Back to top

Web analytics: a cautionary tale

April 13, 2017

Digital Team icon

Published by:

Digital Team

A story about how the numbers we look at lie to us all the time.

Like anyone who manages a website, our team keeps an eye on Google Analytics to track our web traffic. What’s popular? What content are people looking for? What content are they finding? The point, of course, is to use metrics to drive improvements to the site — what we display, where we display it, etc.

We noticed years ago that our food truck schedule is often one of our top three most visited pages, especially during the summer months. We’d assumed this was because of the recent popularity of food trucks and the fact that is the central place to find that information. It was only last week, after our old food truck schedule went on the fritz, that we peeled back that web traffic and learned the truth.

Google Analytics Results

We looked at open source options to replace our food truck schedule. We were surprised to find several options exist. We were even more surprised to find there are a handful of apps focused on Boston’s food trucks specifically… several of them getting their data by using bots to regularly scrape

On the one hand, that means our traffic numbers are artificially high for that page, boosted by bots scraping the site. On the other hand, it means our food truck information is being disseminated to an audience outside our Google analytics, artificially depressing the traffic number. In any case, it’s clear we haven’t been looking at an accurate traffic number for that page and it’s not likely we’ll have one in the future.

It’s situations like this one that keep us skeptical of the metrics we focus on. There’s a lot to unpack behind each number that changes the solutions we build. If our food truck schedule is truly one of our most visited pages, then we should invest time to make sure it’s built well and prioritize it above other projects. However, if users find this info in other places then we should instead maintain a lightweight food truck schedule on our site and prioritize building an API for the data that others can use to update their apps (rather than scraping our site for it).

So that’s what we’ve done. Check out our new food truck schedule, a minimum viable version that we’ll use until we can invest time in making it more user friendly. In the meantime, we’re focused on other projects and building an API so food truck apps can more easily consume our data.