Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
  1. Home
  2. Blog
  3. How Do I Stop Sending Discount Codes to Customers Who Would Have Bought Anyway?
Problems We Solve

How Do I Stop Sending Discount Codes to Customers Who Would Have Bought Anyway?

Blanket discounts reward customers who would pay full price. We build machine learning targeting so offers go to the customers they actually persuade.

Updated 3 min readBy SpiderHunts Technologies

Free estimateNo obligation

Get a free estimate

Tell us what you need. A senior engineer reads every enquiry.

Takes under a minute. We never share your details.

  • Free consultation
  • No commitment
  • NDA on request

Prefer to talk? Book a free 30-minute call →

Quick answer — TL;DR

A blanket discount cuts margin on every loyal customer who would have bought at full price. Machine learning targeting, built from past campaigns where some customers got the offer and some did not, estimates which customers an offer actually changes, so discounts go to them and full-price buyers are left alone.

The campaign that looked like a success

The discount email went to the whole list. Sales jumped that week, and the campaign report looked excellent. But when finance looked closer, a large share of the orders came from regular customers who buy every month anyway. They got the discount on an order they were going to place regardless.

It keeps happening. Every promotion shows a spike, and every spike includes people you have paid to do what they would have done for free. Meanwhile some of the lapsed customers you actually wanted back ignored the offer, because it was not the right one for them.

Why discount reporting misleads

Most campaign reporting counts everyone who used the code as a success. It cannot tell the difference between a customer the discount persuaded and one who was coming anyway. The loyal customer who uses the code looks identical, in the report, to the lapsed one who came back because of it.

Choosing who to target is often guesswork too. Rules like "everyone who has not bought in ninety days" or "top spenders" feel sensible but are not based on who actually responds to offers. The top spenders may be precisely the people who do not need one.

Without a group of similar customers who did not receive the offer, there is simply no way to tell what the discount did. Most businesses have never held one back, because it feels like leaving sales on the table. In fact it is the only way to find out which sales the discount actually created.

What untargeted discounts cost

PatternWhat it costs
Discounts to loyal buyersMargin given away on sales you would have made anyway
Customers trained to waitRegulars learn to hold off until the next code
Wrong offer to lapsed customersThe people you want back do not respond
Misleading campaign reportsPromotions look profitable when they are not
Brand perceptionConstant discounting makes full price feel like a rip-off

The customers-trained-to-wait effect is the one that builds up slowly. Once regulars expect a code, full-price sales shrink between campaigns and the business needs ever more promotions to hit the same numbers.

How we target offers at the customers they persuade

  1. We gather customer and order history from your e-commerce platform or CRM, such as Shopify, WooCommerce, Klaviyo, Mailchimp or HubSpot, along with past campaign records showing who received which offer.
  2. Where past campaigns were sent to everyone, we set up holdout groups in upcoming campaigns, a random slice of customers who do not get the offer, so there is a fair comparison to learn from.
  3. We build an uplift model that estimates, for each customer, how much more likely they are to buy with the offer than without it, rather than just how likely they are to buy.
  4. We sort customers into groups: those an offer persuades, those who buy anyway, those who will not buy either way, and those an offer might put off.
  5. We send the target list back into your email or ad platform, so the offer goes to the persuadable group and everyone else gets normal communication.
  6. Every campaign keeps a holdout group, so the true effect of each promotion is measured and the model keeps learning.

This needs some patience at the start if you have never run holdouts. The first campaigns with a control group are what make everything after them measurable.

What the marketing team gets

Campaign lists chosen by who the offer is likely to change, not who is easiest to reach. Reports that show the real extra sales a promotion created, measured against customers who did not get it. Loyal customers who stop receiving codes they did not need.

Finance and marketing also end up looking at the same number, which makes the conversation about promotions a lot shorter.

Over time you learn which kinds of offer work on which customers: free delivery for some, a percentage off for others, early access for the loyal ones who do not need a discount at all. That knowledge is often worth as much as the targeting itself.

Is this your situation?

  • Discount campaigns go to your whole list or broad segments.
  • Many discount orders come from regular customers.
  • Customers seem to wait for the next code before buying.
  • Campaign reports count every code use as a success.
  • You have never held back a control group from a promotion.

FAQ

Frequently asked questions

The questions readers ask us after this guide.

Still have a question?

Ask us directly — a senior engineer will get back to you.

Ask about your project

What is uplift modelling in plain terms?

It predicts the difference an action makes for each customer, the extra chance they buy because of the offer, rather than just the chance they buy at all.

Do we need holdout groups?

Yes, at least going forward. Without customers who did not get the offer, there is nothing to compare against and the effect cannot be measured.

How big does our customer list need to be?

Big enough that a holdout group still gives a clear comparison. Very small lists may be better served by simple tests than a model.

What drives the cost?

The platforms involved, how campaign history is recorded, and how the target lists need to be delivered.

What do you need from us?

Customer and order history, past campaign send and redemption data, and access to your email or ad platform.

Keep reading

More on Problems We Solve

Start here

Tell us what your data should be telling you

Describe the decision you want to improve and the data you already keep. We will give you a straight answer on whether machine learning fits, and if a report or a simple rule would do the job, we will say so.

  1. You tell us what you needTwo minutes on the form, or a message on WhatsApp.
  2. A senior engineer reviews itAnd comes back with questions, a realistic range and an honest view on fit.
  3. Free 30-minute scoping callWe talk through scope, options and a realistic estimate — with no obligation.
Free estimateNo obligation

Talk to someone who builds this

Send a short brief and we will come back with an honest view and a realistic range.

Takes under a minute. We never share your details.

  • Free consultation
  • No commitment
  • NDA on request

Prefer to talk? Book a free 30-minute call →