What’s new
What’s new in ECHO
Updated
Our change log: what customers are already using. Honest also when we correct ourselves.
This is what has been added to ECHO, each entry dated with the day it became available to everyone. What we are still building is not in this list: it shows only what you can use today. What we plan next is set apart at the end of the page.
The change log entries are written in German first, in the same step as the change itself. Every public entry also has its own English version. An entry without one does not appear here.
Gap sheet: page against pageActions
For every competitor the AI uses as a source for your questions, ECHO puts the page it used next to yours. Per question and AI service you see how often this happens, which terms appear only there and what the AI says about the competitor. It ends with a ready page brief. From the Pro plan. Competitors that ECHO classified but has not yet reviewed are confirmed by you first, in the settings.
Monthly briefing: every decision with its own target figureMonthly briefing
Every decision in the monthly briefing carries the figure it is meant to move. In the review you see this figure then and now and whether the change is reliable.
Google AI Mode: your questions on Google’s answer page tooMeasurement
Alongside the AI box above the Google results, on request we now also measure Google’s AI Mode, Google’s own answer page. For each question you see whether the answer names you and whether it uses your website as a source. The box and AI Mode share one «Google» column on the answers page. Like the box, AI Mode does not count towards your headline figure.
Copilot: what Microsoft reports about your websiteMeasurement
In Bing Webmaster Tools, Microsoft shows how often Copilot has named your pages as a source. You can now upload this report to ECHO. Under Analysis › Sources you then see how often Copilot used your website, with which pages and what Copilot searched for. The figures come from Microsoft and show no competitors. From the Pro plan.
Answers: filter by buying stage and region, companies outside your listCockpit
In Answers you can now filter by buying stage and by region. All figures on the page then refer to these questions only, for example to the decision stage. The detail view of a question now also shows which company websites the AI uses there as a source in at least two answers although they are not on your list of competitors. Editors and admins can put such a company straight on the list.
Answers: a competitor’s view and export as a table (from Pro)Cockpit
From the Pro plan, the Answers page can show the view of a competitor on your list: for which questions they are named and you are not. The overview can also be downloaded as a table, with the same figures as in the cockpit. That way you can pass it on to your agency or your management team.
Tech check on every plan, fact watch and monthly briefing from Pro upTechnical check
The tech check is now part of the Start plan too, so Start also gets the technical tasks under Actions. The Pro plan now includes the fact watch and the monthly briefing for the management team, the briefing with a print view and in beta for now. The price stays the same.
Answers: history and path for each questionCockpit
For each question you now see the history for each AI service, counted in four-week periods. A triangle marks changes at a service or in our measurement, so you can tell whether a jump comes from there. The new filter «Changed» shows only questions where something has moved clearly beyond the usual fluctuation compared with the previous four weeks. Each question now also shows the path to your mention: whether the AI found your website, used it as a source and named you.
An email when something clearly movesMeasurement
After every scheduled measurement you get your figures by email. If something has clearly moved, it now comes first: how often you are named in supplier questions, questions with clearly different answers and the share of you and your competitors. Clearly means larger than the usual fluctuation of your measurement. Without any movement, the email stays as you know it.
Answers: all questions at a glanceCockpit
The answers page now shows all your questions in one overview. For each AI service you see how often it names you or uses your website as a source over the last four weeks. It also shows who the AI names instead of you. One click opens the question with the answers word for word, their sources and the searches the AI ran. Throughout the cockpit it now says «used» instead of «cited». Only what the AI actually uses to back up its answer is counted.
A source now means what the AI uses in its answerMeasurement
«How often you serve as a source» now only counts pages that ChatGPT, Claude and Gemini use to back up their answer. Until now, the figure also counted search hits that Claude had seen but not used, and Perplexity's whole source list. Perplexity therefore has its own line «in the source list», because the service does not show which sources it uses in the text. The «Most used sources» card, the answers and the monthly briefing count the same way. For some projects the figure drops, for others it rises, also for past weeks. The measurement is unchanged, the counting is more precise.
The overview tells you first what stands outCockpit
At the top of the overview, one sentence tells you where you stand, followed by up to three findings from your measurement: for example a wrong statement about you, a shift in how often you are mentioned, or a competitor that gets mentioned more often. Each finding names the figure it is based on and leads to the matching action where there is one. Anything based on only a few answers is marked as preliminary.
Your figures split by region and buying stageCockpit
You now see mentions, your share of all mentions and how often you serve as a source separately for questions that name a place, such as «in Lucerne», and for nationwide questions, and by buying stage: learning, comparing, deciding. That way a home advantage in your region does not look like strength across the whole market, and you see in which stage you are a source and where you are mentioned. How each question is classified is shown in the question list, where you can change it.
Where you stand and how far there is to goCockpit
A single figure says little. The card «Compared with other companies» now sets you against companies from your sector, whether professional services or industry, trade and utilities, once enough companies there are measured, also beyond Switzerland. It now shows not only whether you are ahead of, within or behind the midfield, but also how many percentage points you are short of the top quarter. It leads straight to the actions that address this. Today the AI names the typical company in only about 2 out of 10 answers. We never show individual companies. Because the calculation is now by sector, the midpoint can shift once without anything having changed on your side.
One criterion fewer, but backed by evidenceActions
We have dropped list format as a criterion. The study our page check is based on finds no effect for how a page is structured. The fourth factor is the position of your page in the AI’s search results, and you cannot set it directly. Open tasks to «show it as a list or comparison» have therefore disappeared from your list. Topic fit, visible prices and a current date remain, and we keep checking them.
When the AI says something wrong about youActions
You give us your hard facts, for example founding year, registered office and what you explicitly do not offer. We compare every measured answer with them and show you the sentence word for word when ChatGPT, Claude, Gemini or Perplexity claim something different. You can tick off each point or clear it as «actually correct», and then we will not ask you about it again.
Google’s AI box is checked for errors tooActions
We now also compare the facts you have stored with the AI box that Google places above its results. If something wrong appears there, you see the sentence word for word, marked as coming from the Google box and not from a chat. That matters, because otherwise you would look for the correction in the wrong place.
Where you stand compared with other companiesCockpit
A single figure says little. In the cockpit you now see how often you are named and serve as a source next to the other companies we measure with the same method: the midpoint and the range in which most of them lie. We never show individual companies. The card only appears once there are enough companies for a fair comparison.
Measuring in French, Italian and EnglishMeasurement
You can now also have your questions asked in French, Italian and English. Each language has its own questions and its own figures, and they are never mixed: in the cockpit you switch between languages. You have existing questions translated with one click, check the suggestion and approve it. Only then does the measurement start.
The AI box above the Google resultsMeasurement
When someone googles something, there is now often an answer written by Google above the results. For your questions we now also measure whether this box appears at all, whether it names you and whether it gives your website as a source. It gets its own card in the cockpit and is not mixed with the other figures: where Google shows no box, there is nothing to win there either.
What decides whether the AI uses your page as a sourceActions
Your actions are now tied to what really decides whether an AI uses your page as a source: visible prices, a current date and the answer as a list or comparison. We check your most important pages for these and show you what is missing on each one. We have dropped the old recommendation on llms.txt because no effect can be shown for it.
What the AI searches for before it answersMeasurement
Before an AI answers, it runs its own web searches. We now show you these searches, per question and as a list of the most frequent topics. Each of them is an idea for a page on your site. In the cockpit you also see how often you appeared in the AI’s search results and how often it used you as a source from there. These are two different things, and they need different actions.
Is the AI allowed to read your website at all?Technical check
The technical check now tells you for each AI service, in plain words instead of crawler names, whether it may read your website at all. It also notices when it is not your robots.txt that blocks, but the firewall in front of it. If something is missing, you get exactly the lines to add instead of a block to replace. You also see, with dates, what has changed since the last check.
Same measurement, more honest figuresMeasurement
From today we show your GEO figures more precisely. The measurement itself has not changed, only how honestly we put it in context. Until now, three repetitions of the same question counted as three independent measurements. That made the uncertainty range (±) artificially narrow. They now count as one, which is why the ± bands are wider. The figure for how often you serve as a source is stricter too: it counts per question and measurement day instead of «at least once in recent weeks», which is why it looks lower. Same measurement, more honest figures, there is nothing you need to do.
Every figure says what it refers toCockpit
Position and share of mentions visibly apply within the competitors you have entered, not against the whole market. And if an AI service did not answer during the period, the tile says so instead of the gap quietly disappearing into the figure.
Monthly status even before the first comparisonCockpit
The monthly status now shows where you stand even when there are not yet four weeks of earlier data: all figures of the current measurement window and the date from which we can compare. The monthly status and the cockpit now both say which period they cover. That is why the figures can differ without anything having changed.
Project list and wording brought into lineCockpit
The project list shows the same figure as the cockpit, with the period below it and the role in plain words. Technical terms in the cockpit now have plain names. The note on the change of 19 August only appears for projects that were already measured before that date.
One sentence at the top: where you standCockpit
At the top of the cockpit, one sentence now tells you where you currently stand and what that means. The tiles and charts for it follow below.
The same language and number format everywhereCockpit
The last technical terms now have their explanation. Date and number formats look the same throughout the cockpit.
The cockpit on your phoneCockpit
Tiles, tabs and tables are readable on your phone without having to scroll sideways.
Demo answers no longer count as measurementsMeasurement
Test measurements in demo mode no longer flow into your figures. These answers are simulated, not measured. As a result, values can shift retroactively for projects that were measured for a while without access to an AI service.
Measurement day on Swiss time, your web address counts as a mentionMeasurement
A measurement day now runs from midnight to midnight Swiss time. And if an AI answer only names your website, for example your-brand.ch, that counts as a mention from today: the answer sends people to you, and that is visibility even without your company name next to it. Both apply from today, older measurements stay as they were.
Up to mid-August 2026
4 entriesQuestions from your Search ConsoleMeasurement
We connect your Google Search Console, and ECHO suggests questions from it that your customers really type. If you do not want the connection, you upload the export file.
Every figure shows how certain it isMeasurement
Next to every value there is a range (±) showing how precise it is. That way you see at a glance whether a shift is a result or just fluctuation.
Four AI services, with real web searchMeasurement
ChatGPT, Claude, Gemini and Perplexity answer your questions with real web search. Each answer shows whether a search really took place or whether the service answered from memory.
Perplexity provides better source detailsMeasurement
Since then, Perplexity gives us sources with title and address instead of a bare list. This makes your source ranking more precise.
The note about a change is written in the same step as the change. That is why this list is complete, and why no date on it was tidied up afterwards.
Coming next in ECHO
What we are planning
Our plan, not a promise: order and scope can change, and what customers tell us shifts it. A point only moves to the list above once you can use it.
Planned
next blockYour pages on Google and in AI
For each page and topic you see side by side how often people search for you and find you on Google, and whether the AI names you and uses you as a source. From that comes the lever most likely to work. The view uses the same filters as region and buying stage. With Google Search Console connected it goes deeper, without it the view stays coarser.
What happened after a completed action
For every completed action we show for the page concerned, before and after, how clicks from Google developed and how often AI answers use it as a source. That shows the trend, not proof that the action caused it. Clicks are only available with Search Console connected.
Which review sites the AI reads for your questionsrequested by customers
For many questions the AI relies on review sites such as Trustpilot or industry portals. We will show which of them it actually uses as a source for your questions and whether you are named in those answers. We only recommend a listing where the measurement shows that the site matters for you.
Ideas
under considerationWhat the AI searches for in your sector
Before the AI answers, it runs its own web searches. We will combine these searches across all companies we measure in your sector and show you the topics searched for most often there. No individual company can be identified in it.
A trial run in another language
Whether you are named at all in another language, such as French or German, should be clear before you measure it permanently: with a one-off trial run of your most important questions, with the fluctuation shown alongside.
The same questions, asked from different places
Answers also depend on where someone asks from. We are considering asking the same questions from different places, for example from Lucerne and from Zurich, and showing whether that changes how often you are named.
Your figures in your own dashboards
Your ECHO figures should end up where your team already looks, for example in Data Studio, instead of in yet another tool.
ECHO Academy
Short explanations of how visibility in AI answers comes about and what you can influence, without jargon.
What are you missing? Tell us.
If you already use ECHO, send your request right in the cockpit, under «Neu in ECHO». Otherwise, just email us. We read every message, and whatever comes of it is marked «requested by customers» here.
PriceECHO starts at CHF 140 a month excl. VAT, on the Start plan with 40 questions for one website. The first measurement, including a call to walk you through it, is free. All plans →
Find out whether you are in the answer
We build twenty questions your customers really ask, run them once through every service, and explain the result to you in a call. Free, even if nothing comes of it. If AI answers in your field hardly ever name a supplier, we tell you that instead of selling you a subscription.
Request a free first measurement → ← Back to site