SEO ROI calculators, the six inputs that decide the answer and the two that most of them quietly assume

An SEO ROI calculator multiplies six numbers and divides by one. Written out, that is search volume, times the click share you expect at the position you expect, times your conversion rate, times your average order value, times your gross margin, minus the cost of the work, divided by the cost of the work. Every calculator in this category is that, with a nicer interface. Which means the output is only ever as good as the two inputs people guess rather than measure, and knowing which two is the whole skill.

Google Keyword Planner reports 170 searches a month in the United States for this group of phrasings, read on 31 October 2026 against the United States geo target, taken as the group maximum rather than the sum.

The six inputs, and where each one honestly comes from

• Search volume. From a keyword tool, for the market you actually sell in rather than a global total. A summed figure across countries is the most common way this calculation starts wrong.
• Click share at your target position. The share of searches that click a given result. This is assumption one, and it is the big one.
• Conversion rate. From your own analytics, for organic traffic on comparable pages. Not a category benchmark.
• Average order value. From your own order data, for the segment the page would attract.
• Gross margin. Because revenue is not return. A calculator that stops at revenue has computed something, and not return on investment.
• Cost. The whole cost. Agency or salary, content production, development time, and tools.

The formula itself is not the hard part. The general version and a full worked example sit on our page about how to calculate marketing ROI, the formula, a worked example and where each number comes from, and the structure there carries straight across. What is specific to SEO is the two assumptions below.

Assumption one, the click share, which is why published calculators flatter

Most calculators ship with a click-through curve by position, and that curve is the number doing the most work in the output. Three things make it unreliable as a default.

It varies enormously by query type. A navigational query sends almost all its clicks to one result. A broad informational query spreads them. The same position number means different traffic on different queries.

It varies by what sits above you. A search result page with a map pack, a shopping carousel, four ads and an AI summary gives position one far fewer clicks than a plain ten-link page does. A single curve cannot describe both, and the share of results pages carrying those features has only gone one way.

And it is almost never sourced on the calculators that use it. If a tool will not tell you where its curve came from and when, you have an output you cannot defend to a finance team.

The honest replacement. Use your own Search Console data. Export the queries you already rank for, and build your own position-to-click-rate table from the rows where you have real impressions and clicks. It takes twenty minutes, it describes your category and your result pages, and it is a figure you can source when somebody challenges the business case.

Assumption two, the timeline, which no single ROI figure can hold

The second quiet assumption is that the return arrives in the same period as the cost. It does not. SEO spend is monthly from the start and the traffic arrives on a curve, so a twelve-month ROI figure computed as if month one performed like month twelve is wrong in a flattering direction.

Replace it with a month-by-month sheet. Twelve rows, cost in every row from month one, traffic ramping on your own best estimate, and a cumulative column. Two numbers then come out that a single ROI percentage hides: the month the cumulative line crosses zero, which is the payback period, and the worst cumulative figure before it does, which is the cash you need to have available. Finance teams care more about both of those than about the percentage.

The spreadsheet, in ten minutes

Six input cells at the top for the six numbers above. One table below with one row per target query, holding volume, your own click rate for the target position, sessions, conversions, revenue and margin. A twelve-row month sheet underneath for cost and ramp. Three output cells: payback month, worst cumulative position, and twelve-month return. That is the whole model, and because you built it you can change one assumption and watch what moves.

Run it three times, not once. A conservative case where you reach position eight, a central case at position four, and an optimistic case at position two. If the conservative case is still positive, the investment is defensible. If only the optimistic case works, you have a plan that requires winning, which is a different proposition and should be described as one.

What to do with the number afterwards

Compare it against the alternative use of the same money, not against zero. The honest comparison for an SEO budget is usually paid search on the same queries, because it buys the same intent at a known cost per click and it stops when you stop paying. Put both on the same sheet. The interesting output is the crossover month where the owned asset becomes cheaper per visit than renting the clicks, and that month is the actual argument for the investment.

Do the same calculation for channels with a different shape and the comparison gets more honest still. Our note on the four numbers that make a social media ROI calculation honest covers a channel where the attribution problem is harder, and reading the two together makes clear how much of any ROI figure is method rather than measurement.

Questions about SEO ROI calculators

Which input should I spend the most time on

The click rate at your target position, built from your own Search Console export. It has the largest effect on the output and it is the one every published calculator guesses for you.

Should I count branded traffic in the return

No, or count it separately and label it. Branded search would arrive largely without the investment, so putting it in the numerator credits the work with demand it did not create.

What is a reasonable return to expect

Expectation is the wrong frame, because the figure is entirely a function of your margin and your average order value. A business with a 70 per cent margin and a high order value produces a strong number from modest traffic, and a thin-margin retailer needs far more. Run your own six inputs rather than borrowing a benchmark.

How do I handle the work that has no query attached

Technical fixes and site architecture lift a whole section rather than a query. Model those as a percentage uplift applied to the existing organic revenue of the affected pages, and state the percentage as the assumption it is.

Can I trust a vendor's calculator at all

Use it to learn the structure, then rebuild it with your own numbers. A vendor's calculator with your conversion rate and your own click curve is a useful instrument. The same calculator on its defaults is a sales document.

The visits the model has no row for

Every row in this sheet starts with a search volume figure and a click. A growing share of buying questions now ends in an answer rather than a click, where an assistant names two or three suppliers and the person contacts one of them directly. Those enquiries never appear as organic sessions, so the model above understates the value of the pages that fed the answer and tells you nothing about the pages that did not. AI Knows Us measures which of your pages get used that way, which is the input this calculation is currently missing rather than a different calculation.