Why it shipped first:

A platform like this starts with choosing someone – without a browsing layer there’s no entry point.

Solution:

Profiles as the core unit: photo, age, one line of character. We deliberately skipped algorithmic curation at MVP – first we needed to see who women pick on their own, and only then teach that to a system.

Effect:

70% reached a profile. 4–5 profiles viewed before the first chat, and about 5 more after it. 80% came back to Discover once they had started talking – not to replace anyone, but to browse the tabs and see who else was there.

Why it shipped first:

Not everyone is ready to talk to a stranger. We needed a way in that doesn’t require writing anything.

Solution:

A feed of their posts – a second surface next to private chats. The bet was that reading is easier than starting a conversation, and that interest can grow out of what someone says rather than how they look in a card.

Effect:

76% read the feed, 10+ posts per session. 58% moved from the feed into a profile – the feed worked as a second route to the same conversation.

Why it shipped first:

Everything else in the product exists to get here.

Solution:

A 1:1 conversation with no limits at MVP. We deliberately left it uncapped – first we had to see whether women would reach a conversation at all, not how fast they’d hit a wall.

Effect:

55% opened a chat. Only 25% wrote – and that gap defined the entire next phase of the product.

Trigger:

Women came back for the posts, but could only respond in DMs.

Solution:

Replies, threads and follows – a public layer with its own logic. Answering someone under a post is psychologically cheaper than writing into an empty chat: others are already talking, and the first word isn’t yours.

Effect:

83% of active users engaged with posts – writing their own or replying – with 10+ replies per post, and their D7 was 18% higher. The flip side: around 30% talked less to the man they’d been speaking with. But half of them found a new one – and for the first five days those conversations ran noticeably deeper.

Trigger:

They reached the chat and stopped. The empty input was the barrier.

Solution:

The man writes first. A letter on sign-up, letters waiting on his profile, and one more sent after she visits and says nothing – we tracked time on the profile and followed up with those who left in silence. If a woman doesn’t start the conversation herself, the product should start it for her.

Effect:

Chat activation 25% → 60%. More telling: 35% of started chats grew into deep conversations, against 10% in control. Letters didn’t just open the conversation – they changed how far it went.

Trigger:

A subscription asks for a decision before she knows what she’s paying for.

Solution:

Message limits and one-time packs alongside the subscription. The bet was that a first payment is easier when it’s small and reversible – the subscription can catch up once the value is obvious.

Effect:

First purchase conversion +34%. Subscriptions – weekly and monthly both – were bought around 1.5× less often, but 44% of pack buyers moved to one later and 13% came back for a second pack. Overall revenue grew 10%.

Trigger:

Interest in a single person fades, but women stop actively looking for a new one.

Solution:

Daily and weekly check-ins on how she’s doing. She gets a result about herself, we get a signal on who to suggest next. People take tests about themselves willingly, and it gave a reason to return that wasn’t tied to any one person.

Effect:

D30 18% → 22%, and 2 → 5 people messaged in 30 days. Half never came back to the check-in at all: the mechanic held the women it worked for, and it didn’t work for everyone.

Trigger:

The first payment was the hardest one to get.

Solution:

A time-limited offer right after sign-up, weekly or monthly. A discount at the entrance removes the “is this worth paying for” question at the moment she can’t answer it yet.

Effect:

+27% weekly first purchases, +11% monthly. Renewals held at 35–40% weekly and 25% monthly – in line with regular subscriptions: the offer brought more paying users without lowering their quality.

Trigger:

Conversations flatten once the obvious topics run out.

Solution:

Light prompts and games inside the chat that surface something new about each other. We weren’t trying to stretch the conversation artificially – the goal was to hand it a new topic exactly when the old ones ran dry.

Effect:

+20% time in chat, +18% average session length.