Data Department built dynamic pricing for an e-commerce retailer. The result was about 4% higher profitability. This page is deliberately short: we report the result we can stand behind and describe, separately and in general terms, how we typically run pricing projects of this kind.
The decision
Every online retailer decides, product by product, what price to show today. Set it too high and sales fall. Set it too low and margin is given away. Dynamic pricing uses data to make that decision more often and more consistently than manual price lists allow.
For this retailer, the result we report is profitability, the measure that matters, rather than revenue or volume.
What we built
We do not publish the client-specific details of the model here.
How we typically run such projects. The following describes our general approach to dynamic pricing projects. It is not a description of this specific project.
| Step | What usually happens |
|---|---|
| Agree the metric | Fix the business metric on day one, usually profit or margin rather than revenue |
| Choose the scope | Select a set of products where price changes are allowed and measurable |
| Model demand response | Estimate how demand reacts to price, using sales and price history |
| Set guardrails | Agree minimum and maximum prices, margin floors and brand rules with the commercial team |
| Measure fairly | Compare against a control group or a clear baseline, so the result is not just seasonality |
Results
- ~4% higher profitability.
We report profitability because it is the metric that matters. A pricing model can raise revenue while destroying margin; that is not success.
Lessons
These lessons come from how we approach pricing work in general.
Optimise for profit, not volume. Lower prices almost always sell more. The question is whether the extra sales are worth the lower margin. Agreeing on profitability as the target keeps the project honest.
Guardrails build trust. Commercial teams adopt pricing models faster when they set the limits themselves. A model that can never price below a margin floor is easier to approve than one that might.
Measure against a fair baseline. Without a control group or a clear baseline, it is hard to tell a pricing effect from a seasonal one.
You can read more about our approach on the dynamic pricing service page and see how pricing fits with forecasting and customer analytics in retail and e-commerce.
Frequently asked questions
What result did the dynamic pricing project achieve?
The project achieved about 4% higher profitability for the e-commerce retailer.
Why measure profitability instead of revenue?
Revenue can rise simply because prices fall. Profitability shows whether pricing decisions actually improved the business.
Is the method on this page exactly what was used for this client?
No. The method table describes how we typically run dynamic pricing projects. We do not publish the client-specific model details.