Precision, Accuracy, and Marketing Measurement

If you are talking to a marketing agency about new opportunities, they better ask how you’ll be measuring results. If they don’t, you should just turn and run, hang up the phone, click the red button on Skype, or whatever. Any honest marketer worth their salt knows that having a measurement plan is essential for success. Otherwise you are blind to the results of your marketing initiatives, and lacking the insight to keep your marketer accountable. Thankfully, there are a lot of great tools for marketing analytics, and it doesn’t cost much to get started. I see a lot of companies use Google Analytics because it is free, fairly easy to use, and scalable. There are paid alternatives too that have impressive capabilities when it comes to fancy stuff like tracking conversions across multiple devices. Regardless of the degree of your marketing sophistication, you need some sort of analytics tool and a measurement plan. A marketer who fails to ask about your measurement plan isn’t taking your best interests to heart; they’re most likely just looking for a short-term win.

No matter which marketing analytics tool you employ, the dashboard will feature pretty graphs and compelling statistics. You’ll gain access to a wide array of very precise numbers. You can look at time-on-site down to the nearest second and bounce rates to the tenth of a decimal. You can assign a dollar value to different types of conversions, and calculate an ROI for the whole campaign. It’s all very precise and compelling.

If you are a numbers guy like me, it can be addicting delving into the data. You may find yourself pondering questions like: “how the hell did we get twenty clicks from Mozambique?”, “why is the bounce rate for my bio page so bad in particular among members of the opposite sex?”, or “why are visitors only spending 34 seconds on average reading my latest blog entry? Does nobody appreciate good writing anymore?”

With so many ways to slice-and-dice analytics data, it’s tempting to take the accuracy of the numbers for granted. But it’s important to remember there’s a big difference between precision and accuracy. Analytics data tends to be very precise, but rarely accurate. Now you may be thinking, “come on Devin, obviously there’s a difference, but you’re just splitting hairs here.” To which I would like to point out that one of the main reasons the financial crisis in 2007 occurred was due to bankers relying upon very precise risk management measures, which ended up being inaccurate. Former Chairman of the Federal Reserve Alan Greenspan wrote in 2008: “The most credible explanation of why risk management based on state-of-the-art statistical models can perform so poorly is that the underlying data used to estimate a model’s structure are drawn generally from both periods of euphoria and periods of fear, that is, from regimes with importantly different dynamics.” Indeed, even the smartest quants in the world can be fooled by precision feigning as accuracy if the context of the data is ignored.

Marketing analytics are inaccurate by nature. There are fundamental and arguably unsolvable reasons for this. You can track visitors using cookies; but cookies get cleared out and have a limited shelf life. You can track people by IP address, but IP addresses change. You can track visitors who log into their accounts, but they aren’t always logged in. You can try your best to filter out spam traffic that skews results, but you’re never going to get all of it. Likewise the impact of your employees’ traffic can be minimized but never eliminated. You can come up with assumptions about how much different conversions are worth to you, but they’ll be guesstimates at best. You can see “average time on site”, but even that’s an imperfect measure because if someone bounces they’ll be recorded as a 0 second session regardless of how long they actually stayed on the page. I could go on.

Virtually every marketing metric is inaccurate. But that is not an argument against having a measurement plan and using an analytics tool. You absolutely still need to have those things. But it’s a mistake to look at the analytics data as “truth”. Marketing analytics are valuable because they provide insights into trends, not accurate measurements. If your average time on site goes up from 2:00 to 2:30, the correct way to interpret that is to say “the time on site is up about 25%”, not “people are spending 30 seconds more on the site”. That’s a simple but important distinction. Don’t make the same mistake the bankers did in 2007 to spark the global recession, make sure you interpret marketing analytics in the proper context and without the expectation of accuracy.

Contact eBridge today if you need help developing a measurement plan, selecting or optimizing an analytics tool, or interpreting your analytics data.