Zedo: trust infrastructure for finding a home in Nigeria

1. 01

   Say what it understood, as chips.

   Above the results: “I understood · 3 bedrooms · Duplex · For sale · Ikeja”.

   WhyA free-text box hides how the sentence was read. Chips make the interpretation checkable in one glance, before any card.Judgement

2. 02

   Show the work while it runs.

   Four named steps tick off: Understood your request, Searching listings (126,184), Matching your requirements (5 found), Ranking the best matches, over skeleton cards in the shape of the results.

   WhyThe corpus size and the match count set expectations, and a named step reads as progress where a spinner doesn’t.Judgement

3. 03

   Split Exact matches from Worth a look.

   A green “Matches everything you asked” tag for the exact result; a separate group, “4, close, but not quite what you asked”, for the rest.

   WhyThe split answers “is this what I asked for?” before the buyer reads a word of the card.

   CostThe near group is longer than the exact one, so it can bury the best option if its range is too loose. See “What I’d fix next”.

4. 04

   Write the difference on the card.

   “4 bedrooms, one more than you asked.” “6 bedrooms, 3 more than you asked.”

   WhyA number mismatch is the commonest near-miss for a request like “3 bedroom”. Stating it removes a comparison the buyer would otherwise do by eye. It is amber rather than red because it is a difference, not an error.Judgement

5. 05

   Answer in one sentence, then offer the next question.

   “The Opebi listing is the only exact three-bedroom match; the other four are larger Ikeja duplexes, priced between ₦220m and ₦2.1bn.” Below it, “Narrow it down” chips, including a non-filter question: “What are the title documents for the Opebi property?”

   WhyAnswer first, with the price range as a sanity check. On a phone a second query costs typing, so the likely next question becomes a tap, and the title-document chip points at the question a buyer of property actually worries about.Judgement

6. 06

   Show provenance and freshness on every card.

   “Source: PropertyPro.ng ↗”, “updated Sep 21”, and “verified Sep 28” where an agent is verified, plus a standing line: “Zedo AI suggests properties from available listings. Always confirm details with the agent.”

   WhyAggregated listings are only as good as their source. Naming the source and the date lets the buyer calibrate how far to trust each card.Judgement

Use Reveal system to see the details I’d point to in a design review. Third-party listing photos and agent names are blurred.Design-level

Surface · the product

Reveal system

123456

1. 1The group says what it is“Worth a look · 4, close, but not quite what you asked” names the rule and the size of the group before any card is read.
2. 2Trust stays on the cardThe Verified tag sits on the photo, in the scan path, not behind a profile click.
3. 3The gap, in words“4 bedrooms, one more than you asked” states the difference so nobody has to compare numbers by eye. Amber, not red: it is a difference, not an error.
4. 4Price at full sizeThe gap between ₦350,000,000 and ₦2,100,000,000 is the first thing visible, which is exactly the thing that decides whether a near-miss is worth a look.
5. 5A flaw: missing data shown as a fact“Not Specified” reads like a value. I would replace it with an action, such as “Ask the agent”.
6. 6Who, and how freshThe listing agent (blurred here) and a “verified Sep 28” date let a buyer judge how much to trust the card.

1. 01

   Start from the brief, not the form.

   The first field is “Your brief: Paste it from WhatsApp or describe the property. We fill in the rest.” Photos and the brief sit side by side.

   WhyThe agent already has the text. Asking for it where it already exists removes the retyping.Judgement

2. 02

   Offer to fill the details; don’t do it silently.

   A prompt appears: “Looks like a full brief. Pull out the price, location and rooms?” with Fill details, or dismiss. The result lands in a collapsed “Listing details: Filled in from your brief. Open to check or change.”

   WhyA wrong price published silently is worse than one extra tap. An explicit offer and an “open to check” row keep the agent in charge.Judgement

   CostOne extra tap on every listing.

3. 03

   Make the cover image from the first photo.

   Five styles: Poster, Dark, Light, Minimal, Original photo. “Made from your first photo. Press and hold a photo above to change which one is first.”

   WhySocial posts need an image carrying the price and the contact. The first photo is already the one the agent chose as the cover, so no new choice is needed.Judgement

4. 04

   One caption per network, each rewritable.

   Tabs for X, Facebook, Instagram, Threads and LinkedIn, each with its own caption and a “Write again”. Five small badges show which channels are on.

   WhyEach network has its own length and tone. Regenerating one caption shouldn’t disturb the others.Judgement

5. 05

   Timing is a choice with a safe default.

   “Defaults to posting right away; pick a time only if you want to hold it until then.” Scheduling asks what should go live: Zedo + Social accounts, or Zedo only. A missing date gets a plain fix: “Pick a date and time, or switch back to Post now.”

   WhyMost listings should go out now. Scheduling stays available, with its consequence spelled out and an error that says how to recover.Judgement

1. 01

   Choose who can see a listing at the moment of posting.

   Two options, each with its consequence in plain words. Circle only: “Only your members see it. It’s watermarked with each member’s name and never cross-posted.” Public: “Everyone on Zedo, and cross-posted to your connected accounts.”

   WhyThe choice is made where the stakes are clear, inside the same Post flow, and the watermark makes a leak traceable to a member.Judgement

2. 02

   Approve first, pay after.

   “People request to join, and only pay once you approve them.”

   WhyThe host controls who gets in, and nobody pays for a request that is then refused.

3. 03

   Identity checks as switches the host controls.

   “Verification to join”, and separately “Verification for owner contacts: members must verify their identity before seeing who owns a property.”

   WhyRevealing an owner’s contact is the sensitive moment, so it gets its own gate rather than riding on membership.Judgement

4. 04

   Pricing power is earned.

   A new host starts at ₦10,000 a month. Higher steps open as the host climbs milestones: ₦20,000 at Level 3, up to ₦60,000 at Level 11. “Closest to done: share 3 listings (1 more listing to go).” Listing in Discover unlocks after Level 2.

   WhyA new host can’t charge a premium before building a record, and members are protected from steep prices on unproven Circles.Judgement

5. 05

   Closing a Circle doesn’t cost members what they paid for.

   “Members who have paid keep access until the end of the period they paid for, so no one loses time they bought.” Payouts: “Your share of every payment goes straight to your bank. Zedo never holds it.”

   WhyA clean exit and a payout that doesn’t pass through Zedo are both trust features for the people paying and the person being paid.Judgement

The deep dives sit inside a larger product. These are the other surfaces, each shown once. Dark and light themes are both supported.Design-level

Before the current public site and app, the product was a dark, chat-first web app. These are screens from that beta (2025), designed in Figma under my direction. Third-party listing photos and agent names are blurred.Design-level