Coffee, Groceries, Cash: 4-Dimension Scoring That Boosts Customer Loyalty

I have two friends who like Starbucks coffee: let's call them Liz and Michael.

Liz buys black coffee, 5 days a week, before work.

Michael only buys occasionally, but when he does, he buys the Venti spectacular with all the trimmings. Sometimes, he even brings his family (4 kids and his wife) - this is when he spends $40 in a single visit.

Since Starbucks is a star in personalization, the offers it sends to Liz and Michael are very different.

??The goal for Liz is to (1) get her to spend more per visit ??and (2) come in again in the afternoon?? . As such, her bonus offers ask her to (1) spend at least $10 in a transaction and (2) make a purchase after 2pm

??The goal for Michael is to get him to shop more frequently ??. His offers would ask him to make a purchase 3 times a week to get bonus stars.

I adopted the same principle when working with a grocery store. Customers were scored across 4 dimensions:

1. Risk (probability to attrite - i.e. leave to shop at a competitor)
2. Average ticket
3. Frequency and
4. Profitability/Margin

Then, we created segmented campaigns and offers with a different call to action for each group.

For instance, the group of ??Low Risk, ??Low Average Ticket, ??High Frequency, ??Low Margin was shopping the grocer regularly, but likely as a secondary store. As such, we needed to understand deeper what they were buying and more importantly not buying.

• Were they shopping only 1 category/department? If so, the goal/call to action (CTA) was be to get them to try a new category department.
• Were they not buying staples (e.g. milk, protein, toilet paper, etc.)? If so, the CTA was be a bundle offer on staples
• Were they doing only small shops? If so, the CTA was be to have X number of items in their basket
• Perhaps, they were grazers and buying a few items several times a week, but their overall spend was very high. In that case, the focus turned to profitability and the CTA was focused on private label products

Then, the question arose about offers. Would the customers respond to discounts, bonuses, or free merchandise? All of these had to be tested to determine which customers were motivated by which type of offer, for which CTA, and level of richness (e.g. 5%, 10%, 20%?).

Next, this exercise had to be repeated for each combination of dimensions.

We also needed to learn the optimal frequency of communication, the timing vs. frequency of rescoring the model, and the long-term impact on KPIs and customer behaviour.

After much trial and error (because nothing is perfect at the beginning), this program started meeting and even exceeding its goals.

In the world of AI and generative AI, this kind of multivariate testing becomes simplified and it becomes easier to continue to optimize based on learnings from previous rounds.

Let's chat about how we can use goal-based segmentation to drive personalized communication for your business. Contact me Lia Grimberg, CLMP?, MBA .

#PersonalizedMarketing #CustomerRetention #AISegmentation