The central ideas and much of the following are from Samantha Stone’s excellent book on marketing that works called, Unleash Possible. You could whitewash it and say that data is only misleading, but data by itself, can turn reality on its head. Data is not productively actionable without a human analyst. Here are four (of the many) ways that data lies:
1. It only tells us what, not why: Behavioral data doesn’t reveal the underlying reasons behind actions. To do this, you need qualitative research.
2. It loses sight of the individual: When everything we do groups people together, we lose sight of the some of our best prospects. For example, in sending out mass emails, we look at the open, click through, etc. rates. This assumes that everyone who didn’t open the email was not interested. The reality is that a lot of people who won’t respond to a mass email are waiting for a more personal touch—from either you or one of your competitors.
3. False confidence in the sales funnel: Sales and marketing people love to track prospect data along the mythical sales funnel. In reality, most prospects don’t follow this linear path. While data may show that people are dropping out at particular point; in reality they may have skipped ahead a couple of steps (they may very actively be considering your offering, but they aren’t on your radar anymore).
4. Flat Out Bad Data: Not to pick on Google Analytics, I do find it extremely valuable, but there are so many points where users cause it to fail. Here are a few that apply to B2Bs:
Code is missing from a webpage
Someone applies a new filter or an existing filter is deleted
Not taking holidays into account—you can’t compare week/week stats with Christmas week.
An extra weekend in a month—you can’t accurately compare month/month traffic if one of those months had 4 weekends and one had 5.
Active 3rd party consultants. While you may have excluded URLs of everyone working at your organization, third party consultants probably weren’t excluded. This means that if you have a web developer that is a consultant, the traffic will spike during periods they are working on your site.
Another factor Stone cites is the false confidence that any database is complete. It may be accurate, but it will almost never be complete. So before you can reach any conclusions with data, you have to thoroughly analyze the points where it can fail. This applies to low-level info and instances where the consequences are much greater, such as input for simulation and modeling.