METRO: customer data across online and store purchases
Mindbox’s June 2021 account of Eugene Mischenko’s METRO conference presentation describes connecting store data with four online stores, an app and partners. The case raises a practical question for a retail chain: which customer decisions become possible when those purchase histories can be read together? The analysis below examines identity rules, campaign control and the commercial test of an additional purchase.
Establish which purchases belong to one customer
A loyalty card, app account and delivery order can carry different identifiers. A retailer applying this approach needs a documented matching rule for each pair of systems. A verified account link provides stronger evidence than a shared email address; a household may use one address for several people. Keep the original transaction and the reason for each match so an incorrect merge can be reversed.
Test the rules with guest checkout, a changed phone number, duplicate cards and returns. Include unmatched purchases in the reconciliation. Otherwise an apparently complete customer view can silently exclude the very customers whose behaviour the commercial team wants to understand.
Coordinate offers across the customer history
In his September 2021 Mindbox interview, Mischenko described duplicate online and offline emails as an immediate problem that shared customer data could address. For campaign design, that suggests a concrete first release: one contact policy applied before messages leave each system.
Define the period over which contact frequency is counted, the priority of competing offers and how preferences reach connected tools. Keep service updates separate in the campaign design. A failed delivery requires a useful status message even when a promotional limit has been reached. Inspect actual outgoing events, including partner activity where data is available, to verify the policy operates across systems.
Test the value of an additional channel
A customer who uses two channels may already have a greater need for the retailer’s products. Comparing that customer with an occasional store visitor would confound demand with the effect of the campaign. Select customers with similar prior purchasing opportunities, assign an eligible comparison group and observe both groups over the same period.
Measure total contribution across their purchases, allowing for discounts, picking, delivery and returns. Also check whether the offer simply moved the next planned store purchase forward. The decision is whether the communication created enough additional value to cover the cost of changing behaviour, including the cost of serving that behaviour repeatedly.
Give teams a shared acceptance check
Before expanding the project, have CRM, store operations, support and finance follow the same test customer. Place an order, apply an offer, make a store purchase and return one item. Confirm that the resulting history supports the intended next message and the agreed financial calculation.
Record which team owns a missing event, an uncertain identity match and a conflicting promotion. A useful release review contains the failed scenario, its customer consequence and the correction owner. This makes a customer data investment assessable through observable decisions: a message suppressed, a purchase recognised or a refund reflected before the next offer is selected.
Sources
Apply this to your business
Bring your business question and current figures. We will define the work, the data required and the decisions your team needs to make.