Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
  1. Home
  2. Blog
  3. How to Test Prices Responsibly
AI & Machine Learning

How to Test Prices Responsibly

Price experiments risk customers seeing different prices for the same thing. Designs that produce evidence while staying defensible.

Updated 2 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

Showing different customers different prices for the same product at the same time carries real reputational and sometimes legal risk. Time-based, region-based and product-based designs give usable evidence with far less exposure.

The obvious design is the risky one

The cleanest experiment shows half of visitors one price and half another. It also means two customers can buy the same item at different prices on the same day, and one of them can find out.

Customers react strongly to this, more strongly than to a price rise applied to everyone. It reads as unfairness rather than as commerce, and the reaction is not proportionate to the size of the difference.

There may also be legal constraints depending on jurisdiction, sector and whether the variation correlates with protected characteristics. Take advice before running one.

Designs with less exposure

DesignHow it worksTrade-off
Time-basedPrice A one period, price B the nextConfounded with seasonality and events
Region or storeDifferent prices in different locationsRegions differ; needs matched comparison
Product-basedVary price across similar productsProducts are not identical
New products onlyTest at launch where no prior price existsLimited to launches
Promotional depthVary discount rather than base priceMeasures promotion response, not base elasticity

Time-based is the most common and needs the most care, because anything else that changed between periods gets attributed to price. Alternating in short blocks rather than one long switch helps considerably.

What you are actually measuring

Units sold is not the objective. A price cut that increases units while reducing total profit has made things worse, and a test measuring volume alone will conclude the opposite.

  • Gross profit, not revenue and not units
  • Effects on other products - customers trading up or down within the range
  • Whether demand was brought forward rather than created
  • Returns rate, which sometimes shifts with price
  • Any change in customer mix, which can affect future value

Cross-product effects are routinely missed. A successful price cut that simply moved customers from a higher-margin alternative is not a success.

Elasticity is not a constant

Businesses sometimes measure elasticity once and treat it as a property of the product. It moves with season, competitor pricing, customer mix and how the price is presented.

Treat any estimate as valid for the conditions it was measured in, and re-test periodically. A figure from a test run two years ago during a promotion is not a reliable basis for a decision today.

Be prepared to explain it

The practical test for any price experiment: could you explain this design to a customer who noticed, and would it sound reasonable?

Time-based and regional variation pass that test - prices change, and they differ by location, and customers understand both. Simultaneous personalised pricing generally does not, which is a reasonable signal about whether to run it.

If you would not want to explain the design to a customer, do not run it.

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

Is personalised pricing legal?

It depends on jurisdiction and how it is done, and it becomes clearly problematic where it correlates with protected characteristics. Take legal advice first.

How long should a price test run?

Long enough to cover a full purchase cycle for the category, and to average out weekly patterns. Short tests on infrequently bought items measure almost nothing.

Can we test on a small set of products?

Yes, and it is a sensible way to start - though results from one category may not transfer to another.

What if competitors change prices mid-test?

Record it. It is a confound you cannot prevent, and knowing when it happened lets you interpret the result honestly.

Keep reading

More on AI & Machine Learning

Start here

Want machine learning project details from us?

Tell us what you are trying to predict and roughly what data you hold. We will come back with an honest view on whether machine learning is the right tool, what the work would involve and a realistic cost range. If a spreadsheet 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 →