Our new interactive demo shows how clearMDM and its Data Steward Agent for Agentforce keep retail customer data trusted: safe duplicates merged automatically, VIP clients and consent conflicts reviewed by a person, and every merge recorded and reversible. It takes about five minutes and needs no sign-up.
One customer, three channels, three records
Retailers sell online, in store and through loyalty apps, and each channel tends to create its own customer record. The same shopper can easily end up with two or three, each holding part of the picture.
Most of those duplicates are harmless to merge. Some are not. A Platinum client with a personal advisor, or a customer who opted out of marketing in one channel and in to it in another, needs a person to decide. Merge them blindly and you risk losing a clienteling relationship or breaking a consent promise.
The hard part isn’t finding duplicates. It’s deciding which ones a machine can merge on its own, and which ones need a person.
Meet Maison Aurelia
The demo follows Maison Aurelia, a fictional luxury retailer, and its data steward working with the clearMDM Data Steward Agent in Salesforce. It shows three things:
- VIP and clienteling protection. A Platinum client with a personal advisor is held for review, not merged automatically.
- Consent-aware governance. The agent won’t merge a held or rejected match without an explicit yes from the steward. That rule is enforced in clearMDM itself, not just in the agent’s instructions.
- Context-aware stewardship. Low-value online duplicates are accepted automatically, and every merge records how it was decided: automated, agent assisted or human review.
What you’ll see
- The review queue. Only the matches that need a person wait for review, each with its reason: VIP Customer, Consent Conflict or Low Match Score.
- The governance moment. Charlotte has an online Silver profile and a Platinum profile at the Bond Street store, with an advisor and a different consent. The steward asks the agent to merge them, and it stops. Only after an explicit “yes, merge anyway” does it merge, keeping the Platinum record as the master and recording the decision as agent assisted.
- Safe automation. Olivia’s two online profiles were accepted automatically, so finding and merging them takes seconds.
- Match Fit. Recalculating James Thornton’s group re-scores it against today’s rules and shows how well each record fits the whole group, not just one pair. The weak link stands out, and the steward can easily remove it.
- Respecting a person’s decision. A steward has already rejected Amelia’s match: mother and daughter at one address. The agent is blocked from merging and asks before changing a person’s decision.
- Completing the picture. Eleanor’s loyalty-app profile, “Ellie Price”, matches but isn’t in her group yet. The agent finds it and adds it once the steward confirms.
- MDM Activity. Every merge, how it was decided, how much was automated, and a Roll Back button for any merge that shouldn’t have happened.
Why it matters
AI agents can do a lot of a data steward’s legwork: finding duplicates, explaining matches, re-scoring groups. But in retail, a wrong merge can cost a client relationship or break a consent promise. clearMDM lets the agent help without handing it the final say on sensitive records.
Low-risk duplicates are handled automatically, sensitive ones get a person, and the agent does the legwork in between. Every merge says how it was decided, and every merge can be undone.