Imagine you have a little nephew who comes over to your place one day and breaks a cup. You don’t really mind. He then comes over a week later and breaks a plate. It’s okay, all good. The week after that he comes over and again, your favorite vase lays shattered on the floor. So as the next week approaches, you hide all your breakables– or you keep your doors locked and politely ask your nephew to remain outside. This is precisely what retroactive data analysis is. You gathered information from your past experiences with your nephew, analysed it, made an informed deduction (that he would likely break something again), and you took the essential steps and created a strategy that would benefit you, and your home. Therefore, people wanting to understand you would analyze the information from your past and uncover the role your nephew played to bring you to your current position in which you lock your doors and scorn at little children.
Let’s now dive into a more technical explanation. Essentially, retroactive data analysis is the process of resorting to historical data in order to determine how it impacted various outcomes.Retroactively analyzing data equips you with a baseline for future measurement and enables you to optimize your strategy.
Almost every data analyst has gone through a truly traumatic situation in which they are expected to materialize valuable insights from data that is almost non-existent. For instance, the company may not even have any effective tracking system in place. However, the analyst does not have to resort to crying and screaming while performing this highly arduous task– he can resort to automation.This is where artificial intelligence intervenes and relieves extreme stress for the analyst by automating most of the process.
The main issue is that Google Analytics does not retroactively register most of the changes.
For instance, if you are trying to implement a different custom dimension or are intending to enable content grouping, you will not conveniently have access to historic data. Unless the particular data point you require is being tracked, it will not be sent to Google Analytics, which in turn creates a gap.This is why data collection and measurement planning are critical aspects of the digital analytics process. Let’s shed a quick light on what these processes are and how they work:
1.Page Information: Which essentially means the URL and the title of the page the user is viewing.
2. Browser Information: All the information regarding the browser is tracked, such as:
- The name of the browser being used.
- View port or Viewing pane, which means the size of the browser window.
- Screen resolution.
- Whether or not the user has Java enabled.
- Which version of Flash is being used.
3. User Information: Google Analytics also tracks and determines the language as well as the location of the user. The former is determined by the language settings of the browser, while the latter is derived from the IP address of the user. Since the IP address is regarded as Personally Identifiable Information, therefore the IP itself is not available in Google Analytics as it violates its terms.
There are of course, some reservations regarding this process as Google Analytics simply gets a snapshot of the information at a particular time. So, for instance, if the user changes the size of the viewport mid-session, Google Analytics would only know about the change if the user then triggered an event or viewed another page. If the user changed the size of the viewport and then left the page without firing another hit, Google Analytics would remain utterly unaware.
Essentially, a document that maintains all the real world objectives and goals of your business in the form of metrics and dimensions is called a measurement plan. Google Analytics enables you to measure these metrics and dimensions on your website.
Now, the prerequisite to creating an effective measurement plan is defining viable goals. Fundamentally, goals are supposed to measure the effectiveness of your site when it comes to fulfilling your objectives. Therefore, it can be stated that a goal represents the completion of an activity (or a conversion) that contributes to the success of your business.Examples of goals include making a purchase (for an ecommerce site), completing a game level (for a mobile gaming app), or submitting a contact information form (for a marketing or lead generation site).
Defining goals is a fundamental component of any digital analytics measurement plan. Having properly configured goals allows Analytics to provide you with critical information, such as the number of conversions and the conversion rate for your site or app. Without this information, it’s almost impossible to evaluate the effectiveness of your online business and marketing campaigns. Makes sense?
We can now move onto the process of actually creating a measurement plan. It is actually a quite straightforward, four step process:
- Define your objectives and key performance indicators (KPIs)
The first step is to simply identify and outline the objectives of your business and feed them into any easy to use document like MS Excel. You then have to strategize on tactics that would enable you to achieve those goals and also define KPIs for each goal. For instance, if one goal of your company is maximum revenue, then your main strategies could be to increase sales, to increase average order value and to reduce returns, which in turn form the KPIs.
These goals and strategies are of course not definitive and require periodical revision to ensure that they remain effective and synchronous with the constantly progressing digital world.
2. Segment Data and Set Targets
After you have the basic framework in place, you need to move onto a more in depth analysis of your goals and targets. For example, you may want to split performance targets for mobiles and desktops, or you may want to set targets in accordance with geographics or demographics. This stage is where you come up with answers to such questions and thus segment the data before the implementation. You also need to translate your goals into measurable metrics at this stage, after which you can move on to setting targets for each of those metrics.
3. Create an Implementation Plan
Now that you are aware of exactly what you need to measure and track, it is time to begin the tracking process. You need to conduct an analysis regarding what sort of tracking you need to conduct and what steps you need to take for it. For example, some websites require an upgrade to Google Analytics latest code i.e., Universal Analytics. So you may be required to do this before implementing the measurement plan.
After you know what sort of tracking is required, you can now devise an implementation plan. This can be done easily by creating another tab in the same Excel document. You may use columns to highlight the scope along with what is already being executed and what needs to be done. After all the technical nitty gritty is taken care of, you can now implement the tracking process. It is important to keep in mind though that the digital landscape is constantly changing and developing, therefore another important step that you should consider is planning the ongoing refinement and maintenance of your measurement plan so that it remains accurate and upto date.
Now let’s route back to our initial concern: let’s say you are in the position of the aforementioned data analyst. You are sitting in front of your laptop with your palm on your face, wondering how to generate an effective report for a company that does not have any tracking system in place and thus skipped the processes of data collection and measurement planning; while simultaneously cursing at all your life decisions that lead you here.
Well calm down, there may be no data but there is still hope.
At times like these you can opt for conducting the analysis offline by scraping the missing data. However, this process only allows you to deal with the content attributes of the data. It does not mean you are free to access all the historic data and therefore is in no way a replacement for the proper tracking set up. Essentially, this procedure can help with one-off analysis projects as it gives you a useful historic baseline for you to draw conclusions from. There are various ways to go about this process, for instance you can make the use of a web crawler to fetch the missing data and match it against an export from Google Analytics, or you can also use the IMPORTXML function in Google Sheets to scrape the data using XPath expressions, or you may even come up with other approaches to take. Irrespective of which method you use to conduct this, the fact remains that the ability to scrape data points from websites and matching them up in Google Analytics is an excellent lifeline when it comes to historic data.
Still unclear about how to efficiently conduct a retroactive data analysis? Let us help.