Conversion teardown

4 funnels on zucity.org, where the leaks are

Each funnel below is mapped end-to-end as a visitor would walk it. Findings are tied to things you would see on the live site today. Fixes are tagged by effort. Lift framing is qualitative because made-up percentages are worse than no number.

Three caveats before the teardowns. First, I am working from a few site visits, not from your analytics, so anything that looks obvious from a heatmap will be invisible to me. Second, some findings will already be on your roadmap, treat those as confirmation. Third, mobile assumptions are based on common patterns, ground-truth them on your own device before acting.

Funnel 01 · Highest leverage

Membership signup

Step 1

Landing on zucity.org

Step 2

Click "Membership"

Step 3

Tier select
(VIP / Daily / Global)

Step 4

Account or checkout

Step 5

Payment confirmed

What's leaking

Tier differentiator is buried. VIP, Daily Access, and Global Membership are easy to confuse on first read. Visitors who cannot tell the three apart in 8 seconds bounce to "I'll think about it" rather than pick wrong.
No social proof on the tier-select screen. The decision moment is where alumni names, member count, or a single quote would do the heaviest lifting. Right now the tier page asks for money before it asks for trust.
Pricing visibility before commitment. If a visitor has to start an account flow to see a real price, drop-off compounds. Pricing should be readable from the public tier page without any form interaction.
No urgency anchor. A themed week starting in 18 days, a cohort capping at 40, or a Karuizawa property opening next month would convert hesitation into action. The current page reads as evergreen, which is the enemy of signup.
Mobile scroll length. If the tier comparison stacks vertically on mobile without a sticky CTA, the buy button vanishes the moment a visitor scrolls past it. Common on coliving sites, worth a check.

Fixes

1hr Add a 1-line differentiator under each tier name. Example pattern: "VIP · for the founder who wants a key to every property" / "Daily · for the nomad testing a week" / "Global · for the community member who travels with us." One sentence each, written from the buyer's identity, not the feature.
half-day Stack 3 testimonial cards (face photo, name, city, one quote) directly above the tier-comparison table. Pull from existing Zucity members who already post on X about staying with you. No new content shoot needed.
day Add a sticky bottom CTA on mobile that follows the user through tier-select and into checkout. Single button, current tier name shown, price next to it. Removes the "where did the buy button go" friction completely.

Estimated lift High This is the funnel where money changes hands. Even modest gains compound across every themed week and member acquisition channel.

Funnel 02

Event registration via Lu.ma

Step 1

Landing on zucity.org/en/all

Step 2

Browse event cards

Step 3

Click "Register"

Step 4

Land on Lu.ma page

Step 5

RSVP, exit Zucity context

What's leaking

Brand handoff to Lu.ma is jarring. The visitor goes from a Zucity-branded page to a generic Lu.ma layout with the event creator's avatar at the top. The trust signal that brought them ("Zucity is hosting this") gets diluted at the exact moment they commit.
No return path after RSVP. Lu.ma does not natively suggest "explore more Zucity events." The visitor RSVPs to one event and leaves. A 7-day membership browser who could have seen 6 more relevant events sees one.
Event filtering is thin. With 50+ events booked May to October, a scrollable list is not enough. Without filters by theme (d/acc, DeSci, jp/acc), city (Karuizawa, Tokyo, Chiang Mai), or month, the relevant event is buried under noise for a first-time visitor.
Email data ownership. When RSVP happens on Lu.ma, the email lives on Lu.ma's list first. If Zucity is not auto-syncing those addresses into its own newsletter, the relationship is rented, not owned.
Event cards lack the "who else is going" signal. Coliving events convert on who is in the room. Showing 3 attendee avatars or "12 builders already RSVPed" on the card itself is a stronger pull than the event title.

Fixes

half-day Add filter pills above the event list on zucity.org/en/all: Theme, City, Month. Three pills, multi-select, no page reload needed. Reduces time-to-relevant-event from 30 seconds of scrolling to 3 seconds of tap-tap.
day Build an intermediate Zucity-branded confirmation page between the event card click and the Lu.ma redirect. Two lines of context ("You are RSVPing to a Zucity event in Karuizawa, this opens Lu.ma in a new tab"), one CTA, then redirect. Preserves brand trust through the handoff.
1hr Add attendee count or 3 avatar previews to event cards on the listing page. If Lu.ma has the data via API, this is a half-day build. If not, a manual count refreshed weekly is fine for a v1.

Estimated lift Medium Event registration is high-volume but lower-value than membership. The data-ownership fix is the long-term compounder here.

Funnel 03

Accommodation booking

Step 1

Landing on zucity.org/en/all

Step 2

Filter to accommodations

Step 3

Open a property listing

Step 4

Inquiry or booking form

Step 5

Confirmation or partner redirect

What's leaking

Instant booking vs inquiry is unclear. A nomad comparing Zucity to Airbnb at midnight wants to know in 2 seconds whether they can book now or wait 48 hours for a reply. If the listing does not say which, most pick the option that books now.
Partner properties not clearly labelled. ADDress and MIDORI.so listings should be visually distinct from Zucity-direct properties. A small partner badge ("via ADDress") sets expectation around pricing, booking flow, and what Zucity owns in the relationship.
No response-time guarantee on inquiry forms. If the booking is inquiry-based, "we reply within 24 hours" near the submit button removes the silent black-hole anxiety that kills 30% of inquiry submissions.
Photo quality is the silent killer. Coliving converts on rooms looking like rooms a person actually wants to wake up in. Any property with stock photos, gradient placeholders, or three years old smartphone shots loses to one with real recent photography.
No "stay before joining" trial path. A first-time nomad who wants to test Zucity before paying for Global Membership has no clear "try a week as a guest" pathway visible from the listing page.

Fixes

1hr Add a single-line status above the booking CTA on every listing: "Instant booking" (green dot) or "Inquiry, reply within 24 hours" (amber dot). Two states, no ambiguity.
half-day Add a partner badge to all ADDress, MIDORI.so, and other non-Zucity-direct listings. Top-right of the card, 11px text, low-key but present. Pairs with a one-sentence partner explainer linked from a help icon.
day Build a "Trial Stay" CTA on the membership tier-select page that routes to a curated 3-property shortlist priced for first-timers. Closes the loop between Funnel 1 (membership signup) and Funnel 3 (accommodation), and converts hesitant tier-pickers into committed first-stay guests.

Estimated lift Medium Booking is where reputation is built. The photo and labelling fixes do not move conversion fast but they protect long-term word-of-mouth.

Funnel 04

Themed week deep-funnel

Step 1

Landing on zucity.org

Step 2

Click themed week
(d/acc, DeSci, jp/acc)

Step 3

Scan the week page

Step 4

Application or registration

Step 5

Confirmed attendee

What's leaking

5-second value-prop test. A first-time visitor landing on the d/acc or DeSci page should answer 3 questions in 5 seconds: what is this week, who is there, what happens. If any of the three takes scrolling to find, the page is doing too much work.
Speaker and mentor names missing above the fold. Themed weeks convert on who else is in the room. Even 5 named faces above the fold ("Joining: Vitalik, X, Y, Z") moves the page from "interesting idea" to "I need to be there."
Open registration vs application is conflated. If d/acc takes an application and DeSci is open registration, the pages should communicate that difference in the first sentence. Right now the visitor cannot tell which without trying to register.
No low-commitment follow path. A visitor who is curious but not ready has no "add this week to my calendar" or "tell me when applications open" option. They leave, forget, never return.
Past-week artifacts are underused. If d/acc 2026-03 produced 12 talks now on YouTube and 4 published essays, those should be linked from the d/acc page as proof of what attendance produces. Most themed-week pages forget to show the receipts.

Fixes

1hr Add a single line under the themed-week H1 that answers all 3 questions at once. Pattern: "10 days in Karuizawa with 40 d/acc builders. Workshops, talks, and unstructured time. Application required, opens 1 July." One sentence, every word load-bearing.
half-day Build an "Add to calendar / notify me" CTA on every themed-week page as a secondary action next to the primary apply button. Captures the 60% of visitors who are interested but not ready, into an email list segmented by week.
half-day Add a "Past weeks" block linking to recordings, essays, photos from prior cohorts of the same week. Even 3 thumbnails with names is enough. Converts the "is this real" hesitation into "yes, and here is what they produced."

Estimated lift High Themed weeks are the highest narrative-value surface Zucity has. Conversion gains here also feed the membership and accommodation funnels downstream.

Pick what's tractable, drop the rest. Membership signup is the highest-leverage funnel, so if you only fix one thing on the whole site, fix the tier-select page: clearer differentiator, social proof above the table, sticky mobile CTA. Everything else compounds slower.

Donal · Online Optimisers · 2026-05-28