Cost of invisibility calculator
Work out how much you leave to your competitors every month because customers cannot find you in Google or in AI answers. Six questions you can answer off the top of your head. Instant result, no email needed.
Your situation in six questions
You do not need Google Analytics or Search Console. You can fill in every field from memory in about a minute. The result updates as you go.
Model assumptions you can change
Pew Research Center measured 18% of all searches in March 2025. The figure is growing, so you can raise it.
The share of queries put straight to an AI assistant instead of a search engine. There is no public study of this metric for Poland, so this is our cautious assumption.
Used only to convert the revenue gap into a profit gap.
How many of the people who enquire become customers. Used only to convert lost customers into the number of lost enquiries.
What has changed in Google click-through rates
In 2024 the result in first position on Google got 28% of clicks. A year later, after AI-generated answers were introduced at the top of the results page, the same position gets 19%, and second position has fallen from 20.83% to 12.60%. This is a GrowthSRC measurement on about 74 thousand TOP10 keywords, selected from a set of more than 200 thousand, using Search Console data from more than 30 websites in e-commerce, SaaS, B2B and EdTech, not a forecast.
A panel study by Pew Research Center shows the other side of this change. When an AI Overview appeared on the results page, 8% of users clicked a traditional link. When there was none, 15% clicked. That is a drop of 46.7%. 1% of visits ended with a click on a source cited by the AI answer itself.
This means one practical thing: first position in Google is no longer the same commodity it was two years ago. If someone sells it to you as a guarantee of traffic, they are selling you the situation before the change.
Why SEO alone is not enough today
Between January and April 2026, 68% of Google searches ended without a single click on any result. Two years earlier the figure was 60.5%. At the same time, a channel that Google does not serve at all is growing: Gemius reports 9.3 million real ChatGPT users in Poland in June 2025, almost a third of Polish internet users.
Both trends push in the same direction. A growing share of queries ends with an answer in which you are either mentioned or absent. The calculator splits this loss into two layers so you can see which one affects you: visibility in AI answers is a different problem from your position in the classic results and needs different work.
How we calculate: the full method
Below is everything you need to reproduce the calculator's result by hand in a spreadsheet. Each coefficient has one of four labels.
Step 0. Revenue from the online channel
We calculate everything only from the revenue that can realistically come from the internet, never from the company's total turnover. Without this limit the model produces absurd results for companies that sell mainly through sales reps.
P_online = monthly turnover × internet share, where in “I know my customer value” mode the turnover is the average customer value multiplied by the number of customers per month.
Step 1. Position gap in Google
We do not guess the number of searches in your industry, because we do not know it. Instead we work with a click-through ratio: how many clicks your current position gets and how many the target position would get.
| Visibility state | CTR today | Type | Google layer limit |
|---|---|---|---|
| First screen of results | 11.0% | data | 20% P_online |
| You have to scroll down | 5.0% | interpolation | 30% P_online |
| Rarely or never | 0.5% | assumption | 35% P_online |
| I do not know, I have never checked (default choice) | 5.0% | assumption | 30% P_online |
| Target: stable TOP3 after AI Overviews | 13.0% | derived | not applicable |
The “I do not know, I have never checked” state is calculated exactly like “you have to scroll down”. Not knowing your own position is no proof that you have no position, so we do not plug in the worst case here, even though this is the default choice in the calculator.
We deliberately do not use 28.5% as the target value, even though that is exactly the click-through rate for first position in the SISTRIX study for mobile results. 28.5% is the situation before AI Overviews. We use 13.0%, derived from GrowthSRC measurements after their introduction: in 2025 they measured 19% for first position and 12.60% for second, and we assume a stable TOP3, not a permanent number one, so we put the target value just above second position. This figure does not appear directly in any study. This choice understates the promised gap, and that is deliberate.
Google gap = P_online × (target CTR ÷ CTR today - 1) × reach multiplier, with a negative result clipped to zero.
| Reach | Multiplier | Type |
|---|---|---|
| One city and the surrounding area | 0.85 | assumption |
| Several cities or a region | 1.00 | assumption |
| All of Poland | 1.15 | assumption |
| Poland and abroad | 1.25 | assumption |
With local reach the pool of queries is smaller and the effect of a better position saturates faster. With nationwide reach it is the other way round. The multipliers are an editorial calibration, not a research result.
Step 2. AI answers intercept the click
AIO loss = P_online × share of searches with an AI answer × 0.467 × non-citation coefficient.
The value 0.467 is the drop in click-through rate measured by Pew Research Center (8% versus 15%). We chose Pew rather than Seer Interactive (a 62% drop) or Ahrefs (58%) because it is a primary study on real panel traffic and gives the lowest, and therefore safest, value. The non-citation coefficient is 1.0 when a company has checked and is not mentioned by AI models, and 0.35 when it has checked and is. The third state, “I have not checked”, gets no amount: without checking we do not know whether AI models mention the company, so we do not put a value on the layers from steps 2 and 3 and show them as “we do not know”. This is the first thing we check in the free audit. The same coefficient also applies to the layer from step 3.
Step 3. Queries that never reach Google
Assistant loss = P_online × share of queries that bypass Google × non-citation coefficient.
This is the weakest point of the whole model, and we say so openly. There is no public, citable study of the share of Polish purchase queries that start in an AI assistant instead of Google. The default 3% is a cautious fraction of the ChatGPT user population in Poland reported by Gemius. It is an editorial assumption, which is why it is a slider you can change.
Step 4. Aggregation, limits and scenarios
The Google layer has its own limit, which depends on the visibility state: from 20% of online revenue for companies that are already visible to 35% for invisible companies. This answers a simple arithmetic problem. A company with a 0.5% click-through rate that wants to reach 13% has a ratio of 26. A model without a limit would claim that its online revenue will grow twenty-six times, which is arithmetically correct and makes no business sense.
The limit deliberately covers only the Google layer. If we clipped the whole total proportionally, both AI layers would drop to a few dozen PLN and the breakdown of the result would lie about where the problem really is. A separate, absolute ceiling of 55% of online channel revenue applies to the sum of the three layers. This protects against double counting, because the drop in click-through rate after AI Overviews is already partly built into the low target click-through rate.
We do not show the result as a single number. The cautious scenario is 0.6 of the base result, the likely one 1.0 and the upper one 1.4. We always show the cautious scenario as the main number. In addition, the annual gap cannot exceed 25% of annual turnover, and with turnover below 20,000 PLN a month we round the result down to the nearest full 100 PLN.
What this model does not do
The model does not check your website, does not know your positions in Google, does not know the search volume in your industry and does not measure whether you are cited by ChatGPT or Gemini. We check all of these only in the free SEO and GEO audit, which we prepare in 5 working days.
The second thing to say openly: all the key studies this model is built on concern the US or global market. Polish specifics shift the result both ways. Google has an 88.9% share of the search engine market in Poland, more than in the US, which increases the weight of the first layer. Adoption of generative AI, on the other hand, is lower than in English-speaking countries, which reduces the weight of the third layer.
What exactly we check in the free audit
- Your real positions. Not what you declared in field 3 of the calculator, but a measurement for the phrases your customers actually use to look for you, along with the number of searches per month.
- Whether AI models know you. We ask ChatGPT, Gemini and Perplexity the questions your customer would ask and check who comes up in the answer instead of you.
- What is blocking indexing. Pages Google does not see, missing structured data and technical errors, in order from the most costly.
- A comparison with two or three competitors. Without it, the number of seconds your page takes to load means nothing.
We explain what the price of an audit depends on in the article How much does an SEO audit cost (in Polish). If you would rather compare service packages straight away, they are on the pricing page.
Sources
- SISTRIX, click-through rates of search results across more than 80 million keywords, for mobile results: sistrix.com
- GrowthSRC, the drop in organic click-through rate after the introduction of AI Overviews, measured on about 74 thousand TOP10 keywords, selected from a set of more than 200 thousand, using Search Console data from more than 30 websites: growthsrc.com
- Pew Research Center, March 2025, a panel of 900 people sharing their browsing history: pewresearch.org
- SparkToro using Similarweb data, the share of searches without a click: sparktoro.com
- Gemius, Mediapanel, ChatGPT use in Poland: gemius.com
- StatCounter, search engine market share in Poland: gs.statcounter.com
Questions about the calculator
No. The result appears as soon as you fill in the fields, and you can copy or print it without giving any contact details. The form at the bottom of the page is optional and is for ordering a free audit.
From four public studies: SISTRIX (click-through rates of results across more than 80 million keywords, for mobile results), GrowthSRC (the drop in click-through rate after the introduction of AI Overviews, measured on about 74 thousand TOP10 keywords selected from a set of more than 200 thousand, using Search Console data from more than 30 websites), Pew Research Center (8% of clicks when an AI Overview is present versus 15% without one) and Gemius (9.3 million ChatGPT users in Poland in June 2025). Coefficients not backed by research are labelled as assumptions.
No. It is an illustrative estimate, not an offer within the meaning of Article 66 of the Polish Civil Code, not financial advice and not a guarantee of positions or of presence in AI answers. The model does not check your website and does not know your positions in Google. We check those things only in the free audit.
Because most Polish SMEs do not have analytics set up correctly, and those that do do not know their number of organic sessions from memory. The calculator only asks about things a business owner knows off the top of their head.
For two reasons. The Google layer has a limit of 20% to 35% of online channel revenue, depending on the visibility state, and the sum of all layers does not exceed 55%. On top of that, we show the cautious scenario as the main number, which is 60% of the base result. The range of the other scenarios is shown below the amount.
All the calculation happens in your browser. The calculator does not send the numbers you enter to our server, does not store them in cookies and does not put them in the page address. But let us also say the second part openly: this page, like every Neurise page, has Google Analytics and Google Tag Manager installed, which, with your consent, record the visit together with the page address. That is why the address deliberately stays clean and Google does not receive your turnover with it. The only moment your numbers go further is one you choose yourself: the “Copy link with result” button puts them in the copied link, and the form at the bottom of the page sends them to us if you leave the consent to attach the result ticked.
Not in this model, and we say so openly. The best state the calculator describes is an 11% click-through rate on the first screen versus 13% achievable, so the Google layer always stays positive. In addition, if you have checked your presence in AI, two AI layers are added, which cannot be brought to zero by your position in the results alone. The lowest amounts come either from a small sales scale, from a low share of the internet in winning customers, or from a situation in which you are already visible in Google and mentioned by AI assistants. In that last case the right response is maintaining your position and monitoring citations, not SEO from scratch, because model citations change from week to week.
Related pages
Get a free AI audit
Leave your details and your website address, and we will check how it performs in Google and whether it appears in AI answers. You get the result with specific recommendations.