At the upcoming NRF Big Show, retailers will no doubt hear about ways to improve online sales by focusing on a core metric, conversion rate, that has outlived its usefulness as a key performance indicator. As an industry, we badly need to change the way we measure the performance of digital stores, and look for more modern techniques to explore optimization and growth. We’ve heavily relied upon a handful of metrics to analyze, understand and communicate how the business is doing. As creatures of habit, we continue to rely on a handful of metrics, some more than a decade old, that are swiftly losing their relevance and don’t mesh with our multi-device omni-shopper reality.

At the top of the list of key performance indicators is conversion rate. To be clear, in this case, I define conversion rate as the ratio of number of orders placed on the site in a given time period over the number of sessions (or raw traffic numbers) seen on the website over the same time period.

Why? Well, for starters, we made this case earlier about how conversion rate metrics are losing relevance. To add to that, eMarketer published a summary of a recent research that bot traffic makes up for a significant portion of overall web traffic, particularly with highly trafficked sites.

Bots have been known to corrupt traffic metrics, and in the past several months, we’ve seen an overall increase in the number of sessions that can be qualified as bots (non-human). As this research postulates, this new trend could make up to 50% of your site traffic, especially if your site has more than 100,000 daily visits.

Some bots are necessary – like the ones dispatched by Google to crawl your site for indexing purposes – and are purpose-built for performing “good” functions. But there are many others that may be uninitiated by the site owner, or may have malicious intent.

“Good” bots usually have a predictable signature that can be used to detect them and in some cases, allow your analytics tool to ignore them in your reporting dashboard. This is typically accomplished using the “user agent” field in the session signature. “Bad” bots, however, are not easy to detect and don’t necessarily announce themselves as such, making them notoriously difficult (not to mention resource intensive) to deal with.

The bottom line is, trying to filter out bots from corrupting your conversion rate KPI might be a futile exercise.

This being the season for resolutions, I am hoping that we as an industry rethink the way we analyze, understand and communicate business performance in this new year. By year’s end, I hope we can claim that 2016 has been the year we got into KPI rehab and got rid of our addiction to conversion rate.