How It Works
Six stages and one rule: the evidence is always on the page, and a person approves every merge.
DIO Data Concierge
Evidence first.
A person behind every merge.
CSV and Excel exports from your CRM, ERP, marketing platforms and spreadsheets. Large files stream in rather than loading all at once, and every source keeps its own identity, so you always know which system a value came from.
Columns are mapped automatically for you to confirm. Profiling then shows how full each field really is, which values sit in the wrong column, and what was never captured at all, before a single match is trusted.
Exports that look complete often aren’t. It’s better to know that on day one than after a migration.
Names, emails, phone numbers in international formats, company names and places are standardised inside the platform, so “KESTREL FREIGHT” meets “Kestrel Freight Pty Ltd” and “(03) 5550 1234” meets “+61 3 5550 1234”.
Your source values are never changed. The normalised forms sit beside them.
Likely duplicates are found across every source and scored on the evidence: what agrees, what conflicts and what’s missing. Each pair gets one of four grades, and the evidence behind the grade travels with it.
When new evidence arrives from another source, a pair is graded again automatically. Nothing needs resetting.
Reviewers see both records with the evidence for and against, strongest pairs first. They choose Same person, Different people, Can’t tell or Never match, and every decision is kept. Reversing one keeps its history.
An optional, governed AI second opinion can help on hard cases. It can’t merge anything.
Confirmed identities become a golden record built field by field: the best value from each source, chosen by rules you can read, and each one traceable to the row it came from.
Identities are checked for coherence first, so a chain of shared numbers can’t fuse different people. Emails, phone numbers and addresses are validated, and shared switchboards and team inboxes are recognised for what they are.
Governed enrichment will let an approved provider fill missing fields. Enriched values arrive as evidence with their source, wait for acceptance, and rank below what your customers told you directly.
Nothing about a person is accepted automatically, and nothing is sent to a provider your workspace hasn’t approved.
Every candidate pair lands in exactly one grade. None of them merges anything on its own.
Queued first. A person confirms before anything is merged.Several independent signals agree. Measured precision on real CRM data: 95% in development, 100% on a held-out test set.
Queued for side-by-side review.Some evidence agrees; some is missing or points the other way. This is where an experienced reviewer earns their keep.
Set aside, not guessed.The data doesn’t contain enough to decide, typically a shared name and nothing else. Expert reviewers couldn’t tell 88.5% of these either. Withheld from routine review, available on request, and graded again when new evidence arrives.
Not queued. Available whenever you look.The evidence points to different people. Nothing is discarded.
Send us one export from your CRM, ERP or a spreadsheet you’ve stopped trusting. We’ll show you what’s really in it: the duplicates, the gaps, and the evidence behind every match.
No bulk merges, no black box, and nothing changes in your systems until your team says so.
Phone
