Not seeing a Scroll to Top Button? Go to our FAQ page for more info.

The Shelf Life of Data

I know in these days where data is king, what I’m about to say may seem like heresy, but hear me out. All too often I meet with clients who spend way too much working with data that’s just not going to help them accomplish their tasks at hand.

The mantra, “don’t throw data away because we don’t know what’s important,” doesn’t apply to all situations. To be sure, it’s the right approach when you’re doing a lot of profiling or predictive exercises, like predicting traffic patterns over the next 24 months. In these cases I agree -- use  as much data as you have, and let the data speak.

What I’m talking about is a bit different. Too often I see people struggle to make immediate decisions because they don’t have the right data they need in front of them. As a result, an opportunity is lost.

So rather than display every bit of available data when running weekly reports, I suggest we accept that data has a shelf life, and once it expires, it won’t help you any more. Of course that begs the question: How do you know what the shelf life is?

Here are some rules of thumb that I’ve seen work with our media clients:

Start with the use cases

Let the use case drive the level of information you need for your analytics. Some use cases are bread and butter and require immediate and concise data, such as verifying campaign changes  or launches or checking on time-sensitive campaigns. Others, like detecting market fluctuations, or addressing under-delivery require a deeper level. Be honest about what you really need to address the use case at hand.

Analytics is about speed as much as it is accuracy

Data may provide the insights into why a campaign isn’t performing as expected, but you need to get it in time to make changes. The right answer too late is just as useless as the wrong answer on time.

Access to just the right level of data, intra-day or at least daily, provides an opportunity to salvage a campaign (and potentially a client relationship) so you can drive revenue. If you wait to get that data when you run your other, more extensive reports, you squander that opportunity.

The value of historical information diminishes with time

For many use cases, the questions you need to answer are: What’s happened up to now? Where are we at? What do I need to do?  Understanding how a campaign performed in April isn’t particularly important now that it’s almost July. If you need to create a post-campaign analysis for a client, then obviously historical data is critical, but if you need optimize a current campaign, it’s far less useful.

Know where to see the field and where to see the details

Your field is driven by your role in the company. If you’re a frontline practitioner, a 10-foot view is appropriate, but for a CRO a 50K-foot one is.

For instance, ad ops people are far more concerned about the levers they have available for optimizing the campaigns or areas of the site for which they’re available. What data will provide you that insight? CROs don’t really care about those details; they’re more interested in knowing things like how revenue is flowing on a global scale and whether it aligns with forecasts.

The trick is to bring these analyses into the appropriate level of focus and resolution for their purpose.

Avoid the temptation of the "drill down"

Focus your analytics on what you need to know, not what might be interesting. As a data guy, I know how seductive data can be, and I too love clicking on that data plus sign to get more and more data points. But that is, at best, an exploratory exercise, not a diagnostic or strategic one. On top of that, datasets like that require effort to set-up and interact with on an ongoing basis which may hinder the core business goal. Worse: this approach isn’t particularly purpose driven, which leads to poor ROI. It’s not that drills are bad, but focus on what data relationships matter and avoid a kitchen-sink approach.

Every report should have a clear business purpose, and identify the things you have control over (e.g. campaigns, content, audience) to drive change. To do this, analysts would be wise to borrow from their marketing communications colleagues, who begin each project with a creative brief outlining the exact purpose, target and objectives, of every project.

Pick the right frequency of update for business reasons

Obviously this is intuitive, but too often people forget that they can choose how often to run their reports and do their analysis. If you need data every 30 minutes, design around that. As I said earlier, the right information delivered too late is useless.

Selecting the right data to focus on requires discipline and some upfront thinking about the shelf life of data. But the results of those efforts are enormous, because you’ll be in a position to make better decisions faster.