If your company isn’t selling a product where the customer does any research online and you are using last-click attribution modelling, then you may have fallen into the trap of no longer marketing. You are buying space for when the customer has already made their decision. If you feel this might be something you are doing this article is important to you and your business.
Last click modelling takes whatever the source (PPC, SEO, social, organic, etc.) was that the session started on and gives credit to that source and furthermore that click.
Let’s ignore digital for now and create an example pre-digital with Peter the Predictable Buyer, and his journey buying Producto (Showing my Creative Side)
1. Peter sees a billboard about Producto, barely pays attention
2. Peter sees a TV ad about Producto, mutes it and carries on his life.
3. Peter hears about Producto barely listens.
4. Peter sees billboard again and takes it in.
5. Peter sees a tv Ad again and actively talks to others about Producto
6. Peter gets a mailer about 20% off for first-time buyers of Producto
7. Peter heads to the shopping mall to buy Producto
8. Peter gets a little lost, sees an Employee going to lunch wearing Producto outfit and asks for directions.
9. Peter buys Producto
Let’s give credit to the marketing that Peter came across.
In last click modelling what we see is that the employee that directed to Peter to Producto would receive all the credit for this sale. Is that right?
We should be affecting everything from purchase decision and above, but I would strongly argue that getting someone from information search to the final part to purchasing is much more valuable than taking them at the purchase decision level.
So now let’s take this back to Digital Marketing and Last Click Attribution Modelling.
If you give credit solely to the final thing in the list, what you end up doing is creating an Adwords account that congratulates the person holding the door open to the store. Don’t delude yourself that you are marketing to the right people.
What happens is a business owner, wants more bang for his buck. What happens is the digital marketer optimizes for the best CPA. They look at their stats and of course being in the earlier positions returns a low CPA, because people still haven’t reached that evaluation area. The CPA is amazing at the door holding stage of the buyer decision-making model. What will the digital marketer do when pressured for a better CPA? They will get out there less and spend more time on the door. Even if the marketing they were spending at the higher level was bringing the customer to the door later.
I can’t mention how many times I have got an account where the marketing spend is almost completely spent on Brand name keywords. Where the client is telling me, that it was working, to begin with, then as they optimized the CPA it got exceptionally better for Adwords, but not for the business.
Changing the attribution model is a start, but at its basis, at the moment, the answer has to be it’s not enough. The reason being is that there are too many holes. Cross-device is too poor, people walking into store, session data, and people searching in incognito and so on.
Lots of digital marketers, fall into the trap that their numbers are 100% correct. Reasonably we have to assume that we can follow a number of conversions, but we will lose out on a number more. Meaning that our CPA data is poor and incomplete no matter what attribution model we use. (I will be releasing an article on how to combat this problem next week)
This being said, changing the attribution model is not a waste of time. What should we change it too? Data-Driven Attribution (DDA) Model, is by far and will be the strongest attribution model out there, but it is only available after reaching a limit of spend and conversions. DDA finds what the most important factor was on Peters decision making process and weights credit accordingly. If DDA is not available you have the choice of time decay and position based. I prefer position based, as it gives you an overview of the process that they went through, I feel this model allows you to grasp more of the entry and exit point of a customer. Then you can fill in the gaps.
It could be argued that Peter would never have found Producto and would end up buying a competitor if it wasn’t for the final step. Having a spread of high funnel and low funnel terms allow you to capture the full market.
The only way to truly judge performance is to work closely with the owner of the account, assessing the growth pre and post digital marketing/digital marketing changes.