After Schnucks brought acquired stores under its banner and launched a rewards program, longtime shoppers from those stores scanned in as strangers, while many already existed in the system under a second identity. Matching the customer data into one household view took nine weeks, not the months a manual merge would have needed.
INDUSTRY
Retail (regional grocery)
BUSINESS PROFILE
Over 100 supermarkets · St. Louis, MO · Family-owned since 1939 · ~12,000 employees
Schnuck Markets, known to shoppers as Schnucks, is a family-owned grocery company founded in St. Louis in 1939 by Anna Donovan Schnuck. Now in its third and fourth generations of family leadership, Schnucks operates more than 100 supermarkets across Missouri, Illinois, and Indiana, with around 12,000 employees. The company has grown through a long series of acquisitions and runs the Schnucks Rewards loyalty program across its stores, app, and online ordering.
When Schnucks brought a chain of acquired stores under its banner and rolled out the Schnucks Rewards program across the company, a problem surfaced at the registers. A shopper who had bought groceries at one of the acquired stores every week for years scanned into the new program as a stranger, with no history and no standing. Many of those same shoppers were not strangers at all; they already had a Schnucks profile from an online order or the fuel program, now sitting unconnected to the rest.
The customer data had accumulated in pieces. Point-of-sale systems at the acquired stores, Schnucks' own store data, the e-commerce accounts, and the fuel program each held their own version of a shopper, keyed differently and never reconciled. One household could exist three or four times across those sources, and nothing tied the copies together.
For a loyalty program, that fragmentation is close to fatal. Personalized offers went to one copy of a shopper and not the others, so the savings a household earned depended on which card it happened to use. The promise of the program, that Schnucks knew and rewarded its regulars, broke on the data.
The customer-data team absorbed the difference by hand. Every week brought requests from shoppers who had two accounts, points on the wrong card, or a spouse's purchases missing from the household, and resolving each one was manual detective work. The queue never emptied, and the launch date was fixed.
Schnucks had a deadline and a deduplication problem too large to hand-merge, and the obvious shortcut was risky. Loosely matching shoppers to clear the backlog faster would mean occasionally merging two different people, which in a loyalty program means showing one shopper another's purchase history. The team needed matching that was both fast enough for the launch and careful enough to never do that.
Schnucks chose MatchLogic to build one household view from its scattered customer data and licensed the Server tier, with the Workflow Scheduler rebuilding the master every night. A nightly file then loaded the unified household IDs into the Rewards platform and the marketing tools. Three capabilities mattered most.
Anchored by Jaro-Winkler distance and Levenshtein edit distance, with phone, email, and address standardization underneath, the engine linked the same shopper across the acquired stores, Schnucks' own data, the app, and the fuel program despite typos, old numbers, and maiden names.
Using its own rule definitions, the team rolled individuals into households at a shared address while keeping unrelated roommates apart, so an offer reached a family once and credited the right people rather than a building full of separate accounts.
Because wrongly combining two shoppers would expose one person's purchases to another, high-confidence matches merged automatically and everything short of that waited for review. In a hand-checked sample of a thousand pairs, false matches came in under one percent.
Stores were open and the program was launching, so none of this could disrupt the registers. Everything ran against nightly extracts, with the unified household master written to a separate table that fed Rewards and marketing rather than touching the point-of-sale systems.
The team matched within Schnucks' own data first, then within the acquired stores' data, then across the two, and finally rolled individuals into households. The mess was ordinary and everywhere: a shopper enrolled twice under two phone numbers, an e-commerce account under a work email, a member who had moved and kept both addresses. Validation ran against pairs the team had judged by hand before any merge was allowed to commit.
From the first extract to a unified household view feeding the Rewards launch, the work took about nine weeks, well inside the months a manual merge of that size would have taken.
The customer data from the acquired stores joined the same household view as the rest of Schnucks in about nine weeks, in time for the program launch. A shopper from a bought store now carried a single recognized identity, where a comparable hand-merge would have run for months and still drifted out of date. The deadline that started the project was met with room to spare.
The match drew on roughly 4.5 million records from point of sale, e-commerce, the fuel program, and the acquired stores, and resolved them to about 1.4 million unique households. The loyalty master fell from around 1.75 million separate profiles to those households, a 20% cut, with every original profile kept and linked so points and history followed the shopper. Personalization now runs against the household rather than whichever card was scanned.
The weekly flood of duplicate-account and missing-points tickets the customer-data team had worked by hand dropped about eighty-five percent, because the nightly match resolved most of them before anyone called. The two people who had spent their weeks on manual merges moved to loyalty analysis and campaign work. The job that used to be reconciliation became insight.
With the copies unified, the share of transactions Schnucks could tie to a known household rose from about 64% to 86%. Offers reached the right family, the program recognized the regulars it had promised to recognize, and shoppers from the acquired stores kept the standing they had earned before the sign on the building changed.
Schnucks now runs its loyalty program from one view of each household rather than a drawer full of duplicate cards. The customer-data team is extending the same matching to the supplier and product records behind its shelves, and the nightly job keeps the household view current as new shoppers join and old ones move. For a grocer that has known its neighbors since 1939, recognizing a regular wherever they shop, in the aisle or on the app, is the corner-store relationship rebuilt at the scale of a hundred stores.
Founded in 2003, MatchLogic has matched over 2 billion records across 4,500+ enterprise installations, powering production-scale entity resolution pipelines.
MatchLogic combines deterministic and fuzzy matching with visual profiling and drag-and-drop cleansing workflows to identify duplicates and maintain trusted golden records.
Built for fintech, healthcare, insurance, and government teams, the platform provides REST API integrations, cross-platform deployment, and full auditability across every match operation.
.svg)
