SEO ROI: how to measure and forecast it like a CFO

SEO ROI is the net profit your organic channel generates relative to its cost. But a defensible figure reconstructs revenue from organic conversions, not from a calculator’s assumptions.

Here is the difference that breaks most numbers (illustrative). A page doing 100 visits a month at a 5% conversion rate and a $500 average order value makes $2,500. A page doing 10,000 visits converting at 0% makes nothing. SEO ROI is revenue-first, not traffic-first, and almost everything ranking for this term forgets that.

Search this term and you get two things: a vanity calculator that asks for three inputs and spits out a tidy figure, or an “average ROI” statistic you could never defend in a finance review. Neither survives ten minutes with a CFO.

This guide is different. These are the models we actually run for clients: the four-layer attribution we use to measure SEO ROI, the top-down method we use to forecast it, and the honest hedges we put on every number. We will show you how to build an SEO ROI number you can stand behind across a whole SEO campaign, and how to forecast it without pretending the future is certain. Treat the genre’s SEO ROI statistics as marketing, not method: the numbers below are ones you can rebuild yourself.

How to calculate SEO ROI (and the formula everyone stops at)

To calculate SEO ROI: (1) total your cost of SEO, including retainer, content, tools and internal time; (2) measure the revenue the channel produced, in layers, starting from last-click conversions; (3) subtract cost from revenue and divide by cost; (4) express it as a percentage and attach a confidence range.

The canonical SEO ROI formula is simple. Return on investment equals (SEO revenue − cost of SEO) ÷ cost of SEO, expressed as a percentage. In plain finance terms, it is the same return on investment your CFO calculates for any capital allocation, and the same way you would size the return on any other marketing channels: what you got back, minus what you put in, divided by what you put in. Search engine optimization is just another line item until you can express it this way.

So far, so textbook. The problem is what the formula assumes, and what every calculator quietly skips.

Here is the metric precision point that almost no competitor teaches: clicks are not sessions, sessions are not leads, and leads are not conversions. They are four different numbers, and they get smaller at every step. Measure the wrong one and every downstream figure in your ROI of SEO is wrong. A page can earn 1,000 clicks, log 900 sessions in Google Analytics, generate 30 leads, and close 6 conversions. If you build your SEO return on 1,000, you have overstated the result by orders of magnitude. Most articles use these words interchangeably. We do not, and neither should you.

The second assumption is attribution. The formula needs clean revenue: revenue you can prove came from organic search and from your wider search engine optimization efforts. You do not have that yet. Not cleanly. We will get to why.

What is the ROI of SEO? It is the profit organic search generates against what you spent to earn it. A positive ROI means the channel returns more than it costs; the harder part is proving which revenue belongs to it.

The SEO ROI calculator trap: why a tidy number isn’t a forecast

Every SEO ROI calculator ranking for this term works the same way. The typical SEO ROI calculator asks for monthly search volume, an average order value, maybe a customer lifetime value, multiplies them together, and hands you a confident dollar figure. It looks like maths. It is mostly assumption.

Look at what the typical ROI calculator hides. It assumes a flat conversion rate across every keyword, when intent varies wildly. It uses no attribution model, so it credits organic search for revenue other channels helped close. It assumes results land instantly, ignoring the ramp that real SEO takes. And it gives you a single number with no confidence interval: no honest range around the guess.

Then there is the honesty layer the calculators never mention. Your own measurement tools are lossy. Google Search Console anonymises roughly 46% of queries, so nearly half your keyword data, and your view of keyword rankings, is hidden before you start. And GA4 typically underreports organic traffic by 15–30% (per Suganthan Mohanadasan of Snippet Digital). Any single, tidy number quietly papers over both gaps, which is why a fixed SEO budget set against a calculator output is a guess dressed as a plan.

An SEO ROI calculator is fine for a back-of-envelope sense check on a new SEO campaign. It is not a forecast, and it is not a measurement. The SEO ROI statistics it leans on are industry averages, not your numbers. Treat the whole thing as the conversation starter, not the answer: a real SEO strategy needs a model built on your own data.

How to actually measure SEO ROI: the four-layer attribution model

This is the model we run. When we measure SEO ROI for a client (the way we calculate SEO ROI on every retainer) last-click is the floor, not the ceiling, so we build up from it in four layers. Each layer recovers value the layer below it misses, and together they turn a vague SEO strategy into a number a finance team accepts.

Layer 1, last-click organic revenue (the floor). Start with the conversions Google Analytics attributes directly to organic search as the last touch. This is the most conservative, most defensible number you have. It undercounts, deliberately. That is the point of a floor.

Layer 1.5, audience-based ROAS with a 30-day window. (The half-step numbering is deliberate: it widens Layer 1’s lens without changing the attribution model.) Widen the lens. Look at users who arrived via organic search and converted within 30 days, even if their final session came from elsewhere. This catches the buyer who found you in an unpaid listing, left, and came back direct a week later.

Layer 2, assisted and first-touch conversions (the conversions last-click hides). Pull the assisted conversions and conversion paths report in GA4. Organic is frequently the first touch, the channel that introduced the customer, without ever being the last. Last-click attribution erases that entirely. This layer puts it back.

Layer 3, CPC-equivalent traffic value (what this traffic would cost in Google Ads). Take your non-branded organic clicks and price them at what the equivalent keywords would cost in paid ads. This is the replacement-cost value of your organic traffic. It is not revenue, but it answers the CFO question: “what would we pay to buy this traffic if we lost it?”

Google Analytics mechanics: where to audit each layer

Each layer ties to real GA4 mechanics (assisted conversions, multi-touch conversion paths, channel groupings) so anyone can audit your working. And every layer carries the same honest caveat: this is directional, not accounting-grade. We say so out loud. A model that pretends to be precise is less trustworthy than one that shows its range.

This is the discipline a good SEO agency or SEO consultant should bring to your SEO investment. If you want it built properly for your business, it is core to our SEO services. Mobile optimization matters here too: split desktop and mobile in your reporting, because conversion behaviour differs sharply between them.

Splitting brand vs non-brand revenue

Claiming brand-search revenue as an SEO win is the fastest way to inflate your ROI of SEO and the fastest way to lose credibility in a finance review. People searching your brand name were already coming. Separate branded and non-branded queries in GSC, attribute the branded revenue elsewhere, and report non-brand organic search as the figure SEO actually earned. The honest number is smaller. It is also the one you can defend.

SEO revenue attribution: solving the ‘(not provided)’ problem

Since Google’s 2013 “(not provided)” change, keyword-level conversion data has been dark. Google stopped passing the search term into your analytics, so you can no longer see which keyword drove which sale. Calculators pretend this never happened. It did, and it is permanent.

Here is the reconstruction method we use to estimate organic revenue despite the gap (the approach is attributable to Suganthan Mohanadasan of Snippet Digital). Join your Google Search Console and Google Analytics data in BigQuery. Narrow each match by landing page, date, and device so you are pairing the right query to the right session. Then weight each query by intent, because not all organic traffic is worth the same. These are tunable defaults, not laws:

Intent type	Weight
Branded	3.0×
Transactional	2.0×
Commercial	1.5×
Navigational	1.0×
Informational	0.5×

Finally, score every figure HIGH, MEDIUM, or LOW confidence based on how much of the underlying data was hidden or inferred. A figure built mostly on anonymised queries scores LOW, and you flag it.

The mandatory hedge, on every output: this is an estimate for prioritisation, never for financial reporting. It tells you where to focus the next SEO campaign and where your SEO efforts will earn the most. It does not go in the audited accounts. Done well, this is how you express search engine optimization as a true SEO return on investment rather than a pile of search engine impressions nobody can value.

One more note for lead-gen businesses. If you do not sell online, assign a value per conversion first (what a qualified lead is genuinely worth) or your output is a count of conversions, not dollars. Counts do not calculate ROI. Values do.

Get this in your inbox

One AI search tactic, every Friday.

AI On Fire: a working note from the Firewire desk, sent to a few thousand Australian search marketers every Friday.

Subscribe

No spam. Unsubscribe anytime.

How to forecast SEO ROI like a CFO

Measuring the past is the easy half. To forecast SEO ROI you need a model that does not collapse the moment someone questions an input.

The bottom-up approach (keyword volume × CTR × conversion rate) is what most forecasts use, and it breaks fast. It assumes a flat conversion rate, it ignores the ramp, and it stacks estimate on estimate until the error compounds. It feels rigorous. It is fragile.

We use top-down instead. A forecasting model popularised by Eli Schwartz starts with total addressable market and works down: TAM → the demographic universe → the in-market percentage → purchase frequency → the share who use search → a click-potential modifier → realistic market penetration → conversion rate → average order value. Your CFO recognises TAM-based modelling immediately, because it is how they size every other business case. That is the point: you are speaking their language, not SEO’s.

Then add the AEO modifier. Split organic demand into click-generating behaviour versus answer-engine behaviour, where the user gets their answer in an AI result and never clicks. Baseline this from your own GSC impression-to-click ratio on non-branded queries, and assume it is lower than you think. People are clicking less.

Run three scenarios. Conservative captures 50% of modelled potential, and this is the default we put in proposals. Moderate captures 75%. Aggressive captures 100% and rarely happens. Ramp each over twelve months: roughly 10–20% of full potential in months 1–3, 30–50% in months 4–6, 60–80% in months 7–9, and 80–100% by months 10–12. SEO does not arrive on day one.

Wrap every figure in a confidence interval and present it as a range, never a guarantee: ±15–25% at one to three months, ±25–35% at four to six, and ±30–50% by seven to twelve months out. The further you forecast, the wider the band, which is exactly what a CFO expects to see.

Adjusting your forecast for the AI-search era

The click-potential modifier is not optional any more. Ahrefs’ analysis of 300,000 keywords found AI Overviews cut clicks to the #1 result by about 58%, and even position 10 lost roughly 19.4%. Per the same Ahrefs analysis, AI Overviews appear on around 21% of keywords, climbing from 9.5% on single-word queries to 46.4% on queries of seven-plus words. ChatGPT has roughly 12% of Google’s search volume, yet Google still sends about 190× more traffic (all Ahrefs, Q1 2026; decimals approximate). 2023-era ROI maths overstates returns because it assumes clicks that AI answers now absorb.

Ranking #1 isn’t the same as being seen: the pixel-position reality

Position one is not what it used to be. Moz’s Tom Capper, analysing STAT data across 46,000 terms in March 2026, found the median #1 organic desktop result now sits 590 pixels down the page. On Capper’s viewport assumptions, that puts the top organic result above the fold on only about 65% of typical laptops: a third of searchers never see “#1” without scrolling. CTR curves in every legacy calculator assume a visibility that AI Overviews and SERP features increasingly take for themselves. Rank one, get seen by two-thirds. Build that into the maths.

Is SEO worth it? Realistic returns, honestly

Here is the honest answer: yes, for most established businesses, with no multiple and no guarantee attached. Built and measured properly, an SEO programme earns a positive ROI: the question is the size of that positive ROI, the time horizon, and the SEO cost behind it.

They make good headlines. They make terrible forecasts.

Now the critique the genre earns. Third-party roundups love to publish figures like a 748% median ROI, or 4:1 to 10:1 return ratios. We present those critically, because they are not ours and they do not map to your business: a B2B SaaS average tells a local services brand nothing, and no finance team will accept a borrowed benchmark as a projection. They make good headlines. They make terrible forecasts.

On timing, third-party benchmarks such as FirstPageSage suggest positive SEO ROI typically emerges in 6–12 months and peaks in years two to three. Useful as context; timelines vary by market, competition, and starting authority, and none of it is a Firewire promise. On cost, an SEO campaign with us starts from $2,500 per month; see what SEO costs in Australia for the full picture.

The structural advantage is the part the numbers undersell: SEO compounds. Unlike paid ads, the asset does not switch off when the spend stops, so your SEO efforts keep paying out long after the work is done, and a thought leadership SEO campaign compounds the same way, building authority that lowers the cost of every ranking that follows. A ranking page keeps earning its position in the search engine results after the invoice is paid. Every search engine still rewards the pages that earn relevance, not the ones that rent it.

Frequently asked questions

What is the ROI of SEO?

The ROI of SEO is the profit organic search generates relative to its cost: (SEO revenue − cost of SEO) ÷ cost of SEO × 100. The number is only as good as the attribution behind it: reconstructed organic revenue, not a calculator’s flat assumptions.

Is SEO dead or evolving in 2026?

Evolving, not dead. AI Overviews and answer engines are shifting clicks away from traditional results, so you adjust the click-potential maths and the forecast. The channel still drives revenue; you measure it more honestly and stop assuming 2023-era click curves.

Is a 22% ROI good?

Short version: 22% is a real, positive return. But the figure alone can’t tell you whether it’s good: a number means nothing without the attribution model and time horizon behind it. A defensible 22% built on last-click is stronger than a 700% built on a calculator. Ask what is behind the number.

How long until SEO pays back?

Third-party benchmarks suggest 6–12 months for positive SEO ROI, with peak returns in years two to three. That is industry context, not a guarantee: your payback depends on competition, starting authority, and how aggressively the SEO campaign is funded. We forecast a range, never a fixed date.

Build a model you can defend

A real attribution model, honest revenue reconstruction, and a top-down forecast beat any calculator or borrowed “average”, because they survive a finance review. That is the whole game: an SEO ROI number your CFO trusts.

Want one built for your business? Have Firewire build a defensible SEO ROI model (measurement and forecast) through our SEO services, and check what SEO costs in Australia before you decide. No guaranteed returns. Just numbers you can stand behind.

This guide is general information, not financial advice. Model your own numbers before committing budget.