What are answer engines?

The tools that give an answer instead of a list of links: ChatGPT, Perplexity, Gemini, Claude, and the answers Google itself now shows above its results. The guest does not search for "hotels Naxos", they ask "where should I stay in Naxos with two small children for five nights in June".

Why should I care what they answer?

Because the answer is given whether it mentions you or not. If your property never appears, you lose guests who will never learn you exist. And if you appear with wrong details, the mistake travels without you knowing.

How do I check this myself?

Open two or three answer engines, type five to ten questions of the kind a prospective guest would ask, and record what comes back. Important: the questions must not name your property, otherwise you are measuring whether the engine knows you after you have just told it who you are.

How often should it be checked?

It is not a question of frequency, it is a question of continuity. Engine answers shift from day to day even when nothing on your site has changed, so a check twice a year does not produce evidence, it produces an impression. To know whether you are actually rising or falling, the measurement has to run continuously and build up history. Doing it once yourself is worth it, to see whether there is a problem at all. Beyond that it needs a system.

Does each engine give the same answer every time?

No, and this is the most misunderstood point. The same question can get a different answer on two consecutive tries. That is why a single "it did not mention me" means nothing on its own, while five out of five "it did not mention me" means something.

What do I do if the information about me is wrong?

You fix the source, not the answer. These engines have no correction button. Check what your own site says, your maps profile, and the booking platforms, and make sure they all say the same thing. The wrong information usually comes from somewhere you control and have forgotten about.

Do I need a special tool?

Not to start. A file with your questions and one column per date does the job. You need a tool when you want to track many questions across many engines systematically, rather than by hand.

Can you do this for me?

Yes, this is what we do. We write the questions so they are descriptive and comparable with each other, we ask them repeatedly across four engines over weeks, and we give you a trend with a range rather than a single number. And, most importantly, we also take on the work that changes the result, because measurement alone has never improved anything. Send us your address and your area for a first picture at no cost.

Continuous visibility monitoring is part of the MS-Tools subscription, alongside availability and the technical health of the site. The improvement, meaning the work that gets your property to appear, is priced as a GEO project. The first picture, so that we both see where you stand, is not charged.

Is this the same as SEO?

Related but not the same. SEO is about your position in a list of results. Here there is no list, there is an answer that either includes you or does not. Much of what helps one helps the other, but the measurement is completely different.

What answer engines say about your property: a self check guide

A couple in Munich is planning a holiday. Five years ago they would have opened Google and typed “hotels Paros”. Today, more and more often, they type something like this:

“We want five nights on a Greek island in September, quiet, with a good taverna nearby, no car. What do you suggest?”

And they get an answer. Not a list, an answer, with three or four names inside it.

The question for you is simple and uncomfortable: are you inside that answer?

The good news is that you can get a first picture today, on your own, without buying anything. Here is how.

It is the same method we use ourselves, on our own sites and on clients’. At the end I will tell you what changes when it is done systematically, because the difference between a snapshot and a measurement is large.

Step one: write the questions as a guest, not as an owner

This is where half the value of the exercise is lost, and it is usually lost on the first line.

The natural instinct is to type your property’s name and see what comes back. That measures something entirely different: it measures whether the engine knows anything about you after you have told it who you are. The prospective guest does not know your name. That is the whole reason you want them.

Useful questions are descriptive: they describe the guest and the need, not you.

Good examples, adapt them to your own area:

“Where should I stay in [area] with two small children, near a shallow beach?”
“Quiet accommodation in [area] for a couple, no car, in October.”
“Which are the best areas to stay on [island] if you do not want nightlife?”
“Accommodation in [area] that accepts dogs.”
“Where do Greeks stay when they go to [area]?”
“Is [area] a good choice in May? What should I watch out for?”

And one that is particularly worth it, because it tests accuracy rather than presence:

“What should I know about prices and fees when staying at accommodation in Greece?”

Write five to ten. Fewer gives no picture, more and you will not repeat them.

Step two: ask in more than one

They do not all give the same answer, and they do not all draw on the same places. Ask in at least two or three: ChatGPT, Perplexity, Gemini, Claude, and also look at what Google itself shows at the top of its own results.

Two practical details that change the outcome:

Ask in the guest’s language too. If your big market is Germany, the question in German is the question that counts. Answers differ by language, often considerably.

Do not be logged into your account. If the engine remembers you, the answer is shaped by what you have asked before. Open a private browsing window.

Step three: record it, do not be impressed by it

The third step is the only boring one, and it is what makes the exercise useful rather than entertaining.

Make a simple file. One row per question, one column per engine, and today’s date. In each cell note three things: whether it mentioned you, in what order within the answer, and whether what it said is correct.

The third is the one that gets forgotten and the most dangerous. A mention with wrong details is not half a win. A property presented as “close to the port” while it is twelve kilometres away will bring guests who arrive annoyed.

The most misunderstood point: one measurement says nothing

Answer engines are not deterministic. The same question, in the same engine, five minutes apart, can give a different answer.

That has a consequence you need to accept before you start: one “it did not mention me” means absolutely nothing. Nor does one “it mentioned me first” mean you have won.

What does mean something is repetition. Five tries of the same question, and you count how many times you appeared. Three out of five is information. Zero out of five is information. One out of one is not.

And this is where the real problem starts, which is worth knowing before you settle in to work. Five tries tell you where you are today. They do not tell you whether you are rising.

To know whether you are rising, the same measurement has to run continuously, with exactly the same questions, for weeks and months, so that history accumulates and the trend separates from the noise. Otherwise, when you come out three in five instead of two in two months, you will not know whether you improved or the engine simply had a good day.

It is the same discipline any noisy measurement requires: look at the trend, not the snapshot. And state the range instead of pretending you have a precise number.

Let us be honest about what that means in practice: no property owner is going to open four engines, ask ten questions and log them in a file, every week, for months. Nor should they. That is work for a system, not for a person with a notepad and patience.

What to do with what you find

Depending on the result, three different jobs.

You appear nowhere. The most common outcome, and less alarming than it sounds. It usually means there is not enough written, public, specific content about you. These engines answer from text. If your site says “beautiful rooms in a quiet setting” and nothing else, there is nothing to retrieve. If it says how many minutes on foot the beach is, whether there is shade in the car park, what time the first ferry leaves, and what there is to do there in October, then there is.

You appear with wrong details. You fix the source, not the answer. None of these engines has a correction button. Check what your site says, your maps profile, the booking platforms, and local directories. In practice the error almost always comes from somewhere you control and have forgotten: an old page, a profile with last year’s address, a directory you registered with in 2019.

The area appears but you do not. The most interesting result. It means the engine knows the area and recommends accommodation, just not yours. Look at which ones it recommends and what they have written that you do not. It is usually not a secret, just more and more specific information, written in a way that can be retrieved.

And a warning about what this is not

It is not a ranking. There is no “first place” to win, no official score, and nobody to complain to.

It is a visibility measurement with noise in it. Its value is not in the number, it is in showing you what is missing from your writing.

How we do it, and why it differs

You can do everything above yourself, and it is worth doing at least once. You will get a snapshot.

We turn it into a measurement, and the difference is in four places.

The questions are written by a human, and that is half the work. You give us two to four ideas, we turn them into descriptive questions that do not name your property, we check they do not overlap, and we lock them so they are asked identically every time. A question whose wording drifts cannot be compared with itself.

We ask continuously, across four engines, and we keep the history. Not once, and not twice a year. This is the point where the exercise becomes a measurement: with enough repetitions of the same question, “three out of five” stops being an impression and becomes a number that compares with itself from last month. Without it, no change you make to the site can be shown to have paid off.

We give you a range, not a single number. It is the same rule we apply to every noisy measurement we run: anyone who shows you a visibility score as if it were the truth either has not measured it or is not telling you.

And then we do the work that changes the answer. Measurement on its own has never improved a single result. The missing content, the structured data, the pages that answer real questions: those move the needle, and they are work.

Continuous visibility monitoring is part of MS-Tools, alongside availability and the technical health of the site. The improvement itself, meaning getting your property to appear where it does not appear today, is the SEO and GEO service and is priced as a project.

Want to see where you stand today?

Send us your property’s address and its area. We will run a first set of questions and tell you what the engines answer about you and about your neighbours, at no cost and with no commitment.

If you like the result, you need nothing from us today. If you do not, you will know exactly what has to change, and we can take it on: both the continuous measurement and the work that moves the result.

Ask what the answer engines say about your area