Kyoto in 3 days for 4 people: a composite planning doc

Quick answer: for 4 people in Kyoto for 3 days, plan one anchor sight per day (Higashiyama temples, Arashiyama bamboo grove, Fushimi Inari at sunrise), stay central within walking distance of Karasuma subway (~A$680/night for a 4-bed shoulder-season Airbnb), book one nice dinner from home before flying, schedule a buffer afternoon, and run debt-simplification at trip-end (three transfers max for four people). Honest 4-day rewrite at the bottom — Kyoto really wants 4 days, not 3.

This is not a top-10 list.

Top-10 lists are the worst format for group trip planning, because they hand you ten unconnected places and leave you to figure out how to sequence them with three other people who all want a different five.

This piece is a composite planning document — built from real Kyoto trips we and our beta testers have run, with the day plans grounded in actual logistics, the costs given as representative shoulder-season ranges, and the names and exact numbers stylised. The framework is the durable thing; the specific places and prices are anchors for your numbers, not ours.

The trip we're describing: 4 people, 3 nights, late October (shoulder season), arriving from Tokyo. Representative per-person spend: A$650-900pp excluding flights (verified against Kyoto shoulder-season Airbnb listings as of early 2026 — your own dates will shift this 15-40% in either direction). Settlement at trip-end with three transfers — the n−1 math works the same regardless of the exact numbers.

The crew

Four people, intentionally varied:

• A (Sydney) — has been to Tokyo three times, never Kyoto. Pescatarian. Strong opinions about coffee.
• B (Melbourne) — first Japan trip. Mild gluten intolerance. Wants to walk a lot.
• C (Singapore) — has lived in Tokyo for a year, treats Kyoto as "the day-trip city." Vegetarian.
• D (Auckland) — third Kyoto trip, knows the city as well as any of us. Eats anything. Resident expert.

The four roles mapped naturally: D was the local (no contest), A was the scout (always with new places from food blogs), B was the treasurer (calm with spreadsheets), and C was the decider (least invested emotionally, best at calling ties). Nobody volunteered for decider — we asked.

The shape of the trip

Before any specific places, we agreed on the shape:

• 3 full days (not counting arrival/departure transit days).
• Stay central to minimise daily transit time. Walking distance to a subway station was non-negotiable.
• One anchor per day + 2-3 supporting stops. Not six things crammed in.
• One buffer afternoon somewhere in the trip.
• At least two group dinners, where everyone sits down together.
• One nice dinner booked from home (we learned this the hard way — see below).

This shape is the most durable thing in the document. The specific places change every trip; the shape works for any East Asian city.

Accommodation: ~A$680/night (shoulder season), 4-bed Airbnb, central Kyoto

Representative pricing for late October 2025/early 2026, booked 6 weeks out. Early November (autumn leaves) and cherry blossom (late March / early April) both 40-60% more. Verify current Airbnb / hotel listings against your specific dates.

We picked the area roughly bounded by Nishiki Market (north), the Kamogawa river (east), Shijo-dori (south), and Karasuma subway (west). This means:

• 5 minutes walk to Nishiki Market for breakfast.
• 10 minutes walk to Pontocho for evening drinks/dinner.
• 8 minutes to Karasuma subway for Arashiyama and any trip out of central.
• 15 minutes walk to the Kamogawa for the morning run/walk crowd.

Lesson: pick the transit graph before the specific apartment. Subway proximity matters more than apartment aesthetics for a group trip.

Day 1: Higashiyama (eastern temples)

Theme: the postcard Kyoto. Old wooden streets, lanterns, the temples everyone has seen on Instagram.

Time	Stop	Note
09:00	Breakfast — Inoda Coffee (central)	Their classic morning set. Order ahead, queue otherwise.
10:30	Kiyomizu-dera (temple, the anchor)	30 min via bus or 25 min walk. Arrive before 11:00 for crowds.
12:30	Walk down Sannenzaka / Ninenzaka	The old streets. Slow walk, ~45 min with browsing.
14:00	Lunch in Higashiyama district	Tofu set lunches everywhere. We did one with a garden view.
15:30	Yasaka Shrine + Maruyama Park	Easy, no entry fee, lots of resting spots.
17:00	Walk to Gion via the Shirakawa	Sunset light. ~30 min.
19:00	Group dinner in Pontocho	Booked 2 weeks ahead. Worth it.

Total walking: ~9 km. Illustrative cost per person: A$95 (lunch A$30, temple A$5, dinner A$60). Treat the breakdown as a structure, not exact figures.

What worked: Higashiyama is dense enough that one day in one district is enough. We didn't try to add Arashiyama, despite the temptation. What didn't: Kiyomizu by 11:00 was already busy. Next time, 09:30 entry.

Day 2: Arashiyama (bamboo grove + western temples)

Theme: outside the city. Bamboo grove, river views, a long walk through the temples.

Time	Stop	Note
07:00	Local breakfast near apartment	Convenience store onigiri + coffee. Save appetite for lunch.
08:00	Train to Arashiyama (Sagano line)	~20 minutes. Arrive before 09:00.
09:00	Bamboo grove (the anchor)	Empty before 09:30. Mobbed by 10:30.
10:00	Tenryu-ji temple + garden	Adjacent. Walk through to the back garden exit.
11:30	Walk along the river path	30-min walk to Hozu river boats / monkey park.
13:00	Lunch — Arashiyama tofu lunch	Multi-course tofu, riverside.
15:00	Train back to central Kyoto	Buffer afternoon — see below.

Total walking: ~7 km. Illustrative cost per person: A$85.

What worked: 09:00 arrival at the bamboo grove was transformative. The grove with no-one in it is silent in a way that doesn't survive other people's selfie sticks. What didn't: we tried to add the monkey park and ran out of energy. Skipped it. Next time, monkey park OR temples, not both.

The buffer afternoon

Back in central by 16:00, we deliberately had no plan. What actually happened: A and B took naps. C went to a coffee shop and read. D went for a walk along the Kamogawa. We reconvened for dinner.

This was the single best decision of the trip. Do not skip the buffer. The temptation will be enormous. Resist.

For dinner that night, no booking — Nishiki Market food hall + a walk through Pontocho looking for somewhere with a table. Took 25 minutes to find a place. Acceptable.

Day 3: Fushimi Inari at sunrise + central Kyoto

Theme: the early-morning trip everyone says they'll do and most groups don't.

Time	Stop	Note
05:30	Wake up	Painful. Do it anyway.
06:00	Train to Fushimi Inari	~15 minutes from central. Train runs early.
06:30	Hike up the torii gates	Empty. Watch the sunrise from one of the upper shrines.
09:00	Back in central — second breakfast	Earned. Treat yourself.
11:00	Nijo Castle (the anchor)	The shogun's residence. Not the most exciting castle in Japan but it's here.
13:30	Lunch — Nishiki Market crawl	Standing food, multiple stalls, low commitment.
15:30	Coffee + walk along Kamogawa	Slow afternoon.
18:30	Farewell dinner — the booked nice one	The single splurge. ¥12,000 per person.

Total walking: ~10 km. Illustrative cost per person: A$185 (the ¥12,000 dinner accounts for ~A$130).

What worked: Fushimi Inari at sunrise is the moment of the trip. The lower torii gates with no-one in them is what every photo of Kyoto wishes it was. By 09:00 on the way down, the crowd was already significant — we were already leaving as people were arriving. What didn't: Nijo Castle was a 20% experience after the temples of the previous two days. Next time, replace with a coffee shop crawl or a kimono fitting.

The expense math (worked example)

What the settlement looks like in the workspace at trip-end. Numbers below are illustrative — round figures picked to show the algorithm cleanly; your trip's numbers will be different.

Total trip spend across 4 people: A$3,200. Equal share per person: A$800.

Net positions, after applying everyone's payments against their A$800 equal share:

• A: paid A$1,140 (fronted accommodation). Net: +A$340 owed to A.
• B: paid A$680 (small spend + treasurer). Net: −A$120 owed by B.
• C: paid A$540. Net: −A$260 owed by C.
• D: paid A$840. Net: +A$40 owed to D.

Check: net positions sum to zero (340 − 120 − 260 + 40 = 0). ✓

Settlement using debt simplification — at most n−1 = 3 transfers for 4 people:

1. C → A: A$260 (clears C completely; A now at +A$80 net)
2. B → A: A$80 (clears A completely; B now at −A$40 net)
3. B → D: A$40 (clears both B and D)

Three transfers. Everyone at zero. Posted in the workspace as a single message that everyone reacts to when they've executed their transfer.

The math always converges in at most n − 1 transfers regardless of how many expenses there were — see the expense fronting piece for the algorithm. The point of including this worked example is to show what the output looks like in a real planning doc. The numbers are illustrative; the shape is the real takeaway.

What we got wrong

Four things, in order of severity:

1. We didn't book one nice dinner in advance. Two of our target spots required week-ahead reservations and we were turned away both times. The booked Pontocho dinner on Day 1 was the only "real" sit-down dinner — by Day 3 we were eating standing food at Nishiki. Not bad, but not what we'd planned. Book one fixed-point dinner from home. Build the trip around it.
2. Day 3's Nijo Castle was the wrong choice. After two days of temples, a third temple-adjacent attraction was a marginal experience. Should have been a coffee shop tour or a kimono fitting.
3. We didn't pre-book Fushimi Inari for the whole group. A and D made the 05:30 wake-up. B and C bailed. Should have set the expectation 48 hours out, with a clear opt-in/opt-out moment.
4. We over-packed Day 1. 9km of walking after a travel day was too much. Should have made Day 1 lighter and Day 2 heavier.

What we'd do differently next time

If you're locked to 3 days: keep the structure above, but cut Fushimi Inari (or replace Day-3 Nijo Castle with the Fushimi sunrise). 3 days works if you're willing to skip something — but on reflection, the honest answer is that this is a 4-day city for most groups, not a 3-day one.

A clean rewrite, given everything we learned:

• Day 1 (lighter): afternoon arrival from Tokyo, evening Pontocho stroll, light dinner. No temples.
• Day 2 (Higashiyama, the same): Kiyomizu by 09:30, Sannenzaka, Yasaka, Gion at dusk. Group dinner.
• Day 3 (Arashiyama, half day): 09:00 bamboo grove + temples + lunch. Back by 14:00. Buffer afternoon. Booked nice dinner.
• Day 4 (Fushimi at sunrise + light central): 05:30 wake, sunrise hike, second breakfast, coffee shop crawl, farewell drinks at the Kamogawa.

That's 4 days, not 3 — which is the honest call. 3 days works if you're willing to skip something. 4 days is when it stops feeling rushed.

The template, abstracted

Strip out Kyoto and what remains is a planning template that works for any 3-4 day city trip with a small group:

1. Pick the shape first: number of full days, one anchor per day, one buffer, one nice dinner booked from home.
2. Pick the neighbourhood for transit, not aesthetics: walking distance to a transit hub beats a nicer apartment 15 minutes from one.
3. Schedule by district, not by enthusiasm: group nearby things on the same day.
4. Use the four roles to distribute load. The local is irreplaceable.
5. Vote on activities, not on logistics: use the 80% rule for ideas, the 50% rule for dates and bookings.
6. Log expenses as they happen, settle once at end: use debt simplification, 3 transfers max for 4 people.
7. Build in one buffer. Always.

This same template scales: Tokyo, Lisbon, Mexico City, Buenos Aires, Cape Town. Different cities, same shape.

Where this fits

This is the first Destination Deep-Read — the format for every subsequent destination piece. We'll do Tokyo (with the dietary-tolerance variant), Sydney long-weekend, and Mexico City in the same shape. Each one is the real planning doc, not a top-10 list, not a marketing piece.

If you're using Wendir's beta to plan a real trip, the agents map to roughly the same flow: Scout proposes places, Local verifies (Kyoto's "D" in human form), Moneybags handles the n−1 settlement. Closed beta, iOS-first. Waitlist.

If you're using a doc and a chat: this template works. Copy the day-plan table format. Set the four roles. Pick the shape first. The trip will work.

More from the Destination Deep-Reads

• Tokyo for friends with three different food tolerances — coming next.
• Sydney long-weekend for a hens — the spreadsheet that worked — coming after that.
• Mexico City for 5 people, 4 days, no Spanish — later in the year.

And the Manual pieces this trip ran on:

• The main planning manual
• The 4 roles
• The 80% consensus rule
• How to split travel expenses

---

Written by the Wendir team. Last updated: 15 May 2026.