How It Works

A Match IQ job runs as a single batch. You define the inputs and the rules; Match IQ does the comparison and grouping. Conceptually it's a five-step flow:

1. Prepare your data - standardize and structure the records you want to match.
2. Set up the job - choose which records to include, what to compare, what counts as a match, and what output to produce.
3. Create match keys - Match IQ reads each record and builds a compact key from the fields that matter.
4. Find matches - records are grouped, compared, and sorted into duplicate groups.
5. Process the results - consolidate, output, purge, or post the results according to your rules.

The first two steps are yours; Match IQ performs the last three. You can run these as one job or, while you're dialing in your rules, split them out to inspect results between stages.

Comparing every character of every record against every other record would be needlessly slow - and often counterproductive. Instead, Match IQ builds a match key for each record: the significant, standardized pieces of data you choose to compare, such as first name, last name, street number, street name, and ZIP Code.

For example, a record for George F Hayes, 100 Main St #5, 55424 becomes a normalized key built from its name and address components. Matching then works on these clean keys rather than the messy raw data, which improves both speed and accuracy.

To match efficiently, Match IQ first sorts keys into break groups - clusters of records that have a reasonable chance of matching (for example, sharing a ZIP Code or the start of a last name). Only records within the same break group are compared to one another, which eliminates vast numbers of pointless comparisons and keeps large jobs fast.

Within each break group, Match IQ compares the keys. When two or more records match, they form a match group (also called a duplicate group). Match IQ then ranks the records in each group and classifies them:

• Unique - a record with no matches.
• Master - the highest-ranked record in a match group, chosen by your priority rules.
• Subordinate (duplicate) - the remaining records in the group.

Successful matching comes down to three things:

• Data - clean, consistent input yields better matches. Standardizing names and addresses first makes a large difference.
• Rules - your criteria for when two records are the same. Match IQ ships with default rule sets to start from, and matching is usually an iterative process of running, reviewing reports, and refining.
• Results - the outputs you want: a deduplicated file, a consolidated best record, updated source data, and the reports to prove it.