Context
What is IA ROI, the agent that works out your SEO ROI?
It is an AI agent that answers the question everyone asks before hiring SEO: how much should I invest, and when will I get it back? It does so with your figures and public data, not with a hidden formula.
Most tools that promise to calculate seo roi give a «lost revenue» figure without showing where it comes from. This agent shows the formula, the source and date of every figure, and a range instead of a magic number.
It also accounts for what has changed: AI Overviews keep part of the clicks, and assistants such as ChatGPT or Gemini are starting to send visits. That is why it compares classic SEO with SEO + AEO + GEO.
Why now?
Architecture
How does the agent work out your SEO ROI?
Like my other agents, such as IA Ocupación: code does what code can do, and AI decides and explains only where it adds something.
Facts
From your answers, code describes your business: sector, how customers arrive, sales cycle, clicks, value of a sale, margin, conversion, presence in AI and investment. No AI.
Simulation
Code simulates three scenarios month by month with each source’s benchmarks, and repeats it with their pessimistic and optimistic values to give you a range.
SubQ decides
For each lever of the return (AI Overviews erosion, sales cycle, conversion, margin, lead close rate…) it answers two yes/no questions: does it weigh on your business? Is it a priority now?
Mistral explains
It explains which scenario the figures favour and why, using only those levers and numbers, and suggests a first step.
You decide
You see it in 3D charts, fine-tune your exact figures with sliders and decide your budget.
You do not type anything: you pick closed options. The same combination always gives the same result, so it is cached for a week.
Before you use it, the charts show a demo of the business you are describing, worked out by code only, without calling any model.
Simulation
Which three scenarios does the agent compare?
The agent measures the gain of investing against doing nothing, because doing nothing has a cost too.
| Criterion | What it assumes | What it measures |
|---|---|---|
| No change | You do not invest. Your organic clicks fall where AI Overviews keep part of them, at the pace measured in 2023–2025. | The gross profit you lose against your traffic today. |
| Classic SEO only | You invest in traditional SEO. Your clicks grow with the scenario you pick, but the AI Overviews erosion continues. | Extra gross profit over «No change», minus the investment: payback month and ROI. |
| SEO + AEO + GEO | The same investment, also so that AI picks and cites you: you recover part of the erosion and gain visits from AI assistants. | The same, with its pessimistic–optimistic range. |
Data
Which sources does the agent take its data from?
Every benchmark carries its source, date and a reliability grade. I only use studies with a declared sample, never blog round-ups.
| Criterion | Range | Source | Date |
|---|---|---|---|
| Searches in Spain that show an AI Overview (midpoint of two studies) | 15.3 % – 33.2 % | Ahrefs (15,3 %, nov. 2025) y SE Ranking (33,2 %, may.–jun. 2025) | 2025-11 |
| Clicks lost on a search when an AI Overview appears | 38 % – 58 % | Pew Research (8 % frente a 15 %, mar. 2025), Seer Interactive (−38 %, feb. 2026) y Ahrefs (−58 % en la posición 1, dic. 2023–dic. 2025) | 2026-02 |
| Share of a website's traffic that comes from AI assistants (Spain) | 0.25 % – 0.32 % | SE Ranking, 101.574 webs con Google Analytics (0,30 % en España, ene.–abr. 2026); Ahrefs (0,25 %, 2025) | 2026-04 |
| Share of the traffic sent by AI chatbots (Spain) | ChatGPT 77.8 % · Gemini 14.4 % · Perplexity 3.9 % | Statcounter Global Stats (CC BY-SA 3.0) | 2026-09 |
| Conversion of a visit from AI against another visit (multiplier) | ×1 – ×1.54 | Adobe Digital Insights, EE. UU. (+28 % en viajes, +54 % en retail, may. 2026); Contentsquare la sitúa por debajo de la búsqueda de pago, por eso el mínimo es 1 | 2026-05 |
| Average conversion of an online shop (UK and Ireland) | 1.07 % – 2.83 % | IRP Commerce, datos de mercado (sep. 2026) | 2026-09 |
| Conversion by sector (lead generation) | landing-page medians × 60% | Unbounce, Conversion Benchmark Report 2024 | 2024 |
| Monthly customer churn on subscriptions (derived from annual retention) | 2.1 % – 4.9 % | ChartMogul, SaaS Retention Report 2023 (más de 2.100 empresas; retención anual del 54,8 al 77,7 %) | 2022 |
Assumptions, shown as such
- Part of that click loss that has not reached your site yet (AI Overviews reached Spain in March 2025): 50 % (25 % – 75 %).
- Leads that become customers, if you do not know: 12 % (5 % – 25 %).
- Gross margin, if you do not know: 40 %. Cost per hour of the in-house team: €30.
- Your scenario (you choose it): prudent +10 % organic clicks and ×1.5 AI visits; medium +25 % and ×2.5; ambitious +50 % and ×4, reached gradually between months 2 and 12.
Where two sources disagree, the agent does not pick the more favourable one: it uses the range between them, and that feeds the pessimistic and optimistic scenarios.
Visualisation
What do the agent’s 3D charts show?
Three views of the same calculation. Drag to turn them and replay the months.
Cumulative return
One ribbon per scenario, drawn month by month on a neumorphic base. It crosses a frosted glass plane in the month the investment pays back, marked in orange, and a band marks the pessimistic–optimistic range.
Channels
Organic clicks, the clicks AI Overviews keep (in red, on top) and visits from AI assistants, on their own scale because they weigh little today.
Sensitivity
A landscape with the ROI if your conversion or the value of your sale were different. The dark line separates gain from loss and the orange dot is your case: it shows which lever moves the result most.
Accessible table
The same data in a table, for screen readers and for anyone who prefers numbers.
After the calculation, sliders let you enter your exact figures. Code recalculates them in your browser, at no cost and with no AI.
Reliability
How does the agent check what it tells you?
Mistral only drafts. Before you read it, SubQ checks every sentence.
Nothing beyond the figures
SubQ checks that each explanation rests on its lever and the simulated numbers, adding no causes or figures.
No promises
SubQ checks that the text promises no rankings, citations, traffic or return, and presents the figures as an estimate.
Second draft
If anything fails, a larger Mistral model rewrites it and SubQ checks again. If doubts remain, you are told.
Visible cost
You see how long each answer took and what it cost in SubQ and Mistral.
Privacy
What does the agent do with your data?
It is designed not to need personal data.
Closed options only
You do not type free text, so nothing you enter can manipulate the models.
No personal data
It asks for no name, email or URL. IP addresses are only used in memory for usage limits and are not stored.
Abuse limits
An «I’m not a robot» proof of work, three calculations a day per visitor and a global daily cap.
European models
Mistral is a European model; SubQ only returns yes/no probabilities on data already calculated.
Honesty
What can the agent not do?
A good budget starts by knowing what a tool measures and what it does not.
What it does
- It compares three scenarios with public data, each with its source and date.
- It gives you the payback month and an ROI range, not a magic figure.
- It tells you which levers weigh most in your case.
What it does not do
- It does not look at your website or Search Console: that is what the AI visibility checker is for.
- It does not know whether ChatGPT cites you today: that is measured with real questions, as I do in the audit.
- There is no reliable public data on lead close rates: if you do not give yours, it uses a wide assumption.
If you want the calculation with your real data, I do it as part of Search & AI Visibility.
Key points
- Facts: From your answers, code describes your business: sector, how customers arrive, sales cycle, clicks, value of a sale, margin, conversion, presence in AI and investment. No AI.
- Simulation: Code simulates three scenarios month by month with each source’s benchmarks, and repeats it with their pessimistic and optimistic values to give you a range.
- SubQ decides: For each lever of the return (AI Overviews erosion, sales cycle, conversion, margin, lead close rate…) it answers two yes/no questions: does it weigh on your business? Is it a priority now?
- Mistral explains: It explains which scenario the figures favour and why, using only those levers and numbers, and suggests a first step.
- You decide: You see it in 3D charts, fine-tune your exact figures with sliders and decide your budget.
FAQ
SEO ROI calculator: frequently asked questions about the agent
How do you calculate SEO ROI?
ROI = (extra gross profit − investment) ÷ investment. The extra profit is the difference against doing nothing, not against zero, because without investment your clicks change too. The agent shows the full formula and the data it uses.
How do you measure SEO ROI before investing?
With a simulation from your own figures and public benchmarks, shown as a range. The agent compares three scenarios month by month and tells you whether, and when, the investment pays back.
How long does an SEO investment take to pay back?
It depends on the sales cycle and the profit per customer. The agent assumes the improvement starts to show from month 2 and is complete around month 12, and gives you the exact month the cumulative return crosses the investment, if it does.
Why does it account for AI Overviews and ChatGPT?
Because they change the return: Pew, Seer and Ahrefs measure between 38 and 58% fewer clicks when an AI Overview appears, and AI assistants already send around 0.3% of the traffic of Spanish websites.
Is the estimate reliable?
It is as reliable as its data, which is why you see all of it: source, date and grade. The result is an estimate with a range, useful to set a budget, not a forecast of your business.
How much does the agent cost?
Nothing. You get three calculations a day; combinations already calculated are answered from the cache at no cost. Fine-tuning with your exact figures is unlimited because your browser does it.
Next step
Want this calculation with your real data?
In Search & AI Visibility I measure your traffic, conversions and AI citations, and turn the budget into a 5-month plan with measurable goals.
This article was created with the help of AI and reviewed by José Galán. I take great care over every post and every translation, but the odd mistake can still slip through. If you find one, write to me: you will be helping me improve.