One of the most difficult questions to answer is “Where should I move my budget to get maximum return?” Typically, this results in a scurry off to analytics to pull out the top revenue earners. This can also result in a very lopsided view though, without a proper look at the attribution mix.
This is because Google’s default is last click which gives you a very skewed outlook on the effectiveness of your integrated strategy. In a world where the integration of all your channels is a proven winner, you can’t simply ignore that integration when reporting and budgeting.
So what can you do? Google Analytics has a great feature that allows you to compare attribution models so you get a better idea of each channels’ performance in the overall marketing mix. In order to use it though, you need to understand the different models available and why they could be important to your decision making.
A brief example:
A visitor journey results in a £100 sale from the touchpoints below:
1) The visitor performs a search and clicks on an organic search listing. They like what they see and like the brand on Facebook.
2) They click on a Facebook post a week later that takes them to a blog demonstrating key benefits of the product.
3) They perform a brand search and click on the paid search brand ad. They bookmark it because they don’t have time to purchase right now.
4) They go directly to the site and make a purchase
Which channel should get the revenue and how much?
1. The Last Click Model
I used to assume that Google Analytics used this as standard, but it actually uses last non direct click. For last click 100% of the revenue would go to direct, clearly not the marketing channel that delivered the visitor or engaged with him/her.
2. Last Non Direct Click Model
In this model, 100% of the revenue goes to the “campaign” prior to the conversion. This excludes direct traffic and is also painting an imprecise picture as AdWords would get 100% of the revenue in the above example. Again, no value is given to the channel that delivered the initial visit or the channel that promoted the engagement.
3. Last AdWords Click
No comment. Why anyone would need this is beyond me; it clearly just pushes you to spend more on AdWords.
4. First Interaction
No better than last click in my opinion. It gives 100% of the revenue to organic search in the example and it clearly didn’t act alone in the conversion. If we adopted this as a rule we could make the decision to axe social as it wasn’t “contributing”. Big mistake and you wouldn’t know it until it was too late.
It’s not as bad as first or last click, but it doesn’t really address the issue. All channels in our example above would get £25. No winners, no losers. Better than the above but in no way right.
6. Time Decay
This gets far better than the previous attempts. The channel closest to the conversion gets a lot of the credit and this reduces exponentially across the previous channels. You can also set the time delay between touch points which will algorithmically affect the outcome, giving more recent touch points more value than older ones. It’s a good replacement for last click if you don’t want to delve deeper but do hate first or last click as much as you should.
7. Position Based
By default, position based gives 40% credit to the first and last touch points and distributes it all evenly between the rest. I do prefer this one to time decay as the initial impression should be worth something – first impressions count, right? Well, in our example I feel social is a bit hard done by as it kept our brand top of mind which is very important, if sometimes intangible. You can of course customise this further by creating a custom model and giving less to the first and more to the middle if you feel that is fairer in your organisation.
So what to do?
The biggest takeaway from this is that no model is perfect out the box, but you should keep researching to develop yours further and use either position based or time decay in the short term while that is in progress. Challenge all reporting that doesn’t show the mix of touch points over time, as all the individual channel reporting that could be separate to your analytics will undoubtedly claim more revenue than it’s due.