Why Attribution Modelling Needs to be on your Radar

Traditional tracking, how it works

The beauty of digital marketing as opposed to offline marketing channels, has been the ability to tack and understand how every penny you invest is working successfully.

In the early days of tracking the main way to actually understand what channels of marketing were working was through redirects. Simply apply masking URL or a query string to the end of your URL and this information will be passed to your analytics account.

For example:

3rd party tracking platforms used ways where the whole URL is masked with a redirect e.g. DoubleClick, Atlas, etc.

Although this tracking showed a lot of information and detail, it was missing a vital piece, how was each channel of marketing influencing each other?

For example:

  • I am actively looking for a “Samsung gear 2”
  • I go to Google and type in a search. The results are mixed but I click on the top Organic listing gadgetsRus.com/Samsung_gear as I am confident this will direct me straight to a relevant page. I review the product and question if this is the cheapest price on the market.
  • I go back to Google “buy a Samsung gear 2” I then see several sponsored PPC adverts for the gear ad and click through to each to compare prices and delivery time.
  • A few days later whilst looking for reviews on Samsung gear – I am delivered a retargeting ad from gadgetsRus.com
  • I then do a search on pay day for “Gadgets-R-us Samsung Gear”, the organic listing comes up and delivers me the relevant page and I complete a purchase.

The user path for this would be:

Organic (non brand) > PPC (non brand) > Retargeting > Organic (brand)

This is just one user experience, you will find a plethora of ways that people search so the key is making sense of the data.

The challenge arises when we try to attribute the sale. It was the organic (brand) search that finally converted me to a sale, however had I have not seen the first organic result, I would not know gadgetsRus.com sold the gear.

When I compared prices, gadgetsRus.com needed to be on my radar and achieved this via PPC, so surely these earlier clicks helped lead me to buying from the website?

The second to last interaction was retargeting, also an important reminder to prompt me to purchase from the brand. Finally I searched for the website and found them on a brand term where I had already made my mind up to purchase directly from them.

A complex journey and just one example of one individual. So what does Attribution modelling do?

Attribution, firstly is the allocation of the conversion to the click. So if we felt the last interaction was responsible for the sale, we would attribute 100% of that value to the last click, in this case the Organic (brand) search. But that doesn’t always make sense. The first interaction was the touch point where I realised gadgetsRus.com sold the product and then became the top site to purchase from. Without that interaction I would not have that website on my radar. So therefore I should allocate a percentage of the sale value to the first interaction.

Finally without the other steps in between I may never of come back to that site, so they should also be allocated some value, even if that value is not as high as the first or last interaction.

In simple terms we are trying to apportion a percentage of the sale to each step of the marketing mix. Brand terms are always going to have a high percentage of last interactions converting, however something like display advertising is awareness building and therefore may be a first interaction but not often a last interaction unless it is retargeting or remarketing.

So what is Attribution Modelling?

This is a way of testing and applying a model which demonstrates that user interaction and attributes the value accordingly based on your business model.

This may vary from business to business. If you are selling high end furniture, your lead to sale time is likely much higher than a site selling mobile phone cases and accessories. One is a fairly instant purchase and the other more considered. They may therefore have different attribution models. An example may be this:


First interaction –  “walnut wood dining table” (PPC click- 30% attribution)

Second Interaction – “bespoke design walnut dining table” (PPC click – 10% attribution)

Third Interaction – Remarketing (Display click – 10% attribution)

Last Interaction – “(Brand) walnut dining table” (Organic Click- 50% attribution)

Using attribution tracking I can run reports to see if I attribute all value to first interaction – how this compares to last interaction attribution.  This can help me decide what percentage allocations I could start with, but this is not an exact science and requires understanding, development and reapplication of learnings.

I may then apply this model across all my interactions:

  • First interaction =30%
  • Last Interaction = 50%
  • Other interaction =20% split by the number of interactions.

In monetary terms if the sale was for £6,000:

  • First Interaction = £1,800 rev
  • Last Interaction = £3,000 rev
  • Other interactions= £1,200 rev split but remaining interactions.

It may be that over time, I start to attribute a slightly higher percentage to the first interaction, or less to the last as I understand the data more effectively, as the model adjusts, the results also change.

The bottom line is through attribution modelling I am taking a much more fair and intelligent view of my holistic marketing strategy. Understanding what is working and what isn’t.

Rather than dismissing PPC non brand terms as being ineffective, I am actually able to allocate a value to them and understand what value they may actually have. If they were not running, I may have saved budget but would I have been as likely to convert a sale without them?


Attribution modelling can take into account all platforms of marketing if you have interaction data and understanding of the user paths. The ultimate attribution is to understand what offline advertising value is to digital in a clear and concise attribution model, but to track the whole user interaction path is somewhat of a challenge.

Rachel Mepham
With over 15 years’ experience in Digital Marketing, Rachel heads up the team at Digital Clarity. With a deep skill set in the Paid Search, Social Media and Analytics, Rachel is regarded by both clients and peers as one of the most experienced and prolific women in the UK digital space. Her approach and application to digital strategy planning has been used by some of the biggest brands as well as leading advertising and marketing agencies.