Will a new website pay for itself? The sum to do first
Short answer
Divide the price by what one new customer leaves you after costs. That is how many extra customers the site has to win. Then check whether your traffic can supply them within a year. The inputs come from your own analytics; benchmarks only tell you if they look odd.
Contents 7 sections
Most people ask what a new site will earn. Nobody can answer that before the site exists, me included. The better first question is what it has to earn, because you can answer that today, with numbers you already have.
What is the sum?
Two divisions.
- Customers needed = price ÷ margin per customer. Margin is what one new customer leaves you in their first year, after the cost of serving them. Gross profit, not revenue.
- Months to pay back = customers needed ÷ extra customers a month. “Extra” means the customers the new site wins on top of what the old one would have won anyway.
If one customer leaves you a tenth of the price, the site needs ten of them. If it wins one more a month than the old site would have, it has paid for itself in ten months. Two more a month, five months.
- Fig.
That is the whole calculation. Everything hard is in the inputs.
Where do the inputs come from?
From your own tools, not from a benchmark.
| Input | Where to find it |
|---|---|
| Visitors a month | Your analytics, averaged over the last twelve months |
| Share who enquire, sign up or buy | Your analytics goals, or form entries divided by visitors |
| Share of those who become customers | Your CRM, or your head of sales |
| Margin per customer, first year | Finance: what a customer pays you, minus what serving them costs |
Multiply the first three and you have the customers your site wins a month today. Use the last twelve months, not the best one: seasons are real. If you can only find revenue per customer and not margin, the answer will flatter the site. Use it, but know which way it leans.
What do benchmarks tell me?
Whether your numbers look odd. Not much more.
The median conversion rate of a landing page, across all industries. More than 41,000 pages.
Source: Unbounce, 2024
The median for software (SaaS) landing pages.
Source: Unbounce, 2024
A landing page is built for one action, and its traffic is usually sent there on purpose. I would expect a whole website to sit lower, and an enquiry is not a customer. Unbounce counted conversion events, not what each one was worth.11Unbounce, Conversion Benchmark Report, 2024: more than 41,000 landing pages, 464 million visitors and 57 million conversions. It counted conversion events, not the value of each. On reading level it names one exception: “Financial services is the only industry where more complex language can actually help conversions.” unbounce.com, checked 11 November 2026. If your rate is far from these, check how you count before you conclude anything.
What have measured site changes moved?
These are results someone measured and published, with the method beside them. None is a promise about your site.
More sales from a faster version of one landing page, in an A/B test on paid traffic split 50/50.
Source: web.dev, Vodafone, 2021
More visitors reaching the form on lead-generation sites, for a mobile site 0.1 s faster. 37 brands, 30 days.
Source: 55 and Deloitte for Google, 2019
Rakuten 24, an online shop for everyday goods in Japan, ran the same kind of A/B test on one landing page for about a month. The faster version had a 33.13% higher conversion rate and 53.37% more revenue per visitor.22“How Rakuten 24’s investment in Core Web Vitals increased revenue per visitor by 53.37% and conversion rate by 33.13%”, web.dev, last updated 24 August 2022. Version A and version B looked and worked the same; only the speed work differed. web.dev, checked 11 November 2026.
Two more, with less certainty attached. Baymard says the average large shop “can gain a 35.26% increase in conversion rate” through better checkout design.33Baymard Institute, cart abandonment statistics. The full sentence scopes it to “the average large-sized ecommerce site”. baymard.com, checked 11 November 2026. That is an estimate of what is possible, built from its usability research, not a before and after. Unbounce found landing pages written at a 5th to 7th grade reading level convert at 11.1%, against 5.3% for professional-level writing.1 That compares different pages, not one page rewritten.
What does it look like with real numbers?
Here is an example. The numbers are made up to keep the sum easy, so replace them with yours. Say one customer leaves you a tenth of the price, so the site needs ten extra customers to pay back. Today it wins ten customers a month.
| Lift on today’s rate | Pays back in |
|---|---|
| +5%, half a customer more a month | 20 months |
| +10%, one more a month | 10 months |
| +20%, two more a month | 5 months |
Pick the payback time you would accept, say a year, and read the table from the top. If only the bottom row gets there, the project is a bet on a big lift. If the top row gets there, it is an easy decision. If it takes the middle row, check the inputs twice, mine included.
The payback calculator does both divisions with your own numbers and my published prices. Its lift starts at zero, so nothing is assumed for you.
When will a new website not pay?
The sum fails in the same few places.
- Too little traffic. If 500 people a month visit and 1% enquire, that is five enquiries. No layout turns five into fifty. Spend on being found first.
- Nobody follows up. A site can fill an inbox. It can’t call anyone back. If enquiries wait days for an answer, fix that before the site.
- The offer is the problem. If people read the page, understand it and still don’t buy, the price or the product is the issue. A redesign hides that for a few months.
- The sales cycle is long. If a deal takes nine months to close, payback is real but late. In the first year, count pipeline, not revenue.
What do I measure before and after?
The same numbers, counted the same way, before I change anything and again after launch.
- Visitors a month, by where they came from
- Conversion rate on the pages that matter: the home page, pricing, the main offer
- Enquiries or sign-ups a month, and how many become customers
- Real-visit speed for the main pages
A before and after is not an experiment. Prices, ads and seasons change while the new site is live, and any of them can move the numbers. Nielsen Norman Group says it plainly about design ROI: “They are not financial projections.”44Kate Moran, “Calculating ROI for Design Projects in 4 Steps”, Nielsen Norman Group, 23 February 2020. The sentence before it calls ROI calculations “strategic exercises”. nngroup.com, checked 11 November 2026. They are a way to decide whether a project is worth doing. That is all this sum is, and it is enough.
Do the sum on your own numbers in the payback calculator.
Notes and sources
-
Unbounce, Conversion Benchmark Report, 2024: more than 41,000 landing pages, 464 million visitors and 57 million conversions. It counted conversion events, not the value of each. On reading level it names one exception: “Financial services is the only industry where more complex language can actually help conversions.” unbounce.com, checked 11 November 2026. ↩︎ ↩︎
-
“How Rakuten 24’s investment in Core Web Vitals increased revenue per visitor by 53.37% and conversion rate by 33.13%”, web.dev, last updated 24 August 2022. Version A and version B looked and worked the same; only the speed work differed. web.dev, checked 11 November 2026. ↩︎
-
Baymard Institute, cart abandonment statistics. The full sentence scopes it to “the average large-sized ecommerce site”. baymard.com, checked 11 November 2026. ↩︎
-
Kate Moran, “Calculating ROI for Design Projects in 4 Steps”, Nielsen Norman Group, 23 February 2020. The sentence before it calls ROI calculations “strategic exercises”. nngroup.com, checked 11 November 2026. ↩︎
Written by Dion Zeneli. AI helped with research and editing. Sources are linked where they are used. Read this note as markdown.