Which Attribution Model should I choose?
As previously stated in earlier posts. There are different attribution models to choose from. In earlier posts, we have introduced you to lineal attribution model. Let us now, review the various models that Analytics provides us with:
As you can see, the picture is self-explanatory. Therefore, which one is best for you on a daily basis? Why is truly important all things considered? Let us explain to you why.
How to know each channel’s influence
Unfortunately, the data provided by Google Analytics is not as spectacular or conclusive as that provided by other tools. Here are some screenshots, just so you know this isn’t a (very) biased opinion…
Do you think this is consistent data? Those are advanced optimizations?
Even more likely is you’re still in shock from so many “Not Available” messages. Why do we need to know about conversions per route? Is this information really useful? We must be very careful to avoid one of the dangers of being over-informed: paralysis. This means if we get bogged down by information overload in the form of numbers, the reaction can be inaction, or paralysis. Fortunately, there are technology and tools are available for us to interpret data in a more clear way, allowing us to make decisions and take action in order to achieve our objectives.
——————————————— (arriba para el post anterior)
But, even if you choose to keep working with Google Analytics, you can still get useful information. For example:
- How much influence does Affiliate Marketing have over PPC campaigns?
Let’s create a “Segment determined by the user”, such as this one:
How much does the SEM influence the SEO?
The result would be:
In other words, more than half of the SEO sales have been helped by SEM traffic, and so on throughout each successive channel… Wouldn’t it be better and simpler for everyone if information could seen in this way? In the following example we can see that 63 out of 200 affiliate sales have been obtained due particularly to Adwords Search. As well as 19 out of 351 Bing sales come from Adwords and so on.
Number of clicks per interaction: We’ve just seen the number of conversions per interaction. But, how can we see the number of clicks per interaction?
A tool such as Adinton can easily give us a table where it shows the amount of clics needed per conversion:
See how the percentages only drop considerably after the seventh click. At least, this is how the commercial version of Google Adwords is sold. In other words, those of you who have spoken with Google Adwords salespeople know that when they present their Search remarketing option, their sales message is that more clicks means more chances of completing the sale, since clicks mean the user is interested in our product, bringing us closer to the sale. This table shows us the number of clicks, conversions and % of conversions, according to the order of clicks. In other words, we have 1366 conversions that were the result of 22,323 clicks, and each conversion was the result of 3 clicks. Here are some other interpretations of the same data. Out of a total of 1,156,000 clicks, 991,000 clicks were the first click, 30,111 of them resulting in conversion, 82,904 kept on clicking. The other 878,000 did not return. You might be thinking where we really judge the success of our online marketing campaigns is on the first click, CRO pure and simple, and not on attribution models.
Optimization actions based on attribution model analysis
We are paid to improve the ROI of our company:
Sales Increase +Cost Control =improvement in ROI
Since we have an attribution model system working for us, we are going to start from there.
In the second column we can see that if:
And the number of clicks equals 3000 and we want to keep the CPAs value at €35.96, because we want to sell this CPA at a maximum value.
Consequently, if we’re talking about Adwords, if we increase the cpc amount up to 3.78€, then we will be able to:
- Improve our average position
- Improve CTRs
- Improve QS
- Increase the number of clicks
- If % conversion is maintained
- We will increase sales
- To a CPA we want to work with.
If this data had been lower than the CPC we were already working with, we would have had to lower the CPC.
We have an attribution models system, or at least Google Analytics has put one into place, for the purpose of controlling the full customer/user journey up to the time of purchase. But, what happens to those users who already made purchases or have converted, and continue to click on our ads?
It would be useful to measure the number of post-conversion clicks. We have noticed that these numbers can be either low and so below 1% of the total amount of clicks. Or, It can be a number higher than 10% of the total amount of clicks, which in this case, would be alarming.
What is the probability that that percentage lies closer to 1% than to 10%? The answer lies on the product or service you are selling. For example, if you are a hosting or an airline company, it may be that after a conversion your client needs to return to your website for information or documentation.
A tool such as Adinton allows you to:
1. Know exactly how many post-conversion clicks you have
2. Adopt measures, once we know that the user has become our customer, we could:
a. Send them to a different landing page in order to encourage cross-selling, or to enhance Customer Service.
b. Create an inverse remarketing campaign so we no longer appear in Adwords (Search & Content) for this user for a certain number of days.
Optimizations to the time of the click:
Looking back at the Clicks and conversions Table by order of clicks, we wil add a last column with the Conversion Rate by user:
Keeping in mind that in order for conversion to occur on the 2nd click, the first click has to be made. In other words, a user that converts on the 3rd click:
In other words, according to Google, the conversion rate of users that reach the 3rd click is 100%. Our statistics separates Conversion Rate by user, therefore we got a Conversion Rate (User) of 33%. Imagine how this simple change in perspective could affect CPCs, CPAS, etc. Therefore, there could come a time, after the 3rd or 4th click, that we would simply choose not to appear or to send the users to another landing.
To conlcude, a tool such as Adinton would allow us to create rules such as:
If the user is doubtful and has already clicked 3 times on your ad, the 4th click would:
- send him/her to a landing page with a huge discount.
- send him/her to a landing page that would at least get his/her email (Growth Hacking)
- make your ad disappear, with Inverse Remarketing
Every day there’s more competition, more professionals and companies that say that they “know” about online marketing. Every day there is more budget available and every day there are more analytic tools. We keep investing in more channels: SEM, FacebookAds, TwitterAds, GmailAds, RTB, Remarketing, banners, email marketing, posts, etc. The key to success lies on understanding that marketing campaigns do not function on an individual basis, but rather as a unit. Even when the user goes “jumping” around from one site to another, he/she should end up on our website, and that is where we must be able to convert all users that enter with better success rates than our competition.
We have seen that a change in the attribution model can help us increase our CPC, and therefore produce more sales, be more competitive, etc.
Our aim must simply be:
- Take accurate and organized measurements.
- Give every metric its fair value and not be swayed by trends. In other words, if we say that 50% of our conversions have required multiple clicks, we work with different acquisition sources, which is why it makes sense to carry out an attribution model study. Likewise, it makes sense to analyze those conversions that have only required 1 click for conversion.
- Make decisions, take action. Do not let information overload paralyze you.