How Often Companies Redesign Their Websites: 2026 Data by Company Size

Mid-sized software scale-ups replaced their homepage design about once every 3.2 years between 2016 and 2026, large public software companies about once every 5.6 years, and small bootstrapped teams about once every 8.7 years. Those figures come from our own count of 30 company homepages in the Internet Archive's Wayback Machine, one capture every six months for ten years. Company size changes the answer more than any rule of thumb does.

This page sets that count next to the published surveys (Databox, Orbit Media, HubSpot and Framer), explains how we detected a redesign and where the method can mislead, and ends with what the numbers mean for planning your next one.

How often companies redesign, by company size

We split the sample into three tiers of software companies, because their websites are comparable and their archives are deep. Large means public companies with thousands of staff (Salesforce, Adobe, IBM, Cisco, ServiceNow and others). Mid means venture-backed scale-ups of roughly 150 to 2,500 people (Asana, Gong, Zapier, Webflow, Intercom and others). Small means teams of roughly 50 or fewer, mostly bootstrapped (Plausible, Tuple, Feedbin, Balsamiq and others). Headcounts are approximate.

The middle tier is the outlier, and the reason is easy to see in the archive. Scale-ups change positioning often: new product lines, new segments, a new funding round with a new story. Intercom's homepage changed design four times in the period and Webflow's six times, several of those in the last three years. Large companies move more slowly, probably because their sites sit on enterprise platforms with global templates; ServiceNow's and Workday's 2021 redesigns both arrived with an entirely new stylesheet bundle, the mark of a platform change rather than a reskin. Small teams often keep one design for years and edit the words.

The per-site record

Every included site is listed, with the half-years in which a new design first appeared. H1 is January to June, H2 is July to December.

The spread inside each tier is wide. Among large companies, HubSpot's homepage kept one design system from 2017 to 2026 while Autodesk changed three times in five years. Among scale-ups, Ahrefs changed once and Webflow six times. Size predicts the average, not the individual.

How we detected a redesign, and where the method can mislead

The Wayback Machine's public CDX index lists every capture of a URL with its date and status. For each homepage we took the first successful HTML capture in each half-year from January 2016 to October 2026, up to 22 per site, and fetched the raw archived HTML.

For each capture we collected the set of CSS class names that contain a hyphen or underscore, which is how most design systems name things (for example `hero_title` or `btn-primary`). We dropped bare and numbered tokens because sites built with CSS-in-JS generate random class names on every deploy. We then compared each capture with the previous one using Jaccard similarity: shared classes divided by all classes. A redesign was recorded when similarity fell below 0.2 and stayed below it at the following capture, so a one-off A/B test or broken capture does not count. Changes in back-to-back captures were merged into one redesign.

We checked a sample of detections by hand. Each one we checked came with a new headline and a new stylesheet file: Feedbin in 2018, Workday and ServiceNow in 2021, HubSpot in 2017 and Balsamiq in 2025.

What it misses. A redesign that keeps the old class names, a redesign shipped and reverted within six months, and changes to pages other than the homepage. A brand refresh that only swaps colours and type also stays invisible.

What it may over-count. A platform migration that renames every class without changing the look. We found no such case in our spot checks, but it is possible.

Who was left out. Six sites from the original 36 had too few usable captures: Oracle, Calendly and Carrd (pages built mostly in JavaScript or with hashed classes), Pirsch, and Miro and Typeform, which returned no archive captures at all.

The threshold matters, so we ran the count at four settings. The ranking of the tiers never changed.

At looser thresholds, smaller partial refreshes start to count, and small teams' figure drops fastest because their occasional section rebuilds now register. Mid-sized scale-ups redesign roughly twice as often as either neighbour at every setting.

What the surveys say

Surveys ask a different question, whether a company has redesigned recently, and they mostly sample small companies. They agree with the archive on one thing: a large share of companies rebuild within two to three years.

Databox. Most of the rest of the 145+ companies had redesigned within three to four years. The survey does not break results down by company size.

Orbit Media. Orbit's standing advice is three to five years, depending on the industry. Orbit is an agency, so its clients' longer lifespan is partly a claim about its own work.

HubSpot. Construction and building materials companies redesigned least often. The page does not date the survey, so treat it as older data.

Framer. Framer sells a website builder, so this is vendor research. Its public summary also says 70% of website projects get deprioritised because they are too slow or difficult to ship.

Our archive count is the first of these to split frequency by company size. It also explains why the survey numbers look high: the Databox sample is mostly small companies that self-report, and "full redesign" in a survey can include a new theme on the same structure, which our class-based method would often not count. We summarised the wider survey picture, including cost and reasons, in our website redesign statistics.

Why company size changes the cycle

Three forces show up in the archive and in the surveys.

Positioning changes faster in the middle. A scale-up adds products, enters new segments and raises money on a new story. Each of those is a homepage problem. In our sample, every mid-sized company redesigned at least once, and seven of nine had a new design first seen in July 2024 or later.

Platforms slow large companies down. Enterprise sites run on shared global templates with many stakeholders. When they redesign, it is often tied to a platform migration, and the next one waits for the next platform decision. Budgets and timelines for that kind of project are in our B2B SaaS redesign cost statistics.

Small teams edit instead of rebuilding. A founder-run site with one designer tends to keep its layout and change the words. Plausible, Tuple and Bannerbear showed no structural redesign in six to eight and a half years of captures.

The common thread is the editing model. Orbit's reasons survey lists "updating it is difficult" at 11% of redesign triggers. A site the marketing team can change without a developer absorbs small changes and lasts longer, which is the argument in our guide to a website the marketing team can edit.

When a redesign is due, whatever the calendar says

Interval rules are a poor trigger. Nielsen Norman Group's advice, in a 2015 article by Hoa Loranger, is to make incremental changes and resort to a complete redesign only when the data calls for it: when incremental fixes stop producing gains, when old technology blocks features users need, when years of patches have left the structure incoherent, when conversion is low site-wide, or when competitors serve users clearly better.

Google's own guidance adds the cost side. A redesign that changes URLs is a site move, and Google tells site owners to expect temporary ranking fluctuation while it recrawls; for a small to medium-sized site, most pages take a few weeks to move. That cost is the same whether you redesign every two years or every eight, so frequent redesigns pay it more often.

Search rankings are the part of a redesign that is hardest to win back. Our website redesign SEO guide covers the nine SEO considerations that decide whether the new site keeps them.

What this means for planning your next redesign

Use your tier as the baseline, not a universal number. If you are a growing software company, expect a structural redesign about every three years and budget for it as a recurring cost. If you are large, expect five to six years between rebuilds, usually tied to a platform decision. If you are small, a well-built site can run six years or more if the team can edit it.

In each case, check the triggers before the calendar: positioning that no longer matches the product, a CMS nobody wants to touch, and conversion that small fixes have stopped moving. From our side, the sites we build are set up so the marketing team can edit them, so the next change is an edit and not a rebuild; we would scope a redesign only when those triggers are present, and we quote per project. Our marketing websites service page explains how we run a redesign.