How Do You Know If Your Website Redesign Worked?

Subheader

Decide what success looks like before launch, then measure it

Featured Image

Woman reviewing a printed page with various analytics and graphs

Sections

Most redesign RFPs can’t answer that question, because they never ask it. RFPs read like a checklist: build it in Drupal, pass WCAG, add an events content type and an alert banner, connect the mailing list. Those are legitimate requirements, and they’re easy to check off at launch. But none of them say what the new site should change for the people using it.

When we ask clients directly, universities usually name enrollment, whether that means growing it or holding it steady. Government agencies often name accessibility compliance. Both are good starting points. Neither says much about how the redesign will make the site better for someone trying to renew a license or apply for a program.

So we push for more specific goals: clearer paths to key sections, faster page loads, more forms completed, more visits to the pages that matter most. During discovery, we talk through what constituents need most and whether there’s a way to measure it. But an agency can’t set these goals for you. You know better than we do what a good year looks like for your organization, and you’re the one who’ll explain the results to leadership.

Once you know what you want to change, you need a record of where it stands today. Without a before number, there’s nothing to compare against after launch.

Over the last couple of years, we’ve started projects with a Project Success document that captures how the client will define and measure success. Then we ask for access to the current site’s analytics and run the baseline traffic and performance reports ourselves.

If you’re writing an RFP, put those success criteria in it. Bidders will scope their proposals around them, and you’ll have a clearer way toevaluate what comes back.

With a baseline in place, a handful of numbers will tell you most of what you need to know. We recommend starting with these:

Sessions and engagement rate
Search impressions and rankings from Google Search Console
Performance scores on desktop and mobile
Accessibility test results
One outcome specific to the organization, such as a form submission or a file download

You may have noticed that site visits are missing from that list. Spam and bot traffic clutter a lot of site analytics, and for many of our clients, traffic from outside the US is noise anyway.

The harder part is reading these numbers once they come in. The city site from the opening is a good example. Most people visit a government website to get something done, and then they leave. If they’re finding what they need faster, a shorter visit is a good sign. That’s more instinct than proof, but user testing can firm it up by showing how quickly someone completes a task, and your analytics can show the paths people take through the site.

Sometimes those paths show the opposite problem. In a recent review for a government client, we looked at a classic A–Z index page. Visitors would click one letter and then several others in the same session. They weren’t finding the term they needed, so they guessed. About 75% of that page’s traffic centered on just a few terms, so we recommended featuring those terms directly instead of hiding them under arbitrary letters.

Paths like these only show up if you’re tracking what people do on the site. On every new site, we set up events for forms submitted, videos played, documents downloaded and internal site search terms. We’ve also started measuring traffic from AI tools, which Google Analytics doesn’t separate out by default.

Beyond that, events depend on what the site is for. Take our work with LawHelpMN. We track engagement with training videos and polls, which categories of legal help people use, and how far they get through the online Law Help tool, which walks users through personal questions to specific legal recommendations.

Some of the clearest evidence that a redesign worked shows up somewhere other than your dashboard.

For higher ed, that usually means applications. After we redesigned the website for Minnesota State Community and Technical College (M State), applications to the school rose 13.9%. A lot happens between a visit and a submitted application, but you can see how often people engage with the application path on the site.

For government, one outcome we try to measure is fewer customer-support calls. If people can find answers on the site, the front desk should field fewer questions. If you can count service calls before and after launch, you’ll have a number leadership understands immediately.

Every one of these numbers takes time to settle. Give it at least 30 days before you treat traffic and event data as a new baseline, and six to 12 months before you look for lasting patterns. Compare against your own baseline, too. Every site and sector is different, and industry averages say very little about yours.

None of this works unless if someone is watching. In our experience, very few organizations check their analytics regularly after launch, and most government teams don’t have anyone in that role.

But someone has to own the numbers after launch, whether that’s your internal team or a support partner. We cover what that ongoing work looks like in Keeping Your Website Fit.

That city we mentioned in the beginning of this article saw numbers that looked confusing without that context. A shorter visit means something once you know what the site was supposed to change. Before your next redesign starts, write down the few numbers you expect it to move, and pull this month’s version of each one.

If you’re still deciding whether your current site needs a rebuild at all, our free CMS Fit Assessment takes about five minutes.

About the Author

About the author