What We Heard at ShopTalk Fall 2026

What we heard: 6 recurring product data problems

Six problems came up again and again, whatever the company’s size or sector.

The spreadsheet is still the system of record

Many of the teams we met have no PIM at all. Product data lives in spreadsheets, shared drives, the ERP or the commerce platform.

Several brands use their commerce platform as the source of truth and told us it stops working as they grow. Every new channel means exporting data, editing it by hand and loading it somewhere else.

Not every PIM fits the team using it

Teams that already run a PIM have one big advantage: a single place for the product record. But several described tools that never quite fit how their team works. Editing screens built for larger teams slow down lean ones. Pricing climbs with every new user or feature. Implementations stall before the PIM is fully connected, so people still pull data from two or three places to finish one update.

The value of a PIM depends on fit. The right one matches how your team works and gets fully rolled out, so the manual work actually goes away.

Launches run on email and re-keying

This was the most common theme. Approvals happen over email, with no record of who changed what and no view of how long enrichment takes.

Many of these teams are small. One or two people build every product record, and their capacity sets the pace of the whole catalog.

Assets live everywhere

Images, spec sheets and documents sit on file-sharing tools, shared servers, or a DAM that people avoid because it is slow.

Several teams questioned whether buying PIM and DAM separately still makes sense. They do not want two systems to keep in sync.

Supplier data arrives thin

Retailers described a volume problem. Hundreds or thousands of vendors send product data by spreadsheet and email, and much of it arrives incomplete.

For brands the gap opens earlier, at the handoff from product development to ecommerce.

Manufacturers carry extra outputs

For manufacturers the website is only one destination. They also produce spec sheets, print and digital catalogs, and data feeds for distributors and retail partners.

Most of that is built by hand today. One spec change means updating the same facts in several documents.

What’s changing: new triggers for retail teams

Two shifts stood out.

AI enrichment is now a stated requirement

Teams have stopped asking whether to use AI on product data. They are asking how to run it across the whole catalog.

The requests were specific. Teams want titles, descriptions, attributes and search fields generated to brand guidelines, with a person approving the result. They also want a clear view of what failed and why.

Some asked about images as a data source. They want to fill missing attributes from product photos and make sense of imagery nobody labeled.

Change inside the business opens the door

Many conversations started with an internal change. A platform migration, a new leader or a contract renewal forces the question of where product data should live.

Several teams were also weighing MDM against PIM, unsure where one stops and the other starts. Those moments are when the data model is easiest to fix.

Fundamentals still matter, and they pave the way for agentic readiness

AI dominated the agenda at the event, but most of the problems we heard on the ground were still about product data fundamentals.

The needs we heard were refreshingly familiar. Teams want one product record they can trust. A workflow that carries a product from creation all the way to launch. Clean supplier data. Assets that stay tied to the products they describe.

Every one of them is a problem a PIM was built to solve.

Agentic commerce readiness rarely came up as a requirement. Most teams had their attention on getting the product record in order first.

But the two go hand in hand. The work that fixes the fundamentals is the same work that makes a catalog readable to an AI agent.

The timing matters. Adobe’s 2026 holiday shopping forecast reports that AI-driven traffic to retail sites grew 127% year over year in August 2026. Adobe expects it to grow 130% across the holiday season and 141% on Thanksgiving.

An AI assistant can only work with the product data it can read. Complete attributes, consistent structure and accurate content serve a shopper, a marketplace and an agent alike.

The few teams already thinking about this asked practical questions.

How do we stay a trusted source when an agent does the shopping?

How do we keep attribution intact?

Because so few teams are planning for it yet, the gap is an opening.

Teams that get their product data clean and complete now will be ready before their competitors start asking the question. It is one of the clearest ways to gain a competitive edge this holiday season and beyond.