Like maintenance work on your car, a refresh of your organization’s website is one of those tasks that seems to roll around every few years and entail far higher costs than anticipated. However, also like car maintenance, a website refresh is often critically necessary. Your organization’s branding or messaging may have changed, or maybe your site just looks like something that was built on GeoCities in 1995. Whatever the reason, you’ll want to ensure that you’re measuring the results of a website refresh in order to adequately justify the costs.

The most basic, and most powerful, question we can ask is “is our site better?” The problem of course is that that’s an incredibly vague question. What do we mean by “better?” Typically, we’re looking at KPIs like page views, number of sessions, bounce rate, time on page, exit rate, conversion rate, etc. The sort of numbers you can get from any basic web analytics tool like Google Analytics. I typically regard these numbers as “surface-level metrics;” they provide a good starting point but don’t say much about whether visitors are deriving value from your website.

Read More: Raising the Bar for Corporate Diversity and Inclusion

To illustrate, page views and the number of unique sessions help you understand site traffic. However, they don’t say anything about how “good” your site is, just how many eyeballs you’re getting and perhaps how well you’re doing in terms of SEO. Bounce rate, time on page and exit rate go a little deeper. These metrics can point to the quality of content on your site or, more directly, to whether the content on your website adequately meets the expectations of visitors. For example, if visitors to “bike.com” are expecting to find information about motorcycles but discover content focused on bicycles, they are likely to leave disappointed, and in a hurry. Unfortunately, depending on how engagement is defined in Google Analytics, metrics like bounce rate can result in both ​false positives and false negatives​, leading marketers to draw the wrong conclusions.

Ultimately, one of the best KPIs to consider is conversion rate: the percentage of visitors that complete the action we want them to. This could refer to signing up for a demo, downloading a trial, joining a mailing list, etc. To quote ​WordStream​, “a high conversion rate is indicative of successful marketing and web design: it means people want what you’re offering, and they’re easily able to get it!” More than anything else, conversion rate is the KPI that marketers look at when assessing website performance.

Read More: It’s Time to Give Marketers Control Over their Data

However, even a site’s conversion rate tells only part of the story. It doesn’t say anything about why

​ a visitor did, or did not, take a particular course of action. Yet answering “the why” is critical if we want to deliver more value to visitors and ultimately boost all other KPIs. For example, did a visitor not sign up for a demo of our software because it’s not relevant to them or because the content on our website doesn’t do a good job of speaking to their pain points? That’s why I recommend focusing on something a little unconventional by today’s numbers-focused approach to marketing analytics: examining and gathering qualitative data. One way to do this is through polls/surveys.

Read More: Five Tips to Perfectly Time Your Webinar and its Promotion

Despite the importance of other KPIs in determining how successful a website refresh has been, polls/surveys are unmatched when it comes to answering the question of “why.” It’s easy to deploy a survey using an inexpensive tool like Hotjar and collect visitor feedback both before and after a website refresh has gone live. While these certainly aren’t “scientific” polls, they do provide important nuggets of feedback that can prove useful when reviewing website performance.

In short, determining the success of a website refresh is more of an art than a science and requires a multifaceted approach. Although traditional metrics like page views and bounce rate are an important source of information, they should be complemented by a more “qualitative” approach in the form of surveys, which can add more color to visitor behavior. I would also advise readers to think of website optimization as a continuous process, rather than a sharp “before” and “after.” While there will certainly be a go live date for any website refresh, components like content and navigation can, and should, be continuously tweaked over time to incorporate visitor feedback.

Has your organization recently gone through a significant website overhaul? How did you measure success? Let me know in the comments below!

Read More: The Anatomy of Sales Decisions: Blending Data Science and Behavioral Science to Accelerate Sales Results