The Evolution of Digital Marketing BI
Neil Patel wrote in his Forbes article that “marketing and analytics go together like peanut butter and jelly.” I tend to agree. In the past 10 years, the Digital Marketing industry has been one of the earliest adopters of new BI (Business Intelligence, similar to Data Analytics) technologies. The combination of (1) infinite amounts of structured data, (2) plenty of experienced and tech-oriented personal and (3) A LOT of money, made our industry the ideal playground for innovative data technologies.
They say Rome wasn’t built in a day – well, neither were Digital Marketing BI technologies. I believe the short history of Digital Marketing BI can be divided into 4 major generations. In this article, I will try to explain what happened in each of them in a few simple words:
First Generation – Introducing Data
In the early days of Digital Marketing BI, we didn’t have fancy BI tools, graphs and dashboards. All we had was good old MS Excel spreadsheets and some sophisticated pivot tables. A substantial part of account managers’ workday was devoted to extracting data and creating massive Excel reports. The data itself and the data sources (from Google Adwords and Facebook Ads all the way to mobile app analytics tools such as Adjust, Appsflyer and Tune) haven’t changed much since then, but the way in which we extract and use the data has changed dramatically.
These processes were time-consuming and sensitive to human error. Does “OMG, the ROI dropped by 10%! Oh sorry, I was pulling the wrong formula…” sound familiar?
Second Generation – Data Visualization
The Second Generation of Digital Marketing BI was great news for marketers. Data extraction and processing was done automatically in the background by designated tools. We at yellowHEAD use a tool called Rivery to automatically extract data from the various data sources, process the data to suit our needs and organize it in our data warehouse.
For the first time, Digital Marketing agencies started using BI tools such as Tableau, Qlick and Sisense. By using these tools, BI analysts\developers could generate smart and simple dashboards that helped recognize trends and get actionable insights. Instead of looking at dull pivot tables, account managers started looking at elegant visual dashboards and graphs.
Since account managers no longer needed to clean and organize the data, they could focus on what they do best – optimizing campaigns and generating profits. But that wasn’t enough…
Third Generation – Smart Alerting
As time went by, the Digital Marketing industry grew fast – more users, more profits and a lot more DATA. To stay competitive in this market, account managers had to use their time more efficiently. When you have 20 campaigns, you can check each of them every day; when you are managing 2,000 campaigns – you must recognize the most urgent campaigns first. This is where Smart Alerting comes in. Passively going through various reports and looking for insights is the practice of the past. By using Smart Alerting techniques, account managers can define what they consider interesting or important (in terms of KPIs and performance metrics) and get notified only when their attention is needed.
Fourth Generation – Predictive Marketing Analytics
To keep it simple, predictive analytics is the art\science of looking at data and estimating what will happen in the future. Imagine you can open a new campaign today and by tomorrow you’ll know what the results of this campaign will be in one week’s time. Today, data scientists are integrating smart Machine Learning algorithms with web-scale amounts of data to predict future results of current campaigns. In simple words – take a good data scientist, add tons of data, stir well and voilà – you can predict the future.
What’s Next? – Prescriptive Marketing Analytics
<Disclaimer – This part is for advanced readers>
The most highly discussed business intelligence trends among BI professionals is called Prescriptive Analytics. Prescriptive Analytics takes Predictive Marketing one step further – it provides us with recommendations of the best actions that should be taken in order to achieve a specific goal.
The nice part is that the algorithm makes sure that the recommended actions are indeed the best possible actions. To do so, the model needs to predict the future or to use – you got it – Fourth Generation Predictive Analytics models. Brilliant.