When it comes to measuring the online success of a business, the maturity level of analytical reporting and decision making usually happens in a phased approach. More often than not the first step in this journey begins with the evaluation of more popular web analytics tools such as the free version of Google Analytics or, if the budget allows, a more feature rich tool like Adobe SiteCatalyst. This evaluation process includes defining what data needs to be collected, the expected reporting, the implementation of the tool and the necessary decisions required. Beyond these preconditions, the eventual analysis of this data can be an exhausting but ultimately rewarding process if it all goes smoothly, which is rarely the case.
One of the more common reporting requirements is that of campaign reporting from external sources such as email, social media and paid/organic search traffic. This is usually the first point where analysis of the visitor behaviour steps outside of the confines of the website, and traditionally uses specifically defined query string parameters to identify the sources and identifiers of the campaigns driving this traffic. With these additional dimensions the reporting is no longer isolated to only what happens on the website but can be segmented into what happens with visitors who come from various acquisition marketing efforts. Vanity URLs can also be used to measure visitors who came from a print, television or radio/podcast ad.
Depending on the nature of the business, this concept can be expanded to various other data channels. One of the more advanced features available in web analytics tools is the ability to import data from external sources and have that mapped to existing data points within the your analytics reporting data. For example, SiteCatalyst has a feature called Classifications whereby users can upload additional data to values which already exist in the reports. For instance product SKU might already be captured in your reporting, but for more meaningful reporting you may want to know more about product SKUs, such as the product name, category or any other metadata, such as color, size or design, for example. Simply map this metadata to the SKUs in an Excel document, upload it into SiteCatalyst, and the additional data will be available in all reports related to the SKU data which was originally captured.
The above example illustrates how you can provide additional context to existing data, but what if you have multiple data collection mechanisms apart from your website which will allow for analytical analysis. A good example would be the airline industry. There are various customer data touch points associated with the purchase cycle of a flight ticket. The potential customer could buy a ticket from:
– The airline’s website
– Over the phone via a call centre
– A walk-in flight centre
– Via a 3rd party travel agent
Once the flight has been purchased the customer may then have various options such as modifying, cancelling or checking-in for that flight, all of which could occur via any of the above touch points. As an example, even though a customer may have booked their flight online they may cancel that flight over the phone or via their travel agent, etc.
This could lead to inconsistent data when it comes to reporting on how well ticket sales are doing, both holistically and by individual sales channel. Even though the analytics software is recording the ticket sales on the website, there will also be an ecommerce engine which records the value of the sales in its own database and most likely this database also has data fed into it from the other sales channels such as the call centres. So if a purchase occurs online and a cancellation for that purchase occurs offline the reports from this system will be out of synch with what the web analytics tool is reporting and that may raise some questions about the validity of what is being captured on the website.
This kind of discrepancy is not the fault of the web analytics tool though, as it can only capture data from where the analytics code is set. Thus, if a purchase occurs online, the tool in its isolation knows nothing about whether that transaction has been cancelled or not. It is however possible to aggregate the online and offline data if the relevant intersection points are available. In our example we could use the ticket purchase reference number to upload offline data from the call centre or travel agent CRM systems into the web analytics tool to reflect ticket refunds or cancellations in the ecommerce reports, making them more trustworthy and hence confidently actionable.
This principle can be expanded further to whichever data points are available, so for our example: flight check-ins can happen online, from a kiosk at the airport or at the airline’s service desk. If the check-in happens online then it’s easy enough to tie it to the ticket purchase. If check-in occurs offline, then it would be a candidate for data upload. Even if a customer has checked in, it doesn’t mean that they will actually board the flight. Unforeseen circumstances may mean that they miss their flight and the boarding pass scanner data could be another candidate for including in analysis of how customers progress from searching for a flight, booking a ticket, and then boarding the actual flight. In our previous examples we used the flight booking reference as our data intersection point but if we wanted to expand the scope of our data collection channels we could look at something more generic such as the unique identifier for the customer in the airline’s CRM database, which could tie data between the website, offline channels, kiosks etc. and even activity related to the customer’s loyalty programme (buying accommodation, rental cars, etc. with air miles).
Based on what has been shown above it may start becoming clear that in order to take advantage of multi-channel analytics, the points of data intersection will need to be predefined for each channel. In our example we originally used the booking reference number but looking at the available data touch points at our disposal, the unique customer identifier would offer more expansive reporting and integration options. Similarly, depending on the industry, loyalty numbers or any other value that could uniquely identify a customer can be used to build a bigger picture of that person’s lifecycle within your business.
So far we’ve only been speaking within the context of a web analytics tool, which means that a process of data collection, transformation and upload needs to occur before it’s available within the reports. If you’re dealing with very large data sets, this could be a daunting process, as data formats and upload issues could mean a vast amount of time and effort spent in trying to ensure that the data is not only uploaded completely, but also correctly. When this becomes a concern, it may be time to start looking at tools which facilitate the aggregation of this “big data” such as IBM’s Hadoop or Adobe’s Data Workbench. Hopefully this article has given you some ideas to think about–how to tie data together, if even on a smaller scale, so that you can reap the benefits of multi-channel analysis.