Key Points
- Each new brand, franchise, or acquisition can turn one customer into several people in a company’s systems, and every artificial intelligence (AI) model or agent built on that data inherits the confusion.
- A customer data platform (CDP) fixes this by unifying each customer’s identity data across every brand while letting each brand keep its own relationship, consent, and view.
- Offline franchise data needs consistent formats and cleansed addresses before matching can recognize the same household across locations.
- When a company makes an acquisition, the acquired customer file deserves a full, tested re-match in the CDP rather than a quick append to existing customer records.
Dana has been meaning to paint her house for three summers running. This year she finally books a crew, picks the sage green she’s been saving on her phone for months, and loves how it turns out. The next spring, ants march into her kitchen, so she calls the pest control company her neighbor swears by. Dana has no idea the painters and the exterminators share a parent company.
Then the mail starts. Two postcards from two different locations of the painting brand land in the same week, one offering a discount on the first job she’s already had. The pest control brand calls to welcome her as a brand-new customer, unaware she’s spent thousands with its sister company. When the parent company buys a lawn care business that fall, Dana turns up in a fourth customer file under the name on her old account, the one from before she got married. And when the parent company pilots an AI assistant for its service team, it greets Dana as a new customer too, because it’s reading the same four records.
Nobody in this story made a mistake. Every system did its job inside its own boundary, and Dana slipped through the spaces in between.
Closing those gaps starts with data readiness, the work a customer data platform (CDP) does before any campaign or AI model runs. It resolves every record to one customer identity with brand, location, and source attached, so the parent company sees one Dana and each brand keeps its consent.
Why does one customer turn into five records?
Dana’s experience is built into the way multi-brand companies grow. Every brand, franchise location, and acquired company arrives with its own customer relationship management (CRM) system, its own customer identifiers (IDs), and its own idea of what counts as a customer.
AAA knows this pattern well. Its 45 federated clubs each kept their own member data and business rules, which two earlier technology vendors couldn’t reconcile into a single member repository.
In Redpoint’s research report The State of Customer Data Readiness, one data leader blamed duplicate customers on enterprise resource planning (ERP) systems inherited from past acquisitions. Another had worked at a company where roughly 10% of clients couldn’t be matched across business units, in one case because the same corporate client went by a different legal name in every system.
For Dana, that means duplicate offers and a sister brand that treats her like a stranger. Any AI built on top inherits the same problem. A churn model scores four partial customers instead of one loyal one, and an agent answering a service question pulls whichever record it finds first. Nearly every leader in Redpoint’s research described the same dynamic: AI exposes gaps that were there all along.
Should you merge your brands into one customer profile?
Picture the people around the table. The pest control brand’s marketing lead doesn’t want Dana’s painting history cluttering her campaigns or another brand’s consent rules overriding hers. The parent company’s leadership wants to know what Dana is worth across everything she buys. Both of them are right.
So merge the identity, and leave the brands intact. That starts at ingestion, where each record picks up a pedigree before matching begins: a tag recording the brand, location, and system it came from. Identity resolution then matches every record to one shared identity key, and the Customer 360 can sit at the individual level or at the individual-and-business-unit level. Leadership sees Dana’s full value, and the pest control team still sees its own relationship, consent, and history with her. That’s also the version of Dana an AI model or agent should see: one correct, current, de-duplicated profile, with brand-level consent attached so the AI respects it.
| Separate stack per brand | One merged profile | Shared identity, brand-level views | |
| Recognizes one customer across brands | No | Yes | Yes |
| Keeps brand-level consent and preferences | Yes | Often lost | Yes |
| Onboarding an acquired company | New silo | Large migration | Add sources and re-match |
| What each brand team controls | Everything, including its own matching | Little; a central team sets the rules | Its own campaigns, consent, and audiences; matching is shared |
| Audit trail for merges | Per brand | Varies | Lineage by source |
The data leaders in Redpoint’s research agree: the organizations making real progress centralize the quality and matching engine and leave ownership with the business units closest to the customer.
Keeping brands completely separate is the right call when they’re separate legal entities that can’t share personally identifiable information (PII), or when a brand is headed for divestiture. Resolve identity within each brand, and revisit if the boundary changes.
How do you bring offline franchise data into the same identity?
Every franchise owner runs things a little differently. One location logs jobs in a field-service app, another exports a spreadsheet every Friday afternoon, and a third has fifteen years of customers in a system older than some of its crew. All of them know Dana’s house. None of them describe it the same way.
That’s why standardization comes before matching. Each location’s systems map to the same feed layouts, and a set error threshold rejects any load with too many problems. Addresses run through Coding Accuracy Support System (CASS) certification and National Change of Address (NCOA) processing, because a family that moved two years ago stays two records until something corrects the old address.
One Redpoint franchise customer brought 25+ years of data in ten formats into a Customer 360 covering people, households, and properties, plus commercial contacts and job sites.
What changes when you acquire another company’s customers?
When the deal closes, somebody on the data team opens the lawn care company’s customer file for the first time. Customer 12345 is a retired teacher in Ohio. In the parent company’s CRM, customer 12345 is someone else entirely.
Many mergers and acquisitions (M&A) take a wrong turn here, appending the new file to existing customer records and moving on. Redpoint’s research found that most organizations carry an acquisition’s data debt for years, and one bank was still working through it two to three years after the deal.
A cleaner path starts with those colliding IDs. The source name gets appended (12345-CRM, 12345-ERP) so every incoming record stays unique. Then the combined file gets a full re-match, so the matching engine compares every record against every other one instead of slotting new records into groups built before the deal. Because this decision is hard to undo, data stewards can test match settings against the combined data first in an experiment framework. Labeled changes between runs show stewards exactly how the two customer bases will come together, and they can override any grouping that looks wrong. Survivorship rules then decide whose system wins for each attribute, based on source trust and recency. That re-match is also Dana’s best chance of reconnecting her old account, since probabilistic matching on her address, phone, and email can carry the match where her name no longer does.
The takeaway
Before the next brand joins the family, decide two things: which boundaries customer identity always crosses (all of them) and which ones it respects (consent, visibility, and survivorship). Teams that settle that early can welcome an acquisition as a handful of new sources instead of a multi-year cleanup, and every AI initiative after that starts from customer data that’s ready.
Dana will never see any of this work. She’ll just notice that the companies she trusts with her home finally seem to know who she is.
To see how merge settings get tested before they reach production, walk through how Redpoint Identity Studio matches experiment documentation.
FAQ
Do we need a separate CDP instance for each brand?
No. Brands can share one identity layer while keeping their own sources, consent, and audiences.
Can a customer opt out of one brand without opting out of all of them?
Yes. Each brand’s consent lives on the resolved identity, so an opt-out from one brand doesn’t suppress the others unless the customer asks for that. Global suppressions such as do-not-contact requests apply to every record that resolves to the same person.
Can franchisees see customers who also buy from other locations?
Admins can scope access so a location sees its own customers while the parent company sees the full profile.
Why does unifying customer data across brands matter for AI?
AI models and agents answer from whatever customer records they can reach. If one person exists as four records across brands, the AI gives four partial answers. Ready customer data (correct, complete, current, de-duplicated, and governed) gives every model one consistent view of each customer.
How is this different from unifying customer data across regions?
Regional unification deals with country-specific formats, character sets, and privacy laws. Multi-brand unificationn deals with ownership boundaries inside the company. Most global enterprises need both, and our guide to unifying customer data across regions covers the regional side.
