Thinklytics

National Specialty Retailer · Retail & E-Commerce · Dallas, TX · 14 weeks

Customer 360 data foundation built across 8 channels

A national specialty retailer with 340 stores and an expanding e-commerce presence struggled with customer data spread across eight disconnected systems and no way to link identities. We created a Customer 360 data foundation that merged identity resolution, purchase history, and behavioral data. This groundwork enabled the retailer to launch their first personalized marketing campaigns.

Challenge

The retailer’s loyalty, POS, e-commerce, mobile app, and email systems operated independently. Customers shopping both in-store and online appeared as two distinct entries in reports. Without merging these data sources, personalized marketing was ineffective.

Approach

We combined data from eight systems using deterministic and probabilistic matching to create a single customer profile. We loaded these profiles into Snowflake and linked them to the marketing automation platform. Then, we created 14 customer segments based on purchase frequency, channel preference, and lifetime value.

Outcome

We cleaned up 2.3 million duplicate customer records, ending with 1.4 million unique profiles. Using this data, we launched a targeted email campaign that brought in $4.2 million more revenue in three months. Open rates jumped from 18 to 31 of every 100 emails by tailoring content to specific segments.

How we pulled customer data from eight different systems without losing our minds

Their customer IDs were a total mess, no consistency across systems. So, we started by matching the low-hanging fruit: emails, phone numbers, and loyalty IDs. That got us part of the way there, but plenty didn’t line up perfectly. For those tricky cases, we leaned on some probabilistic matching to figure out which records were likely duplicates. By the time we wrapped up, we’d cut their chaotic 2.3 million records down to a clean set of 1.4 million unique customers.

Getting to know our customers for real and kicking off campaigns that actually work

We took a deep dive into customer behavior, how often they bought things, how much they spent, which channels they preferred, and how recently they shopped. From all that, we identified 14 distinct customer groups. Then, we plugged those segments right into their marketing automation system. This gave the retailer the power to run campaigns that actually connected with the right people.

Getting a clear picture of where our revenue’s actually coming from, and how much each piece brings in

So, we set aside a holdout group to test how much personalization actually mattered. Over three months, the personalized campaign brought in $4.2 million more than the holdout. That gap? It nailed it for us, our data work was truly pushing sales up.

Results

  • $4.2M Incremental personalization revenue in 90 days
  • 2.3M to 1.4M Duplicate records resolved to unique customers
  • 18 to 31 of 100 Email open rate improvement
  • 14 Customer segments defined and activated

We had been talking about creating a single customer view for four years. Thinklytics put it together in 14 weeks, and we rolled out our first personalization campaign the following month. In just 90 days, that campaign brought in $4.2 million, more than paying for the whole project several times over.

Chief Marketing Officer, National Specialty Retailer

In the client's terms

The question

Where does this number come from, and can the path be trusted?

The situation

Loyalty, POS, e-commerce, the mobile app and email all ran independently. A customer who shopped in store and online appeared as two people, which made personalised marketing largely guesswork.

What we found

Personalisation had been treated as a marketing capability to buy. It was an identity problem, and no campaign tool can personalise for a person the data says is two people.

What it produced

2.3 million duplicate records resolved to 1.4 million unique customers. A targeted email campaign on the new segments returned $4.2M in incremental revenue over 90 days, with open rates moving from 18 in 100 to 31.

The roadmap

Eight systems were resolved into a single customer profile using deterministic matching where identifiers allowed and probabilistic where they did not. Profiles landed in Snowflake and were connected to the marketing automation platform, then segmented fourteen ways on purchase frequency, channel preference and lifetime value.

The work behind this result

Related case studies

Frequently asked questions

Why does personalised marketing underperform?

Often because the customer is several records. At this national specialty retailer a shopper who bought in store and online appeared as two people across loyalty, POS, e-commerce, app and email, which made personalisation guesswork. Resolving 2.3 million records to 1.4 million customers fixed the input.

How do you measure whether personalisation actually works?

Hold a group back. A holdout was reserved here before the campaign ran, and the personalised group returned $4.2M more over 90 days than the holdout, which is a measured difference rather than a before-and-after against a moving baseline.

How do you match customer records across channels?

Deterministic first, probabilistic for the remainder. Email, phone and loyalty ID resolved the records sharing an identifier here; probabilistic matching handled the rest. Email open rates moved from 18 in 100 to 31 once segments were built on the unified profile.