Why You’ll Lose with a Data-First Strategy…and What To Do Instead

Blog Post
July 06, 2015


A few weeks ago, Avinash Kaushik, digital marketing evangelist at Google, shared his perspective on how to “de-mystify” misconceptions about digital marketing and analytics on his popular blog, Occams Razor. While he made several excellent points, the myth-busters that resonated most with me were in relation to creating a data-first strategy and…Cookies! Cookies are all we need! I’m going to talk about the proper of use of data in driving effective marketing in this blog, and tune in next week to learn about how to go beyond cookies and use other technologies to create a single customer view.

Looking for data in a gold mine

As a company that specializes in data insights, it may seem odd that we agree that data-first strategies really aren’t the right approach. Let me explain: data on its own is not always useful, and too much data can be equally disadvantageous.

Simply looking for insights in large data sets is like panning for gold. It’s a lot of work for limited treasure. However, if we take a more targeted approach by doing some proper geology, planning and then mining for the gold, it will yield exponentially higher returns.

In other words, to get the most of your data for smarter decision making, data collection should always have a specific purpose, and the data collected should be fit for that purpose

Data will help you find more useful answers to your problems when you follow these steps:

1. Start with the key business issues that you are trying solve. These should have high impact.

2. Develop a specific question or hypothesis that will guide discovery. Some examples are:

  • Is there a correlation between frequencies of site visits and conversion?
  • Are the most loyal customers also the most profitable?
  • What are the indicators of churn?
  • What are the “best” upsell and cross sell opportunities?

3. Obtain the data that will be most useful to the analysis. Be cognizant of changes in the business that impact the data—including mergers and acquisitions, changes in product lines or go-to-market models. Always remember that behavioral data is just that and can be misleading if you are looking to draw absolutes or not taking context into account.

Data can lead you down the wrong path for the right reasons

A few years ago, when I worked as a product manager at another company, we were looking at click stream data trying to figure out why a certain section of the web site was experiencing a high volume of abandons. We spent a lot of time trying to figure out whether there were issues with the content, page design or navigation. Data was telling us the “what,” not the “why.” So, we decided to ask people via a survey tool why they were leaving the site area. The answer: “I got the information I was looking for.” In other words, they completed the task and we couldn’t see the forest for the trees. We finally got the “why” and realized that the page and content were doing their job.

The lesson learned was that it’s really all about people and intent, and sometimes it’s hard to measure interest and intent with data alone.  You often need to connect with the decision-making behavior on a human level. Which leads to my second point of discussion on why cookies alone aren’t the answer – tune in next week for that post.