CASE STUDY · CUSTOMER EXPERIENCE (CX) & PERSONALIZATION
Building a unified customer profile across five touchpoints.
- 18%18% of duplicate profiles resolved
01 · PROJECT OVERVIEW
The client and the brief.
An anonymised mid-market B2B services company. Name withheld under confidentiality.
Identity resolution and a profile store. We matched records across systems using agreed rules, merged them into one customer profile with a clear history, and made that profile available to the teams and tools that needed it.
- 1Fragmented customer identities across systems.
- 2The same client appeared as three differently spelled accounts in the CRM, two in the billing system and again in the support tool.
- 3Contacts changed jobs, shared mailboxes mixed with personal ones, and nobody could see one view of a client.
- 4Support did not know what sales had promised.
- 5Marketing mailed people who had asked to stop.
02 · WHAT WENT WRONG
What went wrong, and how we fixed it.
Every engagement hits problems. These were the ones that mattered here, and what we changed so they did not come back.
- PROBLEM 01The first matching pass merged two different companies that shared a parent-company email domain
Fortunately, the review step caught it before anything reached production. We tightened the rule so that generic and shared domains were never enough on their own, added company-level hierarchy handling, and made the audit report show every merge reason. We also lowered the auto-merge band and let the review queue take more cases at the start.
- PROBLEM 02We also lowered the auto-merge band and let the review queue take more cases at the start
- PROBLEM 03A second hurdle was data ownership
Sales and support each believed they owned the "correct" version of a client's details, and early merges triggered arguments about which source should win. We settled it with a simple rule per attribute: billing wins for legal name and address, the CRM wins for contact roles, and support wins for service preferences. The rule was written down, agreed by the three department heads and stored with the profile, so every value shows where it came from.
- PROBLEM 04The rule was written down, agreed by the three department heads and stored with the profile, so every value shows where it came from
03 · THE PROCESS
How we ran it.
- 01Weeks 1 to 2
inventoried five touchpoints and the identifiers each one held, including emails, phone numbers, account codes and company names.
- 02Weeks 3 to 5
designed the matching approach with the operations team, and ran it on a sample for manual review.
- 03Weeks 6 to 8
built the profile store, connected the systems and set up a review queue for uncertain matches.
- 04Weeks 9 to 10
rolled out to support and sales with training, then enabled segmentation for marketing.
- 05Step 5
A fortnightly steering call with the heads of sales, support and marketing.
04 · RESULTS
What changed.
Measured against the way the work ran before the engagement began.
- NOW 01
Support agents see recent sales, billing and service history on one screen.
- NOW 02
Marketing builds segments from the unified profile, with opt-outs respected in all channels.
- NOW 03
A named data owner reviews the match queue each week and agrees rule changes.
- NOW 04
New systems are connected through the profile layer rather than through point-to-point exports.
- 18%
18% of duplicate profiles resolved
Cross-channel service and segmentation enabled
05 · WHAT WE BUILT
How it works.
- Connectors pulling records from CRM, billing, support, website and email platform into a staging area.
- Normalisation of names, emails, phone numbers and company names before matching.
- Deterministic rules first, such as an exact match on a verified email, then probabilistic matching with a score for the rest.
- Three bands: auto-merge for high-confidence matches, a human review queue for the middle, and no merge for low scores.Merges were reversible and logged.
- A profile store keeping the source of each attribute and a link back to the original records.
- Consent and opt-out status carried in the profile and checked by every channel that used it.
- Monitoring and handover
- We set up reports on match rate, review-queue size and the number of merges reversed, so quality problems show up as numbers. Staff were trained to read the profile and to raise a wrong merge. A runbook covers adding a new source system. The company agreed that no new tool goes live without a plan for how its customer records join the profile.
06 · QUICK ANSWERS
The short version.
- What problem did the client have?
- Fragmented customer identities across systems. The same client appeared as three differently spelled accounts in the CRM, two in the billing system and again in the support tool. Contacts changed jobs, shared mailboxes mixed with personal ones, and nobody could see one view of a client. Support did not know what sales had promised. Marketing mailed people who had asked to stop.
- What did Brandsmashers build?
- Identity resolution and a profile store. We matched records across systems using agreed rules, merged them into one customer profile with a clear history, and made that profile available to the teams and tools that needed it.
- What were the results?
- 18% of duplicate profiles resolved; cross-channel service and segmentation enabled.
YOUR TURN
Have a similar requirement?
Tell us what is stuck. We will come back with a plan and matched, pre-vetted engineers.
