As a practitioner of Web Analytics and Optimization, I’ve spent a fair amount of the last seven years of my career focused on tracking and improving Conversion Rates.This is a noble pursuit for any business, but intense focus on the conversion rate metric can have negative implications – people in your organization (or your clients) may obsess about changes in conversion rate, pester you about them, and even blame you for them!
To avoid this risk and/or annoyance, it helps to have simple, educational “sound bites” for stakeholders explaining why they are maybe/possibly freaking out for no good reason. This is NOT to say that you shouldn’t diligently investigate what you believe to be causes for concern to your business or client, however. Strike a balance, as in all things.
Let’s say someone in your company comes to you with some data about a change in conversion rate. That is a good thing (coming to you with data), right? So, first off, don’t brush them off. Encourage data-driven behavior, even when it might be off-base!
They’re concerned because in a week-over-week report, conversion rate has dropped by 25%. That’s usually a bad thing, so it’s worth some respectful, diligent investigation. Assuming you’ve looked at various data points, and are of the opinion that it’s not a cause for immediate concern, here’s how you might frame the conversation as it continues.
Hi, [Stakeholder], thanks for bringing this to my attention. I’ve dug into the data, looking at YoY behavior, key segments, etc. and I recommend we take a “wait and see” approach. I’m attaching a trended report, so you can see beyond the week-over-week view.
The Stakeholder will almost inevitably ask some follow-up questions about why conversion rate is down. While you will likely have some data points and explanation of your own, here are 9 of my favorite reminders of why your conversion rate may fluctuate from time to time, instead of the constant “up and to the right” trend that we all strive for:
- Campaign changes: If any marketing campaigns have started or stopped, it can have a swift and decisive impact on conversion rate. Remember that conversion rate is an average; a starting-point metric. Even seemingly minor campaign changes like advertising on new networks, testing new creative, or phrasing an offer differently can cause a KPI fluctuation.
- Traffic quality changes: Changes in the nature, or quality, of your site’s traffic can have a noticeable impact on your KPIs. You could consider this a subset of “Campaign Changes” above, but it’s a bit more abstract when you’re talking about multiple changes to paid search bidding tactics, Search Engine Optimization work, or content marketing. If you see this happening, it may be due to “click bait” tactics, so be diligent about the trends.
- Experiments: One of my favorites! If you’re running tests on your website (you should be), these can obviously have an impact on conversion rate. Since you’re splitting traffic among different versions of your User Experience, some designs may lift conversion rate, and some may depress it. Either way, make sure your team is in sync about what is being experimented on, when it launched, etc. It’s good to log these changes over time so you can look at data trends in better context.
- Design changes: One of my least favorites when not subject to testing! Often times, websites roll out design changes at the whim of some executive, or to simply “update” the look and feel of the site/brand. Often, these changes are made without the benefit of controlled experimentation. When this is the case, there may be unintentional impacts on conversion rate. Sorry.
- Tracking/measurement changes: Whether it’s tracking tags being stripped off of pages, or changes to how web analytics filters and profiles are set up, or a new definition of how a metric is calculated, these issues can impact the reporting of conversion rate, if not the performance of the site. When you see “overnight” fluctuations in conversion rate, this type of issue may be the culprit!
- Seasonality: This could be traffic patterns and purchase behaviors throughout the week (e.g. more sales on Sunday evenings). Or, it could be at the more macro level of the year-long calendar (think Holiday shopping or tax preparation). Either way, these trends are sometimes mistaken as true changes in business performance. In these cases, web analytics date ranges can be your friend, e.g. week-over-week comparison or year-over-year.
- World/geopolitical events: Our sites and businesses don’t operate in vacuums. World news, geopolitical events, or even changes in pop culture media can influence how people convert or don’t. This may be more or less important depending on your business, but don’t overlook it when investigating your data.
- Random chance: Despite how trend lines appear, your customers aren’t faceless blobs of flesh that behave in predictable ways over time. Quite the contrary – they are as unique as snowflakes or fingerprints (or QR Codes). They behave in “predictably irrational” ways, to quote a famous book title. The averages that our data shows us on dashboards hide a huge amount of random, mysterious behavior that we’ll never be able to understand. As such, your conversion rate might just bounce around a bit for no good reason!
- Low traffic volume = noise: If you’re Amazon.com, fluctuations in conversion rate are going to be averaged out by the massive scale of your online business. If you have less traffic than Amazon, your data will be “noisier,” and the ups and downs of your conversion rate will be more dramatic, e.g. a 20% drop instead of a 0.20% drop. If your daily volume of conversions is on the low end, you’ll need to set expectations that deviations from the KPI are going to be commonplace. If I buy from you tomorrow instead of today, I might cause that scary dip in your main KPI 😉
In conclusion, paying attention to Conversion Rate is great. Freaking out about its expected fluctuations is not cool. Be diligent in your data analysis, and you’ll learn what causes those blips and bumps in the spark line. If people are giving you grief, try a few of the 9 reminders above to frame the conversation in a sane, data-driven way.
Did I miss any other good causes of KPI fluctuations?