Overview
What stands out
Community size
Active people per day
New accounts per day
Conversations per day
Messages per day
What people did
Actions counted in the last 24 hours and the last 7 days, across the whole community. The most frequent action is the one the app is really used for.
Conversation funnel
Where people drop off between "I saw someone" and "we're actually talking".
Community health
The signals that say whether the app is alive or just full. Thresholds from docs/GROWTH.md.
Unlock ladder
How many pairs reached each threshold and switched the feature on (both had to approve). The gap = people who could, but didn't.
Top places
The base for local partnerships: this is where people gather.
Where the time goes
Share of all foreground seconds, by screen. Explore, People and Thoughts are the three feeds; Chat is an open conversation.
Most frequent actions
Every tracked action, counted across all sessions. Per person numbers are on the right.
Sessions per day
When people open the app
Sessions by hour of day, Romanian time, last 7 days.
Who's still around — by signup week
Each row is the people who joined that week. “Came back” = opened the app at least a day after signing up. The last column is who was active this week.
Where everyone is right now
Every account, in exactly one bucket. Slipping is the group to win back first: they liked it once, and we still have a week or two.
How long today's active people have been here
Ages
From the birth date typed at signup.
Gender & language
Devices & builds
From sessions, last 7 days. Verified contact below.
Thoughts & Journal over time
Check-ins per day
Journal
Photos and videos for close friends (200+ messages together). They live 24 h, then go to the owner's archive.
Thoughts & polls
Only live thoughts count. Settings are the author's choice when publishing.
Chat unlocks
Pairs that reached each threshold, and how many actually used what it unlocked.
Journal streaks
Days in a row with a Journal post. Alive while the last post is from today or yesterday.
Crowns
A crown = #1 visitor of a place. Moves when someone else visits more.
Comeback nudges
At most one per person per day, only to people who weren't in the app lately.
City totals only — no user position ever appears here. Drag to spin.
Cities over time
People per city, the four biggest. The first city to pass 150 active people is where the app starts to run by itself.
Cities
A city comes alive around 150–200 active people. Below that, people open the app and find nobody.
| City | People | Active 7d | New 7d | Critical mass |
|---|
Most visited
Sorted by real visits (one per person per day). These are the partnership candidates.
Best rated
Only places with 3+ ratings — below that an average means nothing.
Average score over time
Ratings given per day
Score distribution
A healthy community leans right. A fat left tail means people are having bad encounters.
Lowest scores
Rated by real people, below 3. Not proof of anything — a reason to look. Click a name to open the account.
| Loading… |
Reports per day
Why people report · 30 days
Where it happened
How this works
Every report opens a chat between the reporter and the official Safety account, which thanks them instantly. You read the evidence here — including messages the sender deleted, kept for 24 hours — and your verdict lands in that chat. While a report is open, the conversation between the two people is frozen; any decision (verdict or dismiss) unfreezes it.
Dismiss closes a report without a verdict (false alarm, duplicate), optionally with a short line to the reporter. Investigating marks it as in progress. Warn sends a message from Safety to the reported person. Suspend puts a blocking screen on their app until you lift it; Ban disables the account. Everything you do is written to the audit log.
Flagged per day · 30 days
Categories · 30 days
Kinds
How this works
Every new or edited thought, bio, place thought and reply is read by Claude within seconds. Clean content leaves no trace here. Needs review stays until you decide: Remove takes it down and tells the author in their Safety chat, in their language; Keep closes the case; Keep & warn keeps it but sends the author a message you write; Dismiss closes it as noise (false alarm, duplicate, AI error). Removed was taken down automatically — Restore puts it back and apologises. Thoughts about self-harm are never removed automatically; they land here so a person can reach out.
AI error means the check itself failed (most often: the Anthropic credit ran out). The content stayed live, unchecked — decide it by hand and top up the credit.
What people write us
Every message here is someone who stopped to write. It deserves an answer. Click a name to open the account.
Send a notification
Rarely, and with a reason. A pointless notification means people who turn them off forever.
How it looks on a phone
Write on @random
A notification disappears. A message on the @random channel stays in the app — people who had notifications off still read it. Everyone gets the welcome message here 10 minutes after signing up.
Who read it
“Delivered” means the message is in their app. “Read” means they actually opened the chat afterwards.
“Having trouble logging in?”
Written from the login screen by people who are not signed in — the only place in the app that can be used without an account. They leave a phone or an email and a message. Reply from your mail app, then mark it handled.
Sent and failed per day · 7 days
By kind · 7 days
Why they failed
How this works
Every email Random sends is written as a document in the mail collection. The Trigger Email extension picks it up, hands it to the SMTP provider and writes back what happened. This page is that collection, live — so if somebody says “the code never arrived”, you can see within a second whether it left us or not.
“Delivered” here means Resend accepted it — not that it reached the inbox. What Gmail or Yahoo did with it afterwards shows in the Resend dashboard. Send again queues a new copy (a login code is not reused — the person asks for a new one in the app). Dismiss hides a failed mail you've dealt with.
Codes are masked here on purpose. A verification code is a key to somebody's account; it has no business sitting on a screen you might be sharing.
Visits per day
Download clicks per day
Time per visit
Counted only while the tab is visible and the person is active.
The film
Of the people who pressed play, how many were still watching at each quarter.
How far people scroll
Share of homepage visits that reached each section. The biggest drop is the section to fix first.
Which button gets the click
Download clicks by where the button sits. “Landing” is random-app.com/download.
Other actions
Where visits come from
The site that sent them, or the ?utm_source= tag on your links. “Direct” = typed, QR or app.
Countries
Country only. No city, no IP is ever stored.
Devices & pages
When people visit
Visits by hour of day, Romanian time. Post and advertise just before the tall bars.
Weekday × hour
Darker = more visits. The pattern tells you when a post or an ad lands on people who are actually looking.
What the website numbers say
Conversion by device
Download clicks ÷ visits, per device. Available from 1 Nov 2026.
Conversion by source
Which traffic actually downloads. A source with many visits and no downloads is the wrong audience — or the wrong promise.
Conversion by country
Engagement by source
Share of visits that stayed 10 s+ or clicked something.
Time per visit by device
Pages that convert
Downloads by the page the click happened on, and after the film.
Weekdays
Visits and downloads by day of the week, in this period.
New vs returning
Returning visitors who download are people who thought about it and came back — the strongest signal the site works.
Contact form — random-app.com/contact
Live. Every message also lands in hello@ (press@ for press). “Reply” opens your mail app with the address filled in.
Everything the dashboard did
Bans, suspensions, verdicts, dismissals, moderation decisions, warnings and messages — who did it, when, and why. Written by the server, read-only here.