How to use predictive analytics in retail to plan next season
The short answer
Predictive analytics in retail uses past behavior to score what each customer is likely to buy next. For high-value customers, score each product category on recency, frequency and spend in that category, plus what similar customers bought after the same history. Products with the most high-propensity buyers deserve depth and early access next season.
Why it matters
Your highest-value customers carry much of your full-price demand, so guessing wrong about what they want next season means markdowns on one side and missed sales on the other. Propensity scores turn their history into a buying and marketing plan.
The question to ask
Copy this into Lexi as written.
What the answer looks like
Illustrative output on sample data
Illustrative example: how Lexi answers this for an apparel retailer. It scores its top 10% of customers by 24-month gross margin, 14,000 people, for each spring category.
- Uses store and online purchase history, with returns removed.
- Scores each category from 0 to 100 on recency, frequency, spend and the paths of similar customers.
- Counts 70 or above as high-propensity.
What the data shows
- Linen shirts: 5,200 high-propensity customers
- Tailored trousers: 3,900
- Knitwear: 1,100
What it means
Knitwear sold well to this group last spring, but many of them bought it in the fall, so demand has already been met.
What to do
Put more depth into linen and send the 5,200 an early-access email.
Scores reflect past behavior, so a new trend can shift them. Want me to push each product group to Klaviyo or Meta as a segment?
Frequently asked questions
What data does a product propensity model need?
At minimum, two years of transactions with product, category, price paid, date and customer ID, across every channel. Returns matter, since a category a customer keeps returning is a weak signal. Email and browsing data can sharpen scores, but purchase history does most of the work for repeat retail categories.
Can you predict next purchase from last season's bestsellers?
Only roughly. Bestseller lists show what sold to everyone, including one-time and promotion-led buyers. Your high-value customers may already own last season's hit and be ready for something else. Scoring each customer on their own history picks up those shifts ahead of any bestseller list.
How is a next best product score used in marketing?
Each customer's top-scoring category becomes the product they see first in email, on the homepage and in paid social. Higher scores earn earlier sends or stronger creative. Keep a holdout group for each campaign, since a good score should lift sales against customers who got the standard message.