Skip to content

Return and measurement9 min read

When does a new website pay back?

When the extra monthly gross profit and the admin saving together have earned back the investment: simple payback is the investment divided by that monthly value. Six numbers, ten formulas and an honest explanation of what a payback model can and cannot show. Use the method with your own figures; do not treat the example as a promise.

Reviewed by
Pierre Griffin
Published
Last reviewed
Updated
Close-up of one hand writing in a ring-bound organiser while the other rests beside a calculator app on a smartphone.

The managing director of a fifty-person manufacturer has two quotes on the desk for the same website. One is twice the other, yet the descriptions are almost word for word identical: responsive design, modern technology, search-friendly structure. The finance director asks a single question: how long until we get it back? Neither quote answers it — which is unsurprising, because the answer does not sit with the supplier. It sits in the company’s own numbers.

Six of them, and every one already exists somewhere in the business. This article sets out how those six build into a payback model, where the model quietly misleads, and what has to be measured before an estimate becomes a claim anyone can check. The payback calculator runs exactly these formulas, so you can put your own figures through it.

The question carries weight because the advertising money has been moving that way for years: Hungarian digital advertising spend reached HUF 231.8 billion in 2025, close to 56% of total domestic media spending. The size of a channel, though, says nothing about whether a particular company gets its money back. Only your own figures answer that.

close to 56%

Digital share of total domestic Hungarian media spending in 2025 (IAB Hungary AdEx, built on market participants’ self-declaration)IAB Hungary · 2026 · Source

What simple payback actually means

Simple payback answers one question: how many months it takes for an investment to return its own cost. The formula is the investment divided by the additional value produced each month. That is all it is. It ignores the time value of money, discounting, tax, the cost of finance and anything happening in year three, so it is not net present value, not an internal rate of return, and not financial advice.

What it offers instead is five minutes of arithmetic anyone can check — precisely what a first screen on a capital decision needs. If the model says eight months, it is worth taking the question further. If it says ninety, a more sophisticated model is only a slower route to the same answer.

The six inputs, and where each number comes from

  1. Monthly traffic

    Visitors to the site, or enquiries received, per month. Take it from analytics rather than memory. Users and sessions are different metrics; pick one and hold to it throughout the model.

  2. Current conversion rate

    Conversions divided by traffic. The common error is to count every form submission, which quietly includes spam, duplicates, job applications and suppliers introducing themselves. Count only what could realistically become a quote.

  3. Average gross value of a new customer

    The average gross value of one new customer or order, over the same period in which you counted conversions. Put lifetime revenue in this field, multiply it by a monthly conversion count, and the model inflates badly. Start with the narrowest figure you are certain of: the first order.

  4. Gross margin

    The share of revenue left after direct costs — materials, subcontractors, delivery, transaction fees. Your accountant can give it to you in two minutes. This is the money that actually repays the investment.

  5. Administrative saving

    How many hours of repeated work the system would absorb — retyping enquiries, answering status questions, entering the same data twice — and what one of those hours costs fully loaded: salary, contributions, tooling. An hour saved only becomes money if it is genuinely spent on something else, so name the task it moves to.

  6. The investment and the assumed improvement

    The one-off spend you are considering, plus two assumptions: how far the conversion rate might improve, and how far administrative time might fall. Neither is data; both are hypotheses. Write down why you believe them — then work out what happens if only half of it lands.

The formulas, exactly as the calculator runs them

The whole model is ten lines. There is no fudge factor, no sector multiplier and no hidden correction, which is precisely why you can check it on paper.

  1. current conversions = monthly traffic × current conversion rate
  2. scenario rate = current rate × (1 + assumed improvement)
  3. scenario conversions = monthly traffic × scenario rate
  4. incremental conversions = scenario conversions − current conversions
  5. incremental revenue = incremental conversions × average gross value of a customer
  6. incremental gross profit = incremental revenue × gross margin
  7. hours saved = monthly administrative hours × the reduction
  8. administrative saving = hours saved × fully loaded hourly cost
  9. scenario monthly value = incremental gross profit + administrative saving
  10. simple payback in months = investment ÷ scenario monthly value

Three points where home-made versions of this calculation usually go wrong. The improvement is relative, not absolute: a 10% improvement takes a 1.5% rate to 1.65%, not to 11.5%. Advertising efficiency is not part of the sum: ad spend ÷ customers from advertising is a diagnostic that shows what a customer costs today, but adding it to the payback total counts the same business twice. Zero is not an answer: if the monthly value is zero, payback is not “infinite months”, it is undefined, and the calculator prints nothing rather than a misleading figure.

One worked, illustrative scenario

The company below does not exist. The assumptions sit deliberately at the cautious end. The point is to make the arithmetic easy to follow, not to forecast anything.

Illustrative assumptions — monthly figures for an imagined B2B service business, not a client result
AssumptionValueWhy this value
Monthly traffic2,000 visitorsFrom analytics, averaged over twelve months.
Current conversion rate1.5%30 usable enquiries a month from those 2,000 visitors.
Average gross value of a new customerHUF 250,000The value of a first order, not lifetime revenue.
Gross margin55%After materials, subcontractors and transaction costs.
Reducible administration30 hours a monthRetyping enquiries, answering status questions, double data entry.
Administrative hourly costHUF 6,000Fully loaded: salary, contributions, tooling, supervision.
Assumed conversion improvement10% relativeAn assumption, not a measured result or an average.
Assumed reduction in admin time15%An assumption, not a measured result or an average.
Monthly advertising spendHUF 400,000, 12 customersUsed only to diagnose today’s cost efficiency.
The calculation, step by step, on those same formulas
StepCalculationResult
Current conversions2,000 × 1.5%30 customers a month
Scenario rate1.5% × 1.101.65%
Scenario conversions2,000 × 1.65%33 customers a month
Incremental conversions33 − 303 customers a month
Incremental revenue3 × HUF 250,000HUF 750,000 a month
Incremental gross profitHUF 750,000 × 55%HUF 412,500 a month
Hours saved30 hours × 15%4.5 hours a month
Administrative saving4.5 hours × HUF 6,000HUF 27,000 a month
Scenario monthly valueHUF 412,500 + HUF 27,000HUF 439,500 a month
Diagnostic: advertising cost per customerHUF 400,000 ÷ 12about HUF 33,300 (outside the payback sum)

The final step divides the investment by that monthly value. We deliberately name no build figure here: price follows scope, and price is not what this article is about. Express it per million forints instead: 1,000,000 ÷ 439,500 ≈ 2.3 months. The model is linear at this step, so substituting your own figure is straightforward: however many millions the quote in front of you comes to, payback is that many times 2.3 months — provided every assumption above holds.

Free assessment

Does this sound familiar? Let us discuss the best place to begin.

Your assessment is carried out personally by one of our experts.

What the model cannot see

The calculation above is accurate. That does not make it true. The factors below sit structurally outside it, each has to be thought through separately, and in practice they decide more investment cases than the formulas do.

Attribution is not accounting

The numbers in advertising and analytics interfaces are not the numbers in your ledger. Google states that modelled conversions use data which does not identify individual users to estimate conversions it cannot observe directly — where consent is absent, where someone clicks on one device and buys on another, or where the browser restricts measurement — and that those conversions take days to settle fully in reporting. In Google Analytics 4, the default lookback window for key events other than acquisition is 90 days, adjustable to 30 or 60. Those are the defaults documented in September 2026: the specific numbers may change, the principle will not. Every reporting window is finite, and it can be shorter than the time your company actually takes to win business.

90 days

Google Analytics 4’s documented default lookback window for key events other than acquisition, as at September 2026 (adjustable to 30 or 60 days)Google · Source

A six-month B2B sales cycle therefore runs longer than that default window: the February enquiry that signs in August is no longer tied back to the original click. That is a limit of reporting attribution rather than a measurement fault — but in a payback model that signature is the whole point. Which is why the conversion here should come from the CRM and the invoicing system, leaving the advertising platform to do what it is genuinely good at: comparing campaigns against each other.

One month of new customers is not the whole value

The model looks at one month of new customers and treats their value as arriving with them. Where there is repeat purchase or a subscription, it understates: the margin on the third order appears nowhere. Where lifetime revenue has been entered in the average-value field, it overstates badly. Churn drops out the same way — a customer who leaves after six months is worth as much here as one who stays five years.

Four factors settled by the calendar and by capacity

  • Seasonality. Comparing an August launch against September is not measurement, it is a seasonal effect. Compare like periods, ideally year on year.
  • Capacity. Thirty enquiries a month becoming thirty-three is comfortable. Three hundred becoming three hundred and thirty may not be. The model assumes every extra enquiry is handled to the same standard; if sales are already full, the increase becomes waiting time rather than margin.
  • Implementation time. The payback clock starts at launch, not at signature, and traffic and rankings can move about for weeks after a migration. Add that time when you plan.
  • Tax and finance. Simple payback is a pre-tax, undiscounted measure. If the build is funded by borrowing or leasing, real payback is longer than the formula shows.

The improvement stays a hypothesis until it is measured

Moving from 30 conversions a month to 33 is invisible in a single month of data: ordinary month-to-month variation can exceed the difference you are looking for. That does not mean the improvement is not real; it means one month is not evidence. A/B testing does not rescue the situation either: at low conversion volumes a 10% relative difference struggles to separate itself from chance, and the test can run for months without settling the question. What remains is a longer measurement window agreed in advance, like-for-like seasonal comparison, and honest wording: this is an estimate, not a result.

How to make the assumption checkable

The difference between a serious payback calculation and a decorative one is not the formulas. It is whether anyone can tell afterwards if the assumption held. That is decided before the project, not after it.

  1. Write down what counts as a conversion. Not “form submission”, but something like: a completed enquiry from a real company that the sales team quotes for. That definition must survive to the end of the project unchanged.
  2. Measure a baseline for at least three months, ideally twelve. Without a baseline, the post-launch number has nothing to be compared with.
  3. Connect the enquiry to the quote and to the order in the CRM. The final proof of a website’s return is an order, not a visitor count.
  4. Choose one source of truth for money — invoicing or the ledger — and align every report to it. If three systems produce three numbers, the meeting is about the numbers rather than the decision.
  5. Allow a settling period after launch, and say in advance that the first weeks will be volatile.
  6. Six months after launch, run the same model again on measured figures. That is the moment an estimate becomes a fact, or turns out not to have held.

When the model says no

A model earns its place by being able to say no. In five situations a new website is the wrong answer to a good question — and it is far cheaper to find that out now.

  • Traffic is too low. At 200 visitors a month, improving the conversion rate produces a fraction of an enquiry. The constraint is demand, not the site, and the money goes further on demand generation or on the offer itself.
  • The margin is thin. At a 10% gross margin, nine-tenths of the extra revenue goes straight back out in direct costs, and payback stretches from months into years.
  • Only the administrative saving creates value. If the gross-profit line is close to zero, this is not a website project. It is an internal system or an automation project, and it pays back elsewhere.
  • Sales cannot handle today’s enquiries. More enquiries then produce longer response times, not more business.
  • You do not know what a customer is worth. That is a missing input rather than a fault in the model, and until it exists every payback figure only looks like a number.

What to do next

  1. Collect the six numbers, and note which system each came from. Mark the ones you had to estimate — later, that will be the most valuable annotation on the page.
  2. Put your own figures through the payback calculator. The values you type stay in your own browser; they reach us only if you deliberately attach them to an assessment.
  3. Write down the improvement you consider realistic and why, then run the same calculation at half of it. If half still pays back acceptably, the decision becomes considerably easier.
  4. Ask every supplier what measurement they will set up, who has access, and who keeps the data when the engagement ends. Anyone vague about that cannot prove the result later either.
  5. If an outside view would help, the website review and recommended action plan modules of the free digital assessment look at exactly these questions: where the site loses enquiries today, and in what order to address it. One of our experts reviews the site personally; it is not an automated software audit.

If the numbers say it is not worth doing right now, that is a usable answer too — cheaper to establish today than to discover a year from now. If it is worth doing, the next decision is no longer whether to build a new site, but what that site has to do for the assumption to hold.

Sources

  1. IAB Hungary — AdEx 2025 and AdExpect 2026

    Hungarian digital advertising spend reached HUF 231.8 billion in 2025, 11.2% more than in 2024, and close to 56% of total domestic media spending.

    Hungary, calendar year 2025, digital advertising spend. Search advertising accounted for HUF 76.3 billion of the total.

    https://iab.hu/2026/03/27/232-milliard-forintos-digitalis-reklampiac-szerenyebb-idei-varakozasok-megjelent-az-adex-2025-es-az-adexpect-2026/Last reviewed: 1 September 2026

  2. Google — About modeled conversions — Google Ads Help

    Google states that modelled conversions use data that does not identify individual users to estimate conversions Google is unable to observe directly, with documented causes including unconsented users, cross-device journeys and browsers that limit conversion measurement. Modelled conversions can take up to five days to fully process and stabilise in Google Ads reporting.

    Applies to Google Ads reporting, from an undated help page, as it stood on 2026-09-01.

    https://support.google.com/google-ads/answer/10081327Last reviewed: 1 September 2026

  3. Google — Lookback window — Google Analytics Help

    In Google Analytics 4 the default lookback window for key events other than acquisition is 90 days, adjustable to 30 or 60 days; for acquisition key events (first_open, first_visit) the default is 30 days, adjustable to 7.

    Google Analytics 4 reporting attribution, from an undated help page, as it stood on 2026-09-01.

    https://support.google.com/analytics/answer/16291704Last reviewed: 1 September 2026

Free assessment

Four areas. One clear next step.

Choose what we should look at. Our experts review it and return a specific recommendation.

Insights