How To Make Attribution Models Work For Your Business

When I talk to many of my industry friends and ask them which attribution model they use, I get a combination of rolled eyes, shrugged shoulders and very occasionally I will get someone brave enough to answer and they almost exclusively will say last click wins.

In this article I am going to show you examples of 6 types of attribution model you might consider for your business, what the differences are, why last click model is fatally flawed and what steps you can take to ensure you are selecting the right model for your business, based on the situation you might find yourself in.

In the wonderful world of affiliate marketing, which traces it roots back to in the 1990’s because the technology wasn’t up to snuff, the last click model was chosen as the fairest method to reward the seller of a product with the commission published.

In those days, the model was completely acceptable as most e-commerce transactions took the form of one and done. A large percentage of sales were completed in a single transaction. If you were a PPC affiliate, you bought your keyword, someone visited a page (often a direct affiliate link), they bought a product, you made a commission, all neat and tidy.

Last Click Attribution Model

last click attribution model

Last Click Attribution Model

No matter what other stages there were in a journey, the affiliate whose affiliate ID was in the target URL last, got all the commission.

This led to a lot of unsavoury tactics with some affiliates coming up with ways to swap out the legitimate affiliate ID of an affiliate with their affiliate ID and stealing the commission.

Later on, as the end-user buying behaviour changed, the introduction of channels like voucher codes, coupon codes, cashback sites, retargeting became the norm, making the last click model less than satisfactory for a decent percentage of the affiliates who were sending traffic to a merchant.

They were having their commissions “stolen” from under them, albeit the manner in which it was being done was acceptable to the merchant. If you talk to most affiliate networks now and ask who their biggest affiliates are, they will tell you it is voucher code/coupon code affiliates, cash back sites, display media buyers who specialise in retargeting.

The reason this model was so popular was that the technology available couldn’t handle multi-channel attribution modelling.

But, all of that has changed. There are now platforms offering this capability, including Performance Horizon Group (PHG), Impact Radius and later this year Google are rolling out multi-channel attribution modelling to all Google Analytics users. We have access to this for some of our test accounts and it is extremely interesting to be able to compare what-if scenarios to arrive at a suitable model for your business.

Now, if the last click model is the norm, the polar opposite of it is the first click model.

This model basically gives all the credit and commission to the first click into the site.

It encourages affiliates to adopt a scatter-gun approach of trying to get anyone to click, for any reason, to drop their cookie on the end user and then hope that the merchant has plenty of other channels available to it to try to close sales.

First Click Attribution Model

first click attribution model

First Click Attribution Model

This model reminds me of the Abbott and Costello classic – Who’s On First…..

That’s first click and last click discussed, now let’s look at some other alternatives.

Level Click Attribution Model

This model distributes the revenue equally amongst all the parties involved in the customer check-out journey.

So, if someone visited your site from an organic search, then a PPC click, followed by a retargeted banner, then a product review site ad and finally finds a coupon code and checks out to buy your product then each of these 5 channels would receive 20% of the commissions payable.

level click linear attribution model

Linear (Level) Click Attribution Model

This model might seem, on the surface, to be fair and equitable.

The issue is always going to be that different channels will carry more importance or weight and will have more of an impact on the likely outcome.

Many end-users go through stages in the process of their purchase from :

Awareness /Discovery

Consideration / Comparison

Preference / Intention


Often, these stages take place on different sites (some will be on review sites, some will be on social sites, some will be on the merchants site) and can take longer than others. Quite often a clever banner ad displayed can act as a catalyst to turn someone stuck at the consideration stage to move on to the next step. That can and should be reflected in the payout, but in the linear click attribution model this will not happen.

Position Based Click Attribution Model

position based click attribution model

Position Based Click Attribution Model

This model looks at the position in the chain the click took place and rewards (or punishes) based on proximity to the conversion event.

So, it rewards some of the sites currently getting all the glory in the last click attribution model, but it is also recognising the impact that the parties that helped moved the buyer through the funnel and giving them some due reward for their input.

Time Lag Click Attribution Model

time lag click attribution model

Time Lag Click Attribution Model

This model is purely based on time from journey to end.

If your typical buying process lasts 5 days then visitors inside the last day will get more credit and recognition than those from 5 days previous.

Custom Click Attribution Model

custom click attribution model

Custom Click Attribution Model

This model does exactly what the name implies.

The idea behind this attribution model is based on the specific requirements of your business you can customise the payout to allow budget to flow to the channels that add the most value, whilst recognising the parts that others play in the customer journey.

Some ideas for this might include :

Brand terms – Internal search team – low percentage recognition

High-profile blogger writing a review – high percentage recognition

Competitive generic terms – PPC affiliate – high percentage recognition

You can and you should dissect your analytics and evaluate which model will work best for your business.

What alternative attribution models have you used?


Leave a Reply