In this first post, I want to outline the basic idea for what deepdive should be and what kind of issues it adresses.
By now I have worked for over 8 years with webdata. I setup consent management and tracking, enrich the data using external sources and visualize it in some dashboard. I also analyzed said data and acted upon it directly myself in my capacity as a Marketing Manager in Google Ads.
Throughout all this time, I felt deeply frustrated by the lack of coherence and accessibility from all of the various tools one has to use to accomplish such setup for any website. The multitude of tools is not only costly, but extremly work-intense for something that is the same for millions of different website owners.
Let us take a simple, very standard scenario. We have a website and offer some products. To direct people to our site, we use marketing campaigns on various channels, such as Facebook or Google. We setup campaigns and spend money, and indeed see people clicking on our ads to visit our website.
Here already we have the first obstacle: While we can see the users coming onto our website, we cannot see from what campaign they came from.
To do so, we must setup our UTM-Parameters in the respective marketing channels – but there is little to no guideline what parameter to use for what. Instead this must be figured out by each marketing team for themselves, and in cases of some analytics apps, support for this it must be installed seperately.
Now that we have hopefully correctly parameterized this, we can see our users from our campaigns in our analytics app and see what they visited and purchased. In our marketing system that we used (like Google Ads), we can see how much we spent and other metrics – but what we cannot do, is bring those figures together and generate figures such a ROAS (ie Revenue / Cost).
To do so, we must either manually write both numbers for each campaign into an Excel sheet every day (not fun if you have dozens of campaigns on multiple channels), or rely on an external third party tools that loads data from both, then manually configure the tool to merge the data sets, to then use yet another third party tool to visualize said data.
But now we do all this – we still only see if the campaign got users to buy instantly something. But if the users keep coming back after that first initial purchase, or perhaps susbcribe to a product, their real revenue is actually higher (Lifetime-Value). Even with our complicated setup, we are still out luck. While our analytics app does track the transaction id of each order, it does not support any shop system whatsoever. And neither do those third party tools. So to pull this data, we must rely on yet another third party tool such as TrueROAS, just to have this single, but vital number.
There are many such infuriating and frankly unnecessary obstacles at every corner.
If at any point whatsoever any faulty data entered your data stream (and it will!), it cannot be deleted or corrected (unless you are prepared to spend 150.000$ minimum per year on Google Analytics 360, or feel confident in fiddling via SQL statements in a closed off database without any documentation).
There is plenty of known bot traffic (like gtm-msr.appspot) that generates huge amounts of wrong conversion data but are not at all excluded – and remember, those wrong conversions cannot be deleted anymore!
Pages are stored together with their parameters, so websites that use unique hashes will result in millions of page hits with each a single visit (and no way to group them without their parameter). Have fun analyzing!
… and many, many more.
deepdive sets out to exactly do these things differently.