Identity resolution overview (BETA)

Ometria matches contact profiles that belong to the same person and merges them into one. The Overview screen shows you what that matching has done to your contacts, and the settings it is running on.

Go to Settings > Identity resolution > Overview.

Choosing a time period

Use the date range control in the top right to switch between Last 28 days, Last 3 months and Last 6 months. The stat cards and the chart both recalculate for the range you pick.

What the numbers mean

Card	What it counts
Contacts evaluated for matching	Contacts that went through matching in the selected period.
Merge rate	The share of those contacts that were merged into another profile.
Contacts merged	How many contacts were absorbed into a surviving profile.
Contacts after merging	How many contacts you have once those merges are applied.

Contacts evaluated for matching minus Contacts merged gives you Contacts after merging. A merge always reduces your contact count by one, because two profiles become one.

💡 A low merge rate is normal. It is not a sign that matching is failing. It means most of your contacts are already distinct people.

Merges over time

The chart breaks every merge in the period down by how it happened. Use the View by control to group the bars by Day, Week, or Month. Hover over any bar to see the date, the total, and the split by merge type.

overview dashboard.gif

Merge type	How the match was made
Deterministic	Two profiles shared an exact identifier value, such as the same email address or customer ID.
Probabilistic	The two profiles scored above your merge threshold on a combination of identifiers, such as name plus address plus date of birth.
Manual	Someone on your team merged the contacts by hand in Merge candidates > Manual similarity check.

You only see probabilistic bars if probabilistic matching is set up on your account.

Note: matching runs once a day, so merges appear in day-sized batches rather than continuously.

Live settings

This section shows the matching configuration running on your account right now. It is marked View only. Your Technical Project Manager (TPM) configured it. Contact your TPM or support to make changes.

Merge threshold

The percentage on the right is the minimum confidence two contacts need before Ometria merges them automatically. A pair scoring at or above it merges. A pair below it does not.

A higher threshold means fewer merges and less risk of combining two different people. A lower threshold means more merges and more risk.

Your identifiers

The identifiers currently used to match your contacts.

Column	What it shows
Identifier	A data field used to recognise and match your contacts.
Unique	Whether this identifier belongs to only one contact. A tick means an exact match on this value merges the contacts on its own. A cross means the value can be shared between people, so it counts as evidence rather than proof.
Status	Whether this identifier is active in identity resolution. Enabled means it is used for matching. Disabled means it is not.
Similarity identifier	How two values are compared. See below.

💡 A disabled identifier is still stored against your contacts. It is ignored during matching, so you can collect values now and start matching on them later without losing history.

Reading the Similarity identifier column

Unique identifiers such as a loyalty ID or customer ID show --. They only ever match on an exact value, because there is no sensible way for two loyalty IDs to be nearly the same.

Every other identifier is compared in levels, listed strictest first. The first level that matches decides how strong the evidence is. An exact match is the strongest, and each looser level counts for less.

For example, an email address compared as Exact then Jaro-Winkler similarity >= 94% matches two identical addresses outright, still recognises a pair that differs by a typo, and treats that second pair as weaker evidence.

Comparison	What it catches
Exact	Two values that are identical once standardised.
Jaro-Winkler similarity	Values that differ by a typo or a small spelling difference. The percentage is how similar they have to be.
Damerau-Levenshtein distance	Values with a small number of character changes, including two characters transposed. A distance of 1 means one edit apart.
+/- days	Dates within a given number of days of each other. Used to catch a date of birth entered a month or a year out.

A postal address is compared component by component rather than as one string, so its levels read as combinations such as Postcode (Exact), House number (Exact), Road (Exact) at the strictest level, loosening to Postcode (Exact) on its own.

Source weights

How much to trust each identifier based on where the data came from. A higher weight means stronger evidence for a match.

Column	What it shows
Identifier	Which identifier this weight applies to. Any means it applies to every identifier arriving from that source.
Source	Where the data came from, for example an order or a contact record.
Weight	The trust multiplier for this source. Values above 1.00 increase match confidence, values below 1.00 reduce it.

The weight multiplies the score the model produces for that identifier. A weight of 1.00 leaves the score unchanged, 1.20 counts the evidence 20% higher, and 0.70 counts it roughly a third lower.

This is how you account for some of your data being more reliable than the rest. An email address a customer typed at checkout and the same address arriving in a bulk contact record do not deserve equal trust, so the same match can carry different weight depending on where it came from.

Note: source weights are separate from the automatic adjustment Ometria makes for common values. Very common names and addresses already count for less on their own, without any weight being set.

Testing a change

You cannot edit any of these settings here. To try a different threshold, similarity function, or weight, select Test a change at the bottom of the screen.

This takes your settings as a draft copy and runs them against your real contacts so you can see the effect. Nothing is applied to your live account until your TPM makes the change for you.

Why contacts merge over several days

Contacts merge in pairs. If four profiles all belong to the same person, two merge first, and the rest are compared against the combined profile on the next run. A large group of duplicates settles over several days rather than all at once. Nothing is lost while a candidate waits.

When your numbers are still moving

If you see the Data is still loading banner, your account is still ingesting data into Ometria. The figures below the banner are accurate for the data received so far, but they will shift as more arrives. Wait until the banner clears before you use these numbers to judge how well matching is working.

⚠️ A merge cannot be undone. Review merge candidates carefully before confirming a manual merge.