The local recommendations in TouchGuides have been a name, a photo and a line from the host for a while. Watching what guests did with them made it clear that was about half of what was needed.
The pattern in the data was blunt: guests opened a recommendation, looked at it, and then left the guidebook to search for the same place somewhere else. They were leaving to find out two things — whether it was open, and how far it was.
So the place cards now carry those.
What is on a card
Opening hours, with today's highlighted. This is the one that mattered most. A recommendation a guest cannot act on tonight is not a recommendation, it is trivia, and hours are the difference between the two. The Sunday and Monday closures that ruin a walk are now visible before the walk.
Walking distance and time from the property, given in minutes. Nobody converts distance into effort in their head at the end of a long day: "eleven minutes" is a decision, where "900 metres" is arithmetic.
A photo strip rather than a single image. One photograph of a restaurant frontage tells you where the door is. Three or four tell you what the room is like, which is what people are actually judging when they decide whether somewhere suits a Tuesday with a toddler or a Friday without one.
A rating and a review, from the place's own public listing. Included because guests were leaving to check anyway, and a number they can see in place is one less exit.
Your line about why. Unchanged and still the most important element on the card. The hours and the distance are facts anybody can look up; the sentence explaining that the food is fine rather than good, and that you go there to talk, is the reason the guidebook is worth reading at all.
The map
Every place you pick is now a pin on one map, with the property in the middle. Tapping a pin opens the card, and each card still has its own directions button.
This is a small change with a large effect on how the recommendations get used, because it answers a question a list cannot: which of these are near each other. A guest planning an afternoon wants the bookshop and the coffee place that happen to be on the same street, and no vertical list will ever show that.
The map is in the printed PDF too, drawn as a static image with the pins on it, so a paper copy carries the same geography. It has been tested through a photocopier, which turns out to be a real constraint for hosts with binders.
Picking places
The picker searches the area around your property — restaurants, cafes, bars, shops and sights — and you choose. It now looks out as far as the one-click fill does, which sounds like housekeeping and fixed a genuinely confusing behaviour where a place the automatic fill had suggested could not be found by searching for it by hand.
Two things this deliberately does not do:
It does not pick for you. There is a one-click fill that proposes a starting set, and it exists to give you something to edit rather than something to accept. A list of the twelve highest-rated restaurants within a mile is the thing every guest can already generate on their own phone. The value you add is subtraction, plus the sentence explaining the choice.
It does not keep the list in sync automatically. Hours change, places close, and we show what was fetched. Re-check your list a couple of times a year. The audit post on this blog has the full routine.
What did not change
The recommendations are still host-picked. Nothing is inserted into your guidebook that you did not choose, no sponsored entries exist, and nobody pays to appear on your map. That is worth stating plainly because it is the obvious way a product like this would make money, and it is a road we are not going down.
There are also no paid add-ons involved. Place cards, hours, distances, the map and the printed version are part of the plans as they stand.
What is next
The thing guests ask for that we cannot yet answer well is proximity to each other rather than to the property — a walking route through three places rather than three separate walks. That is a harder problem than it looks and there is no date on it.
In the meantime, the highest-value thing you can do with this release takes ten minutes: open your recommendations, delete the ones you have never been to, and write one honest sentence under each of the five that remain.