When is it time to invest in a website redesign?

For this decision, a targeted improvement keeps the site's overall structure and changes one part of it: a page, a form, sometimes a single shared template. A redesign reworks a large share of what the pages have in common, such as the menu, the page templates, the way content is organized and often the message the site is built around. It doesn't have to replace all of it.

That difference tells you which one a problem calls for. If the problem lives on one page, fix the page. If it lives in something many pages share, each page-level fix has to work around the same limit, and the larger project starts to make sense.

A change in the business is the clearest case for the larger project. Suppose a site was built to sell one product to small companies, and the company now sells three products to larger ones. The menu, the homepage and the page layouts were organized around the old offer. The team can rewrite individual pages, but visitors still arrive at a structure that describes the company as it used to be.

A site that's slow to change is another case. If every new page needs a developer, edits wait on developer time, and the site can fall behind what the business sells. Reworking the templates so the team can make routine changes itself is one reason to take on a larger project.

The last case is a set of task failures with one cause. Suppose testing shows that visitors struggle to find pricing and to tell which product fits them. If both problems trace back to how the site is organized, reorganizing it once makes more sense than patching each page.

Many redesign conversations start with a number that dropped, usually leads or demo requests. A drop tells you something changed. It doesn't tell you the design caused it, so it's worth checking the simpler explanations as well.

Start with traffic. If a campaign ended or a popular page lost search visits, fewer people are arriving, and the site can be converting them as well as it ever did. Then check the tracking. A rebuilt form or a change in the analytics setup can lower the count that gets recorded while the real number holds. These are useful first checks, and they don't have to finish before you study what visitors do on the site.

That study has to look for causes, because the obvious guess can be wrong. In an article on A/B testing and UX research, Jen Cardello of Nielsen Norman Group uses the example of a loan application that people abandon. A team could assume the application has too many screens and shorten it. In her example the applicants were missing interest-rate information, which is a different problem with a different fix. A redesign built on the wrong guess can carry the real problem into the new site.

CXL's guide to conversion research makes a related point. It treats what a reviewer notices on a page as an area to investigate and not as an established cause. We follow that kind of investigation through one funnel in whether a website redesign will fix low conversion rates.

Suppose a software company's site brought in 60 demo requests a month last spring from 3,000 visits, and it brings in 40 now. The marketing lead wants a redesign.

The team looks at traffic first. Suppose a paid campaign ended in June and visits fell to 2,000 a month. Forty requests from 2,000 visits is the same 2% the site converted in the spring. In this example, the fall in visits accounts arithmetically for the lower request count. An unchanged overall rate doesn't show that every page performs as it did, so the team can still look closer. The lower count alone is no evidence of a design problem.

Now suppose visits held at 3,000, so the rate fell from 2% to about 1.3%. The team confirms the form is still being counted correctly. Then it watches a few people who match its buyers try to choose a plan. The company added a second product over the winter, and the menu still lists one. Most of the testers land on the first product's pages and can't tell whether the second is meant for them. Sales reports hearing the same confusion on calls.

That evidence points at the site's structure, which was organized around a single product. It's a diagnosis and not yet proof. The testers and the sales calls give the team a likely explanation. Measuring a change later can show whether the numbers moved, and even that won't settle whether the explanation was the right one.

So the next question is how small a change can test it. If the menu and templates can take a second product without being rebuilt, adding it properly is a targeted improvement, and the redesign can wait for the result. If they can't, the company is in the first situation described above: the business has changed and the site can't follow it. That is the case for funding the redesign.

Lean Labs, an agency that builds HubSpot websites, answers this question in its own article on redesign timing with a sequence. Evaluate how the site performs now, prioritize the improvements, set measurable goals against a baseline, and work in sprints.

The agency's article on the three stages of growth-driven design lists a first site build, strategy included, at $30,000 to $70,000 or more over 9 to 13 weeks. The method calls that first build a launchpad. The article also lists a Design Blueprint, a four-week project that sets visual direction, at $6,000 to $12,000. If you're quoted for both, ask whether the Blueprint fee is included in the build price.

Those are one agency's prices and not a market range, but they show what's at stake. Before committing to a build of that size, ask what the investigation would cover and cost, and compare that with the build quote. We go through the budget in how much a HubSpot website redesign costs for a B2B SaaS company and the preparation in how to plan a HubSpot website redesign.

Before work starts, record how the current site performs. In an article on UX benchmarking, Kate Moran of Nielsen Norman Group describes comparing a design's measures against a reference, such as an earlier version of the same site. Her examples include whether people complete a task, how long it takes and how easy they rate it. For the software company above, the record would hold the share of testers who reach the right product page, demo requests per visit and the number of those requests sales accepts.

Measure the same things the same way after launch. A better number afterward fits the idea that the redesign worked. It doesn't prove it, because traffic, prices and the season can all move in the same months. Keep the traffic mix and the tracking setup in the same record, since those are two of the first things to check the next time a number moves.