Why and how you should A/B test your pricing

Matt Scaysbrook
Director of Optimisation

Why & How You Should A/B Test Your Pricing

Let’s look at one of the central truths of any business:

To survive, every business must be able to change successfully.

History is littered with the cautionary tales of businesses that were unable to do that. Blockbuster, Kodak & Polaroid, to name a few.

If we can accept that change in business is inevitable, we can also accept that that applies to individual elements within a business. And pricing is one of those elements.

In accepting that business change is inevitable, the question is no longer whether we should change, but rather:

How do we change with the greatest confidence & the most limited risk?

This is where testing comes in.

The Challenge to A/B testing your pricing

A recent experience with a prospective client highlighted to us that there are many perceived barriers to A/B testing pricing.

They were a service business with a tiered product offering. So they could not improve their margins by selling multiple products to the same customer.  

  1. Sell more of their higher-tier (and higher-margin) products as a % of total sales
  2. Improve the margin of their lower-tier (and higher-popularity) products

Much of their historical CRO programme focused on the former. They had been warned off of the latter before.

But they clearly recognised two important elements:

  1. Pricing was a key part of the success of their business
  2. Making significant changes to their business (including pricing) was best done off the back of empirical proof of the value of the change

The good news for this prospect was that the barriers they perceived are just that; perceived. As with many things in the A/B testing world, what they could achieve technically outweighed what they had initially believed. This lead to a number of possible routes forward.

The Solutions to A/B test your pricing

#1 Voucher code entry

If you run a transactional website, your ecommerce platform provides the ability to enter a voucher code in exchange for a discount.

But what most businesses don’t consider is that those voucher codes can usually be used to make a product more expensive too! And herein lies the opportunity for price testing via voucher codes.

In its simplest form, your AB test using this method needs to do three things:

  1. Change the appearance of the price of a product / service on the front-end. It’s only a series of text changes, so fairly easy to achieve.
  2. Apply the appropriate voucher code when an eligible product is purchased. This ensures that the end-customer gets charged the appropriate amount
  3. Hide the fact that a voucher code has been submitted so the visitor isn’t aware of the presence of the test

But there are some drawbacks to this method too:

  1. Some voucher code systems will not allow negative-value codes. In short, you can only make things cheaper, not more expensive. So double-check the system you are using before you get too far into this route.
  2. Some voucher code systems are limited in the specificity of product / service that they target. If you can’t make them specific, it can lead to a lot more technical complexity. You may have to work out how a voucher code can be removed when the requisite product is removed from the basket.
  3. Some voucher code systems only allow one code to be used at any one time. So because you’ll be hiding the fact that a code has been added at all, visitors who intend on using another voucher may realise that they can’t. And they won’t understand why. Depending on your system & how widely you use voucher codes, this might be a minor-risk edge-case, or it could be a total showstopper for this method.

#2 Back-end triggered switch

If your website already operates a dynamic pricing model, then this may be a better solution for you. With this model, updating the price in one place on the back-end will deliver that change across the site, from product pages all the way through to sale.

To achieve this, you will need to:

  1. Use your A/B testing platform to set a specific variable (often a cookie) to identify visitors who are in your B experiment
  2. Create a back-end setting to update the price of the product / service whenever that variable is detected
  3. You may need to apply the front-end text change to any prices as an interim measure before the back-end switch kicks in on the next request. This will help you avoid having to force a refresh of any page to reflect the new price.

The drawbacks of this method:

  1. The most obvious is the dynamic pricing model. Not every site will have this, so if you don’t, this probably won’t be a good option
  2. The requirement for back-end access can also be a challenge. If you’re a third-party you may be restricted to front-end changes via an A/B testing platform. So you’ll likely need your client to make the back-end switch for you.
  3. If you have a lot of landing pages that show the impacted price, there will be a lot of prices to change on the front-end to avoid the forced page refresh. Not an insurmountable challenge of course! But more variables = more time to execute & a greater opportunity for things to go sideways!

#3 Duplicate product creation

If duplicating & editing an existing product on your site is a simple task in your CMS, then this may be the route for you.

What you will need to do is:

  1. Duplicate the product you want to test the price of, and alter only the price itself
  2. In your A/B testing tool, use your usual front-end changes to alter the price of the original product to mirror the price of the duplicated product
  3. For any Add to Basket CTA for the original product, alter the action to add the duplicate instead of the original

As with the other options, here are the drawbacks:

  1. Duplicating a product / service isn’t simple, especially if they are not CMS-driven - check this before going too far down this path.
  2. If you’re looking to test across a lot of products, there will be a lot of duplicates floating around. So your CMS might look pretty messy for a while.
  3. If you’re looking to test across a lot of products, there will be a significant number of front-end changes to be made.

To sum it up…

Let’s bring it all together:

  1. You will need to change your pricing at some point - that’s inevitable
  2. You will need some means of mitigating any price-change risks when the time comes
  3. Knowing how you could achieve this before you need it could be a major bonus
  4. If you find out that you could charge 10% more with no loss in conversion, you will be hailed as a company hero!