Making visitors always feel in the right place: a guide to orientation

If a visitor writes "the audio guide doesn't work", one time in four they are talking about neither the content nor the technology. They are saying something much more specific and much easier to fix: I'm not sure that what I'm listening to is what I'm looking at.

It's the difference between a content problem (expensive to solve, requires time and curation) and an orientation problem (often fixable in an afternoon, at zero cost). This guide helps you tell them apart and close the second one.

Most of the interventions can be done from the amuseapp console in a few hours: they require no budget, no re-installation of the exhibition and no rewriting of content.

1. What the data say

Across more than 49,000 ratings collected through end-of-visit questionnaires in 258 venues that use amuseapp (October 2026 data), the average score is 4.34 out of 5. Overall satisfaction is high. The problem, when there is one, is not spread evenly: it is concentrated, and it is almost always the same one.

Among the items visitors name as "what I liked least", looking at those that depend on the venue and the route:

• Ticket price: 1,164 mentions
• Signage and directions: 753 mentions
• Length of itineraries: 667 mentions
• Exhibits: 367 mentions
• Accessibility: 341 mentions

"Signage and directions" is the top item after price: among the things the cultural venue controls directly, it is the most frequent complaint.

And it weighs on the final score in a measurable way:

• All visitors: 49,212 responses, average score 4.34
• Those who cite "Signage and directions": 753 responses, average score 3.55 (-0.79)
• Those who cite signage and also the audio guide: 134 responses, average score 2.94 (-1.40)

The last figure is the most interesting. When signage is missing, the blame falls on the audio guide: those who complain about both things together give the lowest score of all. They are the same complaint seen from two sides. An audio guide without physical reference points and signage that doesn't match the audio are the same problem, and it is solved from the same side: from the space.

In the free-text comments the picture is less uniform, and more useful. We read and classified one by one 1,111 comments to the question "what can we improve?", which only appears for those who give a score from 1 to 3. Orientation in permanent routes is the top theme (27%). And among the comments that mention the audio guide, one in four is actually about orientation: the person writes "the audio guide doesn't work" and then explains that they couldn't find the artwork the voice was talking about.

Some real visitor comments (translated and anonymized):

"The audio guide jumps too much from one painting to another without a real guiding thread, I often struggled to follow it and find the artwork it was talking about."

"Why didn't you put a number on the artwork? At least it would have helped."

"You get lost in the room looking for the artworks being discussed."

"I didn't know which room I was in for half the time. You should clearly mark the number of each stop, especially the first ones."

"The staff didn't tell me where the route starts, so I wandered for 30 minutes looking for the starting point. Once I found it, it was easy to follow, but it would be useful to have directions between one stop and the next."

"More orientation in the space was needed, maybe arrows on the floor. The audio guide was a bit confusing about where to go."

"The audio guide needs to say better where to look and how to recognize each painting."

"Link the audio guide information to the numbering of the rooms and paintings. Without it, we had to feel our way around."

"The audio doesn't make clear which room I should be in, so I wasn't in sync with the audio. You need signs with a number that matches the stop number."

None of these visitors is asking for better content. They are asking for a key linking the physical space and the digital content.

The case where the problem is most acute: temporary exhibitions

Temporary exhibitions don't on average have more orientation complaints than permanent routes, but they are the context where the problem most easily explodes: in one analyzed exhibition, orientation accounts for 76% of negative comments (16 out of 21), in another almost half.

The reason is structural:

• the exhibition layout is new and has never been walked by a real visitor before opening;
• the audio guide is written on the catalog, that is, on a thematic and chronological order, while the artworks are hung according to an exhibition order that responds to constraints of space, light and loans;
• loans change until the last minute, and sometimes even after opening;
• there is no room signage consolidated over the years to lean on.

If you are about to open an exhibition, section 7 of this guide is the one to read first.

2. Why it happens: an implicit pact that broke

The old rental audio guides had an ugly interface and a brilliant idea: the number.

The number on the wall was the interface. The visitor didn't need to understand anything: they saw "24" next to the painting, typed 24, and listened. If they heard about a different painting, they knew immediately they had typed the wrong number, not that they were lost. The number was a verifiable contract between the space and the content.

When you move to the smartphone, the interface becomes nicer, richer, multilingual, with images and in-depth material. But if nobody puts up the numbers, the contract disappears: the visitor scrolls a list of titles ("The Triumph of Virtue", "Audience Hall") and has to guess which physical object they correspond to. In a room with forty paintings, guessing doesn't work.

The key, counterintuitive point: the problem wasn't born from the app, it was born from the removal of the number. Nobody removed it on purpose, it simply dropped out of the project because the digital audio guide seemed not to need it. It needs it exactly as before, and amuseapp has a dedicated field to manage it (see 4.1).

The visitor's three questions

At every moment of the visit the visitor has three questions in mind. They rarely voice them, but if one goes unanswered they stop, look around, get distracted, and the blame falls on the audio guide.

1. "Where am I?": in which room, in which section, at what point in the route.
2. "Is this it?": is the object in front of me the one the voice is talking about.
3. "Where do I go now?": what is the next point and how do I get there.

A well-designed route answers all three at every single stop. It isn't a matter of the amount of content: it's a matter of where you put the first and last ten words of each audio track.

A useful way to check: try telling the visit as the film the visitor lives through, not as a list of artworks. If in the story there are moments when the protagonist looks around and doesn't know where to go, you have found the points to fix.

3. Overview of interventions

From the cheapest to the most structural. The first ten rows cost nothing and are done from the console.

1. Number the POIs and align them with the signage. Where: Console. Time: 1-2 h. Cost: zero. Impact: high.
2. Confirmation opening on every audio. Where: Console. Time: 2-3 h. Cost: zero. Impact: high.
3. Closing with a transition instruction. Where: Console. Time: 2-3 h. Cost: zero. Impact: high.
4. One POI = one artwork. Where: Console. Time: variable. Cost: zero. Impact: high.
5. Itinerary order = actual walking order. Where: Console. Time: 30 min. Cost: zero. Impact: high.
6. Photo of the artwork as it looks in the room. Where: Console. Time: 1-2 h. Cost: zero. Impact: high.
7. Physical description of the artwork in the first seconds. Where: Console. Time: 2 h. Cost: zero. Impact: medium.
8. POI zero: "how this visit works". Where: Console. Time: 30 min. Cost: zero. Impact: high.
9. Service POIs (restrooms, elevator, exit, benches). Where: Console. Time: 30 min. Cost: zero. Impact: medium.
10. Declare returns and detours. Where: Console. Time: 30 min. Cost: zero. Impact: medium.
11. Physical numbers in the room next to the wall labels. Where: Museum. Time: 1-2 weeks. Cost: €. Impact: high.
12. QR codes in the room next to the artworks. Where: Console + printing. Time: 1 week. Cost: €. Impact: high.
13. Numbered floor plan. Where: Museum. Time: 1 week. Cost: €. Impact: medium.
14. 20-second script for ticket office and room. Where: Museum. Time: 1 h. Cost: zero. Impact: high.
15. Align physical wall labels and audio. Where: Museum. Time: variable. Cost: zero. Impact: medium.
16. BLE beacons with autoplay. Where: amuseapp + museum. Time: 4 weeks. Cost: €€. Impact: high.
17. GPS for outdoor areas. Where: Console. Time: 1 day. Cost: zero. Impact: high.

4. Level 0: what you do from the console, today, at zero cost

If you are short on time, these ten interventions are 80% of the result.

4.1 Number the POIs, and make the number match what the visitor sees

In the list of stops the visitor sees the position in the route ("Stop 3") and, inside the stop, the counter "3 of 12": this is the numbering they read, and it must be the same one they find in the room. The POI also has a code field, separate from the title, but it is used for per-artwork QR codes (5.12) and does not appear in the list. If the museum already uses a different numbering (for example inventory numbers on the wall labels), put it in the POI title as well.

Rules:

• Continuous, increasing numbering along the real route. No gaps, no numbers "reserved for the future", no alphanumeric codes like "S2-A14": the visitor must be able to hold it in memory for ten seconds.
• A single numbering system across the whole venue. If a room numbering already exists, used by signage or by the paper floor plan, adopt it instead of creating a new one. Two parallel numberings are worse than no numbering.
• Low numbers. If the route has more than 40 stops, consider splitting it into itineraries by floor or by section: "12" is memorable, "137" is not.
• The number also goes in the POI title in the list, so that the visitor scrolling the list sees it without opening the stop.

If you have no physical numbers on the walls, the in-app number is still useful: it becomes the reference for the floor plan (4.13) and for staff (4.14). But the leap in quality comes when the number is also on the wall (5.11).

4.2 Open every audio with a position confirmation

This is the single change with the best ratio of effort to result in the whole guide.

The first words of every track should not be the title of the artwork. They should let the visitor verify in two seconds that they are in the right place, using what they see.

Before:

"The Triumph of Virtue by Paolo Veronese, oil on canvas, 1567."

A visitor in a room with twelve canvases has to find which one. They search instead of listening, and the first twenty seconds of content are lost.

After:

"You are in the Audience Hall, stop 12. In front of you, on the wall opposite the entrance, the largest canvas in the room: a female figure dressed in red trampling a chain. It is the Triumph of Virtue by Paolo Veronese."

Now the visitor has three independent verification keys: the room, the wall, the visual attribute. If one doesn't match, they notice immediately.

The three-element formula:

1. Where you are (room and stop number)
2. How to recognize the object (position in the space + an attribute visible to the naked eye)
3. Only at the end, what it is (title, author, date)

Apply it to all stops. It costs one line at the top of each extended description, and amuseapp translates it automatically into all enabled languages.

4.3 Close every audio with the instruction for the next stop

The other half of the work. The moment when the visitor gets lost is not while listening: it is in the three seconds after the audio ends, when the app goes quiet and they have to decide where to go.

Before:

"...and this concludes the story of the room."

After:

"For stop 13, leave through the door under the arch to the left of the fireplace and cross the corridor: you will find a staircase. Stop 13 is at the top of the stairs, the large tapestry on the landing."

Rules for transition instructions:

• One movement per sentence. "Go out, turn right, climb two flights, enter the third door" is a shopping list: the visitor remembers the first part. Break it up: "Leave the room. You will find a staircase on the right. Go up to the first floor."
• Fixed reference points, not relative directions. "Right" depends on which way you are facing. A visitor who went back to look at a detail again has "right" reversed. Use elements that don't move: the fireplace, the stained-glass window, the door with the stone frame, the checkered floor.
• State the distance or the effort. "It's about fifty meters", "there are two flights of stairs", "you cross three rooms without stops": it removes the anxiety of having gone the wrong way.
• Name the number of the next stop. It closes the loop with the physical numbering.
• If there is no next stop, say so. "This is the last stop of the route. The exit is at the end of the corridor, past the bookshop."

4.4 One POI, one artwork

The pattern that generates the most confusion of all is the POI that talks about five different artworks in three minutes. The visitor hears the first described, finds it, then the voice moves on to the second and they have to look for it while listening, then to the third, and by then they have given up.

Visitors say it clearly:

"It would be better to dedicate one track per artwork. This way it's very confusing."

"I get on better when there is a general introduction to the room followed by individual in-depth pieces on each artwork, marked by a number."

The structure that works:

• One room POI (short, 40-60 seconds): where you are, what this room is, what you will find in it, in what order to look at it.
• One POI for each artwork that deserves an in-depth look, each with its own number and its own position check.

It costs more POIs, not more content: the same words distributed differently. And it has a positive side effect: the visitor can choose what to explore in depth, and the Analytics tell you which artworks really work.

If you can't split (for example it's a display case with twenty objects), declare it at the start: "In this display case there are twenty finds. I won't describe them all: I'll talk about the three most important, the first is the one at the top left."

4.5 The itinerary order must be the order in which people walk

It seems obvious, yet it is the most frequent cause of the worst feedback.

The order of POIs in the itinerary is often inherited from how the content was entered (chronological order of the artworks, catalog order, alphabetical order, the order in which the curator wrote the texts). The visitor, instead, walks through the space in the only order the museum physically allows.

Practical check, 30 minutes: print the list of POIs in the current order, walk through the museum with the sheet in hand following the mandatory route, and mark the actual order number. Then reorder in the console with drag & drop.

Redo it every time you:

• change an exhibition layout;
• close a room for maintenance (the mandatory route changes, the itinerary must follow it);
• reverse the direction of the visit to manage flows;
• a temporary exhibition occupies rooms that were previously free.

4.6 Photos: the artwork as the visitor sees it, not as the catalog sees it

The POI photo has no aesthetic function. It has an identification function: it lets the visitor compare it with what is in front of them and confirm they are in the right place. It is a second verification channel, which works even when the audio is ambiguous.

For it to work:

• Frontal photo of the whole artwork, not a detail, not an evocative particular. The detail is beautiful and useless for orienting: the visitor can't map it onto the whole.
• Include a bit of context around it: the frame, a piece of wall, the object next to it. A photo showing only the perfectly cropped canvas doesn't help find it again on a wall with twelve canvases. One showing the canvas in its position does.
• Taken from the point where the visitor stands, with the museum's real light. Catalog photos, taken in a studio with flat light, sometimes look nothing like how the artwork appears in the room.
• First image = the identification one. The beautiful photos, the details, the in-depth shots go from the second onward.
• Technical requirements: JPG or PNG, minimum 1200x800 px, landscape, under 2 MB.

For room POIs, the best photo is the overall view from the room entrance: it is literally what the visitor sees when they walk in.

4.7 Say what the artwork looks like, not just what it means

A detail that curators tend to skip because it is obvious to anyone who knows the artwork: the observable physical characteristics.

The visitor doesn't know whether they are looking for a 15-centimeter miniature or a four-meter canvas, whether it is in oil or tempera, whether it hangs at eye level or above a door.

Include in the first seconds at least two of the following:

• perceived size ("an enormous canvas, almost three meters wide", "a very small painting, a little larger than this screen");
• position in the space ("high up, above the door", "in the central display case", "at eye level on the long wall");
• dominant color or element ("dominated by blue", "the only portrait on a black background in the room");
• what is next to it ("between the two windows", "to the right of the marble bust").

This intervention has a double benefit: it is also an accessibility intervention. Physical descriptions are exactly what visually impaired visitors need, and those who visit with audio description.

4.8 POI zero: "how this visit works"

The first stop of every itinerary should not be an artwork. It should be the user manual of the visit, 60-90 seconds:

1. Where the route starts, so that whoever listens from the lobby knows where to go. ("The route starts on the first floor. Take the main staircase on your right just past the ticket office.")
2. How many stops there are and how long it lasts (manages expectations, reduces complaints about duration and price).
3. How the stops are numbered and where to find the numbers in the room. ("Every stop has a number. You will find it on a small plate next to the artwork's wall label.")
4. Whether the route is linear or free, and whether there are returns.
5. What to do if you get lost. ("If you realize you are in a different room from the one you are listening to, go back to the list of stops and choose the number of the room you are in.")
6. An invitation to download the content before entering if the internal network is weak: the web app offers the download when the itinerary starts and then works even without signal.

Point 1 is not a detail. One of the harshest feedback comments we received tells of 30 minutes spent looking for the starting point. The visitor was already angry before hearing the first word.

Put this POI as stop 0 or stop 1, and make sure it is the first item in the list.

4.9 Service POIs

Not all stops have to be works of art. You can create POIs (even without audio, with just text and image) for:

• restrooms, elevator, cloakroom;
• rest points: where you can sit, where there is a change in level, where the route is wheelchair accessible;
• bookshop and cafe;
• exit, and how to reach it from mid-route if you want to cut the visit short.

It seems off-topic in a guide about orientation. It isn't: a visitor who doesn't know where the restrooms are, where to sit or how to leave spends the second half of the visit looking for those things instead of listening. One of the feedback comments we collected comes from a pregnant visitor who simply asked to know where she could sit.

4.10 Declare returns, detours and excluded rooms

Many routes are not linear: you go back, you cross the same room twice, some rooms have no stops, some are closed.

A visitor who finds themselves going back thinks they made a mistake. In doubt they stop, check, and set off again unsure.

Just say so:

• "From here we go back through the room you have already seen: the next stop is beyond it."
• "You will cross two rooms without stops. Keep going to the gallery with the windows on the courtyard."
• "Room 8 is temporarily closed: from stop 7 go straight to 9."

On the last point: if a room closes, update the itinerary the same day. A visitor who looks for ten minutes for a stop in a closed room writes negative feedback, and rightly so.

5. Level 1: small physical interventions, big difference

5.11 Numbers in the rooms

This is the physical intervention with the highest return, and visitors explicitly ask for it: "why didn't you put a number on the artwork?", "equip the places on the visit with numbered panels matching the audio guide".

Practical specifications, learned from those who have done it well:

• Large and readable from three meters. Digits at least 4-5 cm tall. A visitor in a crowded room does not step up to the wall label: they look from a distance.
• High contrast, dark on light or light on dark. Bronze-colored numbers on beige marble are invisible.
• Next to the existing wall label, not in place of it, and always in the same relative position throughout the museum. Consistency matters more than elegance: if the number is always at the top left of the wall label, visitors learn where to look within three stops.
• One recurring piece of feedback is numbers that are too small: "the route numbers should be bigger so they can be identified more easily and quickly while listening to the audio guide". If you already have numbers and still receive this complaint, the problem is the size, not the absence.
• Room numbers at the entrance of every room, in addition to the numbers for the works: they answer the question "where am I?" even before the question "is this it?".

If you cannot hang anything on the walls (heritage protection constraints, historic installations, Soprintendenza requirements), the alternatives that work:

• numbered floor stickers, often approvable where a wall intervention is not;
• lecterns or flag-style totems, standing on the floor, with no fixings;
• numbers printed on the existing wall labels at the next reprint (marginal cost of zero);
• QR codes (5.12), which do not require numbering;
• a numbered floor plan (5.13), which moves the numbering from the wall to paper.

5.12 QR codes in the rooms, next to the works

The QR code solves the orientation problem in a different and stronger way: it eliminates the search. Visitors do not have to look for the stop in the app, they scan and listen to the thing physically in front of them. The match between object and content becomes impossible to get wrong.

It is particularly effective:

• in rooms with many works, where numbers get confusing;
• in free routes with no mandatory order;
• for visitors who join midway (groups, secondary entrances, repeat visits);
• for those who are lost and want to restart from where they are.

Practical notes:

• The QR code goes next to the work, at reading height, with a micro-instruction ("Scan to listen"). A QR code without an instruction is ignored by people who are not familiar with it.
• One QR code per work or per room is better than a single QR code at the entrance: the entrance QR code gives access, the in-room QR code gives orientation. They are two different functions and you need both.
• Test the print on site before producing everything: sufficient contrast, non-reflective surface, size readable at arm's length (generally at least 3x3 cm), adequate museum lighting.
• Remember that the QR code opens the web app without installing anything, so it works for everyone.
• The web app has a built-in scanner: those already inside the route scan the work's QR code from the app, without reopening the phone camera. It is activated per museum: ask us.

For how QR codes are generated and the types available (open, universal, single-use), see section 3. QR codes and access to content.

5.13 The numbered floor plan

A paper map with the stop numbers answers the question "where am I" in one go, and visitors ask for it often ("a map should be provided to understand where you are").

Two ways to have one without producing anything new:

• If you already have a paper floor plan, add the stop numbers at the next print run.
• If you do not have one, upload the floor plan image as the photo of POI zero in amuseapp (4.8): visitors have it on their phone, can zoom in, and it is free. It is not an interactive map, but it covers 90% of the need.

There is also a third way, available to all museums: the map inside the web app. From the console you upload the floor plans, draw floors and rooms, place stops and services on the plan, and visitors see where they are and where to go, in their own language. If you want to activate it on your route, write to us and we will set it up together.

Make sure the numbers on the map are the same as the numbers in the app and on the walls. Three different numbering systems are the most effective way to make things worse.

5.14 The 20-second script for the ticket office and room staff

Orientation starts before the app. Several negative feedback items have nothing to do with the product: they are about a visitor who asked for information and got a wrong, vague or annoyed answer.

Prepare one written sentence, not a training course: whoever sells the ticket must always say the same three things.

"The audio guide opens by scanning this QR code, nothing to install. The route starts on the first floor, from the staircase you can see over there. Every work has a number: you will find it on the small plate next to the wall label."

And a second sentence for room staff, for the most frequent question:

"If you are not sure which work you are listening to, look at the number on the plate and look it up in the list of stops."

Add an internal rule: staff must know where the route starts and how to download the audio. These are the two questions they will receive every day.

5.15 Aligning physical wall labels and audio

Small but insidious. When the wall label and the audio guide disagree on a date, a title or an attribution, visitors do not think "the label is old": they think "I am listening to the wrong work", and from that moment they stop trusting anything.

We have a piece of feedback that points to exactly this: a two-year discrepancy in a date between audio and wall label, and the visitor's comment is "it creates confusion".

During the content quality check, verify that audio and wall say the same thing on titles, authors, dates and attributions. If the wall label is outdated and you cannot reprint it, have the audio say so: "on the wall label next to the work you will find the old attribution, while more recent studies assign it to...". You turn an inconsistency into content.

6. Level 2: when orientation becomes automatic

The previous interventions help visitors find their way. These take the problem away from them.

6.16 BLE beacons: the audio starts by itself in the right room

Beacons are small battery-powered Bluetooth transmitters, applied near the works (double-sided tape or small discreet mounts, out of sight). The amuseapp mobile app detects them and automatically starts the audio corresponding to where the visitor is.

Why it solves the problem of this guide at its root: the question "is this what I am looking at?" no longer arises. The visitor walks, the audio starts, and the match is guaranteed by the system. It is the "hands-free" visit: you look at the work, not the screen.

Particularly suited to:

• large or labyrinthine venues, where it is easy to get disoriented;
• complex thematic routes with many closely spaced POIs;
• visitors with visual impairments, who cannot scan a QR code;
• family itineraries: children run from one work to another and autoplay keeps pace;
• venues where signage cannot be installed because of heritage protection constraints.

What to know before deciding:

• It requires the mobile app (it does not work on the web app) and Bluetooth turned on. A share of visitors have Bluetooth off or deny the permission: that is why the right pattern is beacons plus QR codes as a fallback, not beacons alone.
• The hardware is the museum's responsibility, generally one beacon for each POI to be automated. They are commodity devices, but we recommend tested models.
• Configuration (beacon ↔ POI pairing, distance threshold) is done in the console, with our support.
• The dashboard receives anonymous detection data: you also get a real attention map, showing which POIs get more time.
• Realistic timing for a first site: about four weeks between on-site testing, configuration and fine-tuning. On-site testing is not optional: stone, metal and the presence of visitors change signal propagation, and thresholds are calibrated on the spot.

Details on the page Indoor positioning with BLE Beacons.

6.17 GPS for outdoor areas

Where the route leaves the building (gardens, courtyards, terraces, historic villages, archaeological parks, routes spread across several sites), GPS does what beacons do indoors: turn-by-turn navigation with voice directions, with the route drawn in the console in the itinerary's dedicated tab.

The two technologies coexist on the same itinerary: beacons inside, GPS outside. It is the typical case of a castle with courtyards, or a museum spread across multiple sites. Automatic start of stops with GPS also works in the web app, not only in the app: it is activated per itinerary.

Note that in open spaces the orientation problem is worse than indoors, not better: there are no walls, no rooms, no wall labels. If your route has an outdoor section and you receive complaints about orientation, they likely come from there.

6.18 The chatbot as an escape route

If you have activated the AI virtual guide, a visitor who is lost can ask it in plain words where a work is or what is in the room they are in. One thing must be said clearly: the guide answers only with what it finds in the museum's content. If route directions are not written anywhere (in POI zero, in the transitions, in a practical information sheet uploaded to its knowledge base), it answers "I don't know, ask the staff": in our data this happens in almost half of practical questions. It does not replace anything above, but it is a zero-cost safety net if the feature is already active and the content is there. It is worth mentioning it in POI zero ("if you can't find something, ask the AI virtual guide").

7. The special case: temporary exhibitions

As mentioned at the start, here orientation can become the dominant problem: in one analyzed exhibition it accounts for 76% of negative comments. The good news: the cause is known and the checklist is short.

What goes wrong, and what to do:

• Cause: The audio is written from the catalog, and the catalog order is not the order of the walls. Countermeasure: A site visit with the exhibition layout finished, before opening, reordering the POIs on the real route.
• Cause: The audio covers only a selection of works, and the visitor does not know which. Countermeasure: Say so in POI zero: "this route explores 15 of the 60 works on display, those with the number next to the wall label"
• Cause: The audio is organized by thematic sections, and works from the same section are in different rooms. Countermeasure: Switch to one POI per work (4.4), with the section as an introductory POI.
• Cause: Loans change after opening. Countermeasure: Weekly check during the first three weeks: works that are replaced, moved or removed must be updated in the console.
• Cause: Many visitors are members or regular visitors of the museum and do not know the new exhibition layout. Countermeasure: Do not take knowledge of the spaces for granted: transition directions are useful for them too.

One piece of feedback perfectly sums up the typical failure:

"The works are not in the order explained, there is no real structured route, with the result that you get lost in the room looking for the works being talked about."

And another, on the emotional effect:

"We found ourselves running through the rooms to keep up with the explanations."

This second point deserves attention: if visitors feel they are chasing the audio guide, the visit becomes stressful even when the content is excellent. The cure is the same: a confirming opening, one POI per work, explicit transitions.

8. Writing directions well: rules and counterexamples

Directions end up in audio tracks that are automatically translated into dozens of languages. Simple sentences translate well, convoluted sentences translate badly, and in a foreign language an ambiguous direction becomes unusable.

Five rules:

1. Short sentences, one piece of information per sentence. These are also the sentences that speech synthesis reads best.
2. Fixed reference points instead of right/left, or at least anchored ones: not "on the right" but "to the right of the fireplace", not "at the end" but "at the end, past the glass wall".
3. Name the appearance first, then the title. Visitors recognize "the large painting with the white horse" before "The Triumphal Entry".
4. Numbers, always. "Stop 14" is verifiable, "the next work" is not.
5. Do not take the space for granted. Whoever wrote the text knows the museum by heart; the visitor is entering for the first time and in a language that is not their own.

Counterexamples, with the correction:

• Don't write: "Let us now observe the next work." Write: "Go to stop 15, the small portrait to the right of the window."
• Don't write: "On your right you will notice..." Write: "On the wall to the right of the room entrance you will notice..."
• Don't write: "Let us continue into the next room." Write: "Leave this room through the door next to the tapestry. Stop 9 is the first display case on the left."
• Don't write: "Above is depicted the Allegory of Justice." Write: "Look up at the ceiling, above the door you came in through: the Allegory of Justice."
• Don't write: "This masterpiece by the master from Urbino..." Write: "The painting in front of you, almost two meters tall, with the Madonna in blue in the center: it is the masterpiece of the master from Urbino..."
• Don't write: "The route continues." Write: "The route continues on the upper floor. Take the staircase at the end of the corridor, stop 20 is on the landing."

9. How to measure whether it is working

You do not need to wait for feedback to know whether orientation is working: the data you already have in the Analytics section of the console, and in the monthly report you receive by email, tells you sooner.

1. Drop-off between stops. In Analytics, look at the most listened-to points of interest and the completion percentage. A healthy route loses visitors gradually. If you see a sharp drop between two consecutive stops, almost always it is not the content of the stop that is bad: it is the transition that is impossible to follow. Go and look at that passage physically.

2. Itinerary completion percentage. If many visitors stop halfway, ask yourself whether there is a point midway where it is hard to understand where to go (a change of floor, a crossing, a passage room).

3. The "Signage and directions" item in the questionnaire. It is the direct indicator, and you find it among the questionnaire answers in Analytics. Look at it as a time series: if after applying section 4 it does not drop within two months, the problem is probably physical (section 5) and not editorial.

4. The average rating of those who cite the audio guide among the least appreciated things. If it rises while the number of mentions stays the same, you are solving orientation: the remaining complaints are lighter.

5. The five-visitor test. The most effective and the least used. Take five people who do not know the museum (relatives, interns, staff from another site), hand them the phone and say nothing. Follow at a distance of two meters with a sheet of paper and note only one thing: every time they stop and look around. Every stop is a point to fix, and you generally find five or six in an hour. No data analysis gives you this information with the same precision.

What to expect. The average rating of visitors who report signage problems is about 0.8 points below the overall average, and those who report both signage and audio guide issues are 1.4 points below. These are the points you are leaving on the table: a route that solves orientation brings those visitors back close to the average of their peers.

10. Action plan

Today, two hours in the console

• Number the POIs in real walking order (4.1)
• Check and correct the itinerary order (4.5)
• Create POI zero "how this visit works", with the starting point (4.8)
• Rewrite the first two lines of the 5 most visited stops with the confirmation formula (4.2)
• Add the transition instruction at the end of the same 5 stops (4.3)

This week

• Extend opening and closing to all stops (4.2, 4.3)
• Replace catalog photos with photos taken in the room from the visitor's point of view (4.6)
• Add physical descriptions of the works (4.7)
• Split the POIs that talk about multiple works (4.4)
• Add the service POIs (4.9)
• Hand the 20-second script to the ticket office and room staff (5.14)
• Run the five-visitor test (9.5) and fix the stopping points

This quarter

• Physical numbers in the rooms, consistent with those in the app (5.11)
• QR codes next to the main works (5.12)
• Numbered floor plan, on paper or as the POI zero image (5.13)
• Align wall labels and audio (5.15)
• Evaluate BLE beacons for the most critical areas (6.16)
• Activate GPS on the outdoor parts of the route (6.17)
• Re-read the drop-off metrics and the "Signage" item of the questionnaire

11. Frequently asked questions

We cannot put numbers on the walls, heritage protection rules forbid it. Are we cut out?

No. The in-app numbering remains useful with the floor plan and with staff, and QR codes or floor stickers solve it without touching the walls. In the most constrained cases, beacons eliminate the need for signage completely.

Do we need to redo the content?

No, and in most cases it is the wrong thing to do first. Interventions 4.1 to 4.5 do not touch the content: they add a line at the opening, one at the closing, and reorder. The content is enriched afterward, when visitors are able to follow it.

If I edit the extended description, is the audio regenerated in all languages?

Yes. amuseapp regenerates the audio and translations, generally within 1-2 working days. In the meantime the previous version stays available, so there is no gap for the visitor. It is best to make changes in bulk across all stops rather than one at a time.

Our route is free, with no mandatory order. How do we number it?

With a free route sequential numbering makes less sense and in-room QR codes become the main solution (5.12): visitors listen to what they encounter. Still keep POI zero and the room POIs, and number by room rather than by work.

We have many guided groups. Does the same apply?

Yes, and more: in groups, a visitor who loses the thread cannot stop to look around, because the group moves on. Transition directions and position confirmation matter even more.

Better numbers or QR codes?

They do different things. The number is faster to read from a distance, works with any visitor and supports reading the floor plan. The QR code eliminates the search but requires taking out the phone and scanning. The best combination, if you can, is a large number on the wall label, QR codes on the main works.

I already have numbers in the rooms and visitors still complain.

Three typical causes, in this order: the numbers are too small or low-contrast (specifications in 5.11); the numbers in the rooms do not match those in the app; the stops lack the transition instruction and the visitor finds stop 12 but does not know where 13 is.

How do I know whether my problem is orientation or connectivity?

The distinguishing signal is the distribution of ratings. Orientation produces many 2 and 3 ratings with comments about route, order, "I can't find the work". A technical problem produces a U-shaped distribution (many 5s and many 1s) with comments about audio that won't start, downloads, connection. If you see the U, talk to us: it is a different case, with different solutions (pre-download before entry, wifi coverage at reception).

12. Printable checklist

For each stop of the itinerary:

• It has a number, and the number is the one the visitor sees in the room
• The audio opens by saying where you are and how to recognize the object
• The audio closes by saying where to go and which stop comes next
• It talks about a single work (or declares itself an overview)
• The first photo shows the work as seen from the visitor's position
• There is at least one observable physical attribute (size, position, color)
• The position in the itinerary matches the real walking order
• Title, author and date match the wall label

For the itinerary as a whole:

• There is a POI zero that says where the route starts
• Returns, detours and rooms without stops are declared
• Service POIs are in place (restrooms, rest areas, exit)
• Closed rooms are up to date
• Staff know where the route starts and how to access the audio
• The drop-off between consecutive stops has no unexplained collapses

If you are reading this page because you received negative feedback on orientation, the fastest way to start is the "Today, two hours in the console" list in section 10. If you prefer, we can look at your route's data together: the Analytics section of the console shows exactly at which passage visitors stop, and from there the intervention can be targeted.