If the platform’s view of sales and payouts does not match the ledger, nobody will trust anything else it says.
What to watch
Quick commerce is changing pack sizes, pricing and the meaning of availability, and the brands that can see sell-out by channel daily will adapt faster than the ones reading monthly distributor reports. Watch marketplace data policies, which shift with little notice and change what a customer model can know. And be skeptical of AI personalization offers that cannot be tested against a holdout group. If the vendor cannot tell you what the customers who did not see the recommendation did, they do not know if it works either.
Read more at /industries/retail-fmcg/, or see the case studies: A Mid-Scale Retail & FMCG Brand and A Global Fashion Retailer.