AB Test Experiments

8 simple AB test experiments to boost your referral metrics


Life is full of choices. Especially at this time of year.

Wine or champagne? (Champagne obviously).

Yule log or trifle.

Ant or Dec.

Whatever floats your boat.

As marketers, we mustn't be presumptuous about what our customers might like, particularly when it comes to referral marketing. As the saying goes “one man’s meat is another man’s poison”. Thank goodness then for AB testing.

Our referral platform enables AB testing of all elements of a refer-a-friend campaign. Often, something that works well for one brand, can have the reverse result for another. So it pays to plan out a series of tests to establish what resonates best with your customers.

Our platform has delivered hundreds of AB tests for our clients over the past few years. Although these aim to deliver an increase in the number of referrals, they can also have an impact on a number of marketing metrics, including brand share rate and AOV. (in fact, we’ve recently updated our dashboard so that clients can set the objective for each AB test they run).

As our festive gift to you, we thought we’d share a selection of successful AB tests that our clients have experimented with over the past year. They could prove food for thought.

Incentive tests

1. Discount type

A common test among our clients is to test a % discount vs fixed value discount.

Test  against fixed discountAt first glance it seems a simple test, but more than twice as many new customers were generated from the % compared to the $ off. Plus a % discount can often increase the AOV. In this case the AOV was 46% higher.

Indeed we have run tests for clients, where a fixed discount generated more new customers than a % discount, but the % discount significantly outperformed the fixed discount in terms of driving AOV.

Knowing what you want to achieve from your experiments is crucial in defining success.

2. Discount level

If a percentage discount works better for your brand, then the next test might be to see what level of discounting performs better.

Test percentage discount levels

In this experiment the 15% offer drove 47% more new customers. Whilst this might seem an obvious conclusion, it can depend on the level of discount offered. In research we recently undertook, some customers suggested that too high a % discount was viewed with suspicion as it was likely to require a high spend value.

In some brands referral performs so well that they can look to optimise the margin by decreasing the amount they give away to measure the impact on referral. Our client Bloom & Wild did this to great effect.

3. Free Gift

Referral doesn’t have to involve discounting. In this test, cosmetics brand Benefit tested different free gifts as rewards.

Free gift incentive

In this test, although far more customers shared the eye cream offer with friends, over twice as many friends went on to become customers by redeeming the make-up bag offer. So the offer that appeals to your existing customers may differ from that which appeals to prospective customers.

Copy Tests

4. Referrer vs referee lead

Are your customer the generous kind, who like to share something special for their friends? Or are they more likely to be driven by an incentive for themselves?

This is a popular test with many of our clients.

Self interested vs generous offer

In this case, leading with the reward for the friend resulted in more customers being inclined to share the offer.

Customers are often more comfortable to recommend something if they are seen as being generous to their friends. We talk about this concept of social capital in our blog post on The Psychology of Referral.

5. Concise copy

Another great test is the amount of copy used in the referral message. In this case the client undertook a test where one version had the share message pre-filled and the other was left to the customer to complete.

Test pre-filled copy versus blankDoes giving the customer flexibility to write their own referral message improve results? Actually no. The version with the pre-filled text generated a third more shares. And in fact, across all clients, we have found that whilst customers can always change the content of the message, over 90% do not bother.

However, whilst the blank field drove less shares, it did give the brand some useful insight on what their customers really thought of them.

Image tests

6. No image

Whilst most of our fashion clients use impactful images which appeal to their customers, there are some categories where product images aren’t necessarily so eye catching. In this example a health brand tested a background image with a text only share message.

No Image test

Surprisingly, the text only message outperformed the images based message, driving more customers to share the brand and in turn, more friends to become customers.

Share Options

Whilst the focus for referral marketing is often on the incentive. Sometimes small tweaks to layout can make a bit impact. In particular, the sharing messages and how they are presented. Seemingly small things, but often impactful.

7. Number of share options

In this test, the first group were presented with 4 different ways to share the offer. The second group had a stripped down version with just email and share by name options. Proving the theory that sometimes less is more, the version with less share options resulted in more customers sharing the offer, and 20% more introduced friends becoming a customer.

Test number of referral share options

8. Layout of share options

Client Biscuiteers have continued to optimise their programme by running numerous tests. One test this year was the layout of the share options - testing vertical compared to horizontal layouts.

Horizontal vs vertical share layout

Despite being a small change, the vertical share options drove 18% higher customer acquisition than the horizontal version.

So there you have it. 8 simple tests which delivered significant improvements. If your referral offers were stuck in a rut in 2017, maybe these experiments will inspire you to build a testing strategy and identify exactly what resonates with your customers.

You might have a few surprises. Maybe your customers are more Cava and less Champagne. Heaven forbid!

Learn more about how outsourcing refer-a-friend with Mention Me platform works. Request a demo

Guide to a successful referral strategy

White paper

How to launch a successful refer-a-friend programme

Download the guide

Become a referral expert

Blog subscribers get occasional emails with our latest learnings on all things referral.