As data tools become more impressive, we have to understand their limitations

We see this practice as obsolete and unacceptable, so we formulated our strategy with one objective: integration. For that reason, there's almost no place in our company that isn’t currently using data to inform some part of the strategies we develop for our clients. We have PhD data scientists contributing to creative ideation, and our strategists and creatives spend plenty of time looking at performance metrics. We know that analytics and insights must go hand-in-hand with nuance, art, creativity, and interpretation.

As data’s popularity has skyrocketed among marketers over the last few decades, it’s given us the ammunition we need to be more thoughtful, focused, and strategic in our work. Data will continue to be relevant to every part of marketing, from decisions on creative to targeting and spending. But it’s time we acknowledge that all data isn’t equally high-quality, or equally accurate. It will be up to data-savvy marketers to know when to rely on the data results and when to ignore them based on the type of decision they need to make.

Know where the data is coming from

Very few people (if any) within a marketing practice actually look at raw data. They're usually looking at some output of an analysis or a model. As a result, there are layers between the marketer and the data, and in those layers are all kinds of implicit assumptions. Those assumptions can render data relevant, or not.

Understanding the process that brought you the data points enables you to ask: ‘is this something I should use to make a decision?’ This type of questioning will become only more important. Far too often, spurious correlations in marketing are presented as if they are causal facts.

Give yourself permission to ignore the data results

Being comfortable enough with data insights to understand when they’re relevant (and when not) is a challenge for marketers and agencies.

Just because you’ve been given a set of data points doesn’t mean you have to follow them off a cliff. In fact, the best marketers know that some analyses are better left ignored. So it’s important for anyone in a marketing team to be more than just conversant in data.

For example, your data model may imply that it knows what will happen if you dramatically cut ad spend. But if you’ve never spent so little, can it really predict what will happen? The model may be lacking the necessary historical data to make an accurate prediction. In some cases, you may be better off trusting your intuition. Your model may be right two out of three times, but it’s critical to be discerning enough to recognize the one case where you simply shouldn’t trust it.

Always bring data into the creative process

Historically, in our industry, creatives and data scientists rarely collaborated; some creatives worried that analytics could stifle the art. But we’ve found that data-driven decisions allow us to unleash even more creativity.

When our Studios teams have as much information as possible at the beginning of their ideation process, it only makes their work better. Our rapid test-and-learn approach to experimentation can be freeing for creatives, because it gives them the space to be more innovative as they throw ideas into the mix and get feedback on what’s performing well and what’s not. It’s proof that art and science can work well together to breed originality.

Be picky about your measurement tools

Increasingly, we find ourselves educating clients on which measurement tool is good at answering which questions. We want them to see the value in having a variety of measurement tools at their disposal. Most of these tools won’t tell you that they aren’t designed to answer a particular question accurately. Instead, they’ll spit out the best answer they can, which could unknowingly lead you to make less-than-optimal conclusions upon which you base future efforts.

At Known, we encourage every client to go through a measurement audit so we can understand what first-party and third-party tools they’re using to measure the efficacy of their marketing. In most cases, clients aren’t using enough tools to get the most accurate insights.

There might be one measurement approach that's good at helping you figure out which creative will perform better than another, but it may not be able to tell you how much you should spend on your next campaign. Another tool might be better at telling you how much to spend and on what channels, but it could be all wrong at picking the most effective call-to-action.

It’s very rare that there's one measurement tool that you should trust for all your questions. A marketing team must know which tool provides the most accurate data analyses for each problem. This is the crux of where data analysis is headed: we have to be willing to admit its limitations. If we don’t, we run the risk of putting far too much weight on computer-generated insights without the right context. And that’s one of the riskiest things any business can do.