Time piece measuring hours, days, and astrological movements

When Is A Metric Worth Your Attention & How to Improve Data Interpretation

Measuring is only half the battle. You also need judgement to determine if, when & how your metrics matter. Here are some things to keep in mind.



November 9, 2019


I’ll spare you the Drucker quote, and just say that marketing metrics are important.

But I also want to caution you to keep 3 things in mind:

  1. Not everything is (easily) translatable to business value, but that doesn’t mean you shouldn’t do it.
  2. Metrics need context (e.g. YOY, cohorts), especially if you want to base decisions on them.
  3. Metrics aren’t created equal, and their relationships to each other need to be kept in mind.

Not everything is (easily) translatable to business value

Some metrics are fuzzy or hard to clearly connect back to sales.

Brand metrics are a great example of this.

While we measure brand sentiment, recognition, mentions, etc., the direct business value may seem unclear or only apparent when you’re company is at the ends of a brand awareness spectrum — either your brand is weak and means little to buyers, or its strong enough that people buy because of it. Everything in between isn’t obviously moving (or stopping) the needle.

But are you NOT going to try building your brand? Of course not. While you can’t necessarily see the impact yet, you know that brands matter & that brand building can pay off.

So what do you do? While you probably have to focus on performance-based marketing, you still make sure that the Brand is well-represented because you know that it will matter even if you can’t put a dollar value on it yet.

Metrics need context

For example, consider your metrics after a site redesign.

Whether you see initial improvements in performance or not, you can’t call something a success or failure without refining/filtering to make accurate comparisons.

You have to:

  • Compare to prior years
  • Identify outliers or random events that might have hurt past performance (remember the last AWS outages?)
  • Filter out campaigns or channels that didn’t exist before, or that have drastically changed
  • Be sure that improving one metric didn’t hurt another one.

If we don’t do this work, then we risk repeatedly making bad decisions. Why repeatedly? Its “Garbage in, Garbage out“: you start from the wrong perspective and incomplete information (Garbage In), and you make decisions and take actions based on that bad info (Garbage Out).

Many new marketers get this wrong because they want to show impact & will prematurely mark something as a win. The pain comes when they look at things 6 months later and see that key metrics are actually trending the wrong way.

Avoid issues around this by having a measurement framework that includes recurring historical analysis and normalized values.

Metrics aren’t created equal

Think about engagement metrics. Pageviews, time on site, page scroll depth.

We like to look at them to assess how well people like our content, but these metrics are usually only important as they relate to conversion metrics (yes, even for publishers).

If you’re doing lead gen or ecom, then “weak” numbers on these metrics might even be a good thing:

  • A good landing page might have really short visit durations, but great conversion rates
  • Declining pageviews or low pages/visit might mean your site is efficiently driving people to key content.
  • Minimal scrolling can mean the right piece of info for the audience is easily spotted on the page.

I’m not saying that engagement or dwell-time metrics are never important, but deciding to fix one of them without thinking about the conversion metrics is a surefire way to screw up your site.

Rather than focusing on these metrics. Start with conversions and work back:

  • What’s the difference in average pages/visit between converters and non-converters?
  • What percentage of visitors are seeing your first call to action?
  • [By campaign and/or channel] what’s the bounce rate for new visitors vs returning?
  • Does bounce rate decrease when I change the headline or hero image?

The key idea is to put these in the context of the business and then decide if they need work.

Caution: You can go overboard on this and end up with “analysis paralysis” because you’re trying to factor in too many things. Make sure that you put some boundaries around what to include. Conversions are usually a safe place to start, but even there you will want to drilldown (e.g. time of year, content type, product category) to make sure you’re not over-analyzing.

Do Your Metrics Matter?

We should always strive to look at the metrics that drive our decisions & honestly speak to their value. If you can’t, then there is an underlying problem in either strategy or process that you need to address.

That said, don’t ever let ambiguity in your metrics keep you from experimenting. Instead, try to keep in mind why you’re looking at a particular number and what a “good” value is for your channel, test, program or campaign.

You may also like…

Goodbye Prince.

Goodbye Prince.

Remembering the super talented artist that was almost a music genre onto himself.