It’s very interesting how a new methodology, business model, or whatever else could impact the people that they, in most cases, adopt it without even analyze it.
On e of the most mentioned word, within the analytics industry is Funnel. The market adopted the funnel model but we all really know why this model could be useful for analytics proposes? It’s only useful just for analytics proposes? When could we use the model and when not?.
First of all I find deeply important to know a bit about the throughput accounting model (you could find more information in this blog about this model). The above mentioned model analyze the flows through a system or network, identify the bottle necks and determine the throughput (quantity that could flows through a particular system or network).Once it identify the throughput focus all the available resources to solve the more restrictive bottle neck, this allows focus all our resources in just one at the time.
Then the idea is solve the bottle necks one by one, allowing a broader flow of work through the system or network.
Well, even when throughput accounting is a topic I find amazing, this post it’s about funnels so let’s talk about funnels. A funnel is a graphic representation of a process, that even when it’s normally used for conversion processes, it could have different steps and goals. When we talk about performance of a define funnel we are talking about nothing more than the throughput of that particular process. For instance, if we have a funnel defined by the following steps:
1- A banner is displayed into a publisher’s site.
2- Users click the banner.
3- The user lands into your site (even when it seems to be the same step as number 2 it provides us with a very rich information, I’ll mention it later)..
4- The user fills the form and register into your database (this is the conversion or goal in a typical lead generation campaign).
To obtain the performance for the mentioned system we should design the system into a graphical way (funnel) and measure the flows through it, as follow:
1- Impressions: 1,000,000.
2- Clicks: 500,000.
3- Pageviews (landings): 460,000.
4- Registrations: 2,000.
Well I think at this time is important to tell you about two difference from the theory of throughput accounting and the particular way I do it for this cases. Analyzing funnels we shouldn’t find always the most restricting bottle neck but the one that shows the highest restriction compared with the preview step and not with the total model. Not so clear?. Here I go again, if we analyze a factory production, what we should ensure is that the entire system is flowing the same quantity of product, or the closest possible than that. In a funnel model, we could do it that way, but I recommend analyzing the difference between steps and solve the more restricted step. Let’s see an example with the above mentioned data:
1- Impressions 1,000,000.
a. CTR: 50%
2- Clicks: 500,000.
a. CTR Path Efficiency: 92%.
3- Pageviews: 460,000.
a. Conversion Rate (4%).
4- Contact form (thank you page).
In the above mentioned example we have a campaign with a CTR of 50%, a CTR Path efficiency of 92% and a conversion rate of 4%. It clearly shows that something happen with those 460,000 users that as soon as they arrive to they arrive to the form page, they leave and do not register. This really help us be focus on the most restricted bottle neck, the website could no generate leads, why? Well that’s another point, at least we know that something is wrong and where. In this particular case one possibility could be analyze the users experience in that lading page and at the form, another is taking into consideration if we have correctly chosen our target market, or user profile.