Article Customer Acquisition

Why affiliates are breaking your attribution reporting (and how to fix it)

When brands need to work out which marketing channels are driving spend, affiliate programmes consistently rear their head as a huge pain point. They typically barge into the sales process, ‘claiming’ sales regardless of whether they had any impact at all.

This was a hot topic back at the DCA’s Silver Market Day and was raised by Nigel Swabey of Scott’s of Stow. Nigel cited that plugins like Join Honey are used to remind people to use voucher codes at the point of sale (even though the person may buy the product anyway). This means that the affiliate programme is eroding margins, rather than generating new sales. It’s a topic we have revisited many times since.

To compound the issue, the standard methods of data-driven attribution modelling either overestimate or underestimate the impact of affiliates on sales.

Brands are then left unclear on the effectiveness of their channels, let alone where to prioritise spend. We have seen this frustration time and time again, and it inspired us to create a tailorable attribution tool that works.

How affiliate programmes work

Affiliate programmes offer vouchers and discounts that can be used on retail sites. Some offer voucher and discount codes that can be input at the point of purchase; and others simply require an impression (page view) in order to ‘claim’ the purchase.

Some of these voucher programmes are particularly aggressive. They offer plugins so, when you are buying from any shop, they will tell you a voucher code you can use.

Hacking the sales process

The problem with affiliate programmes is that they often claim responsibility for sales that would have gone through regardless of their involvement – leading to misleading attribution.

For example, if you were driving sales with a direct mail or social media campaign – a previously installed voucher plugin could offer a voucher code right at the point of sale. In this scenario the affiliate adds no value (in fact, unnecessary discounts erode your margin), but claims all the glory.

You may find that a signification proportion (if not all) of your sales have a last-click attribution of affiliates, when you may know intuitively that this does not reflect the true customer journey.

Why standard data-driven attribution doesn’t work

There are a couple of problems with automated attribution modelling tools such as data-driven attribution. They have to look at all data points, and they need an automated way of looking at this. This means that the model often overlooks the experience and intuition that people can add.

The two standard methods of data-driven attribution modelling are time decay and uplift modelling, and both have their shortfalls – either overestimating or underestimating the return on investment (ROI) of affiliates.

Time decay will weight affiliates strongly because it looks at the most recent activity (and people tend to search for voucher codes just before purchase). This approach assigns all value to affiliates.

Uplift modelling will struggle to determine what would happen if affiliates were not present. For example, did the customer buy the product because they found a voucher, or were they going to buy the product anyway and just wanted to ‘chance’ the voucher for a lower price? This approach often struggles to assign any ROI to affiliates.

Understanding the true value of a channel through testing

The best way to understand the true value of affiliates is to set structured tests. This could be through a detailed analysis of the user journey, or through turning affiliates off and on (a bold move if a high proportion of your site uses them).

Once you have done your test, you should be able to derive an assumption, which can be fed into your model.

Alternatively you can protect that budget, leaving it out of your model and reporting on it separately. Many marketing managers do this for other channels like SEO – so if vouchers are a required element of the sales process, they become a cost of sale rather than marketing budget.

This approach can also be used for other channels such as retargeting, or testing the outputs of a marketing mix model. The reality is, there is no single best attribution tracking approach. You need to combine multiple approaches to gain a true understanding of your marketing budget.

A better way forward

If you are looking for an attribution tool, it is important to choose one that can be customised to your business. Black box attribution tools rarely cut it, because they provide very little control over modelling.

We built WebFusion for this exact purpose, to provide accurate attribution modelling for even the trickiest of channels – including affiliates. WebFusion allows you to adjust the approach to fit your business, and apply assumptions and knowledge from other sources.

If you are frustrated with inaccurate attribution modelling and feel unclear on where to focus your marketing spend, we can help. Get in touch on and we will be happy to share details of our approach and modelling.