YT YouTube2Text

YouTube2Text → Transcripts → How Top 1% Apps Earn Your Second Visit

How Top 1% Apps Earn Your Second Visit — Transcript

by Tim Gabe · 3,478 words · 500 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00This app's onboarding completion rate
0:02beat the industry average. Still, half
0:04the people who finished it never once
0:06used it for the thing they downloaded it
0:08for, recording a therapy session. But
0:11buried in their own analytics was one
0:14moment every single week where users
0:16open the app on their own. The thing is,
0:19most products have a moment like that.
0:21[music] And just like Undertone, most of
0:23them don't take advantage of it. So, in
0:25this video, I'm going to walk you
0:26through the whole case study, how their
0:29own numbers pointed straight at that
0:31moment, and the seven things we built
0:33into it to bring people back. Every one
0:35of which you can steal for your own
0:37product. I'm Tim, by the way. I've been
0:39designing software for over a decade,
0:41for both big tech and for startups. And
0:44this is one of the core questions I dig
0:46into when we work with companies who
0:48want to improve retention. Where do your
0:51users already show up on their own?
0:56Now, the first job was working out what
0:58was actually going wrong. And on the
1:01surface, stuff looked pretty good.
1:03Again, when they came to us, their
1:05onboarding completion was way above
1:07average. People got in at a high rate,
1:10even though they had a long onboarding
1:12that required 10 plus minutes to get
1:15through. But once inside the app, users
1:17didn't stick. Here's what one of the
1:19founders said about their app. as well
1:22as the functionality of the app. We were
1:24referring to it as very utilitarian. It
1:26it did what it needed to do, but there
1:29was really not much beyond that. There
1:31wasn't much in the way of uh flare or
1:34panache, I'll say.
1:35>> So, on the first call with our team, I
1:38asked the obvious question. Why do you
1:40think the onboarding is doing so well?
1:42The answer had nothing to do with the
1:44onboarding. I think the reason though
1:46isn't necessarily because we have a good
1:49onboarding uh per se, but I think it's
1:52because we have just very high intent
1:55users that download an app like this
1:57because there just there just isn't
1:58necessarily another option, right? Look
2:00at your calories or you want to go to
2:02the gym, like a thousand to choose from.
2:04>> And it makes sense, doesn't it? People
2:06don't have that many options when it
2:08comes to therapy recording apps. If you
2:11download Undertone, you might have a
2:12session on Thursday and you need
2:15something, anything that you can bring
2:17with you for that session. So, a high
2:20intent user like this might make it
2:22through your long onboarding just
2:24because there's a sense of urgency and a
2:27lack of alternatives. So, that good
2:30onboarding number was essentially
2:31telling them how badly people needed the
2:34app that week. That's what it was
2:36measuring. And that tells you nothing
2:38about what happens once the session is
2:40done, which is the exact part we had to
2:43go and find
2:46to help us. Their team walked us through
2:48the data. So, double the amount of
2:50people that are kind of coming through,
2:53hitting the the mic test or whatever it
2:56is. It doesn't actually say mic test.
2:57So, they're going to the recording page,
2:58they're hitting record, they're trying
3:00something. Um, I went and measured the
3:03average time across all of these, uh,
3:06120 and it seems to only be around 20
3:08seconds. They're going in there, testing
3:10the mic, getting stop, being told that
3:12it's not long enough, and that it's not
3:15going to count as a session. 20 seconds
3:17of talking, a popup, and then a
3:19processing screen that spins for a
3:22couple seconds and gives up. This is one
3:24of the first touch points in the app and
3:26it's 100% going to determine how much
3:28you trust everything after it. One of
3:31the founders has his own number for what
3:33this dead end cost them. So, we were
3:36hard cutting people off and I think we
3:39we probably lost, you know, I couldn't
3:41put a number to it, but maybe 10% of
3:44people who downloaded the app and got
3:46into it because they tried to test
3:49something got nothing but a not quite an
3:52error, but a wall that says that's not
3:54enough. Uh, we need 10 minutes and and
3:57never came back.
3:57>> So, that's one problem. The first
3:59interaction with a core feature feeling
4:02kind of disappointing. But there's also
4:04a second one and it's actually the
4:06bigger of the two. Go back to why anyone
4:08downloads this thing. You've got therapy
4:11coming up and you want something to
4:13bring with you. That's the intent we
4:15just talked about, but your next session
4:17might be a week out and nothing happens
4:20in Undertone until therapy happens.
4:22>> You have to remember that the entire
4:24premise of our app is for forgetful
4:26people. Literally built for people that
4:28can't remember things. And so we
4:31shouldn't expect them to necessarily
4:32remember that they had downloaded an app
4:34a week ago.
4:36>> Now to be fair, most people are
4:38forgetful these days. There's just too
4:39much going on everywhere. So there's a
4:41very good chance the thing that
4:43activates your product sits outside of
4:46it on a calendar you don't control or a
4:48schedule you can't see. You can chase
4:51that with notifications and later in the
4:53video we do that properly. But the more
4:56valuable question came first. Was there
4:58any moment where the app could catch
5:00them on its own without asking for
5:03anything?
5:06One of the founders had already worked
5:08out the shape of this question and
5:10problem.
5:11>> It's very likely that some therapy
5:14attendees go once a month, right? So
5:16maybe they do download the app the day
5:18before therapy and they get in there and
5:20hey, they love it and then a month goes
5:23by and they've completely forgotten
5:26about it. And this makes sense. Nobody
5:28hangs out in a therapy app. It's not a
5:30social network. So, anything that
5:33depends on someone choosing to open it
5:36is tough. So, I asked where the
5:38analytics said people showed up on their
5:41own today. And there was exactly one
5:44place.
5:44>> I was actually asking Post Hog what the
5:47exact number of minutes is between when
5:49a session ends and when somebody is
5:51opening the uh the session summary.
5:54And it looks like uh people are doing it
5:57remarkably quickly. It's uh the median
5:59is about 10 minutes after the recording
6:01ends.
6:02>> 10 minutes. Fastest was six and the
6:04slowest 44. They walk out of therapy and
6:07they open the app on their own very
6:09quickly. This is a real payoff moment.
6:13>> You're waiting for your recap to be
6:14generated and then it's it's generated
6:17and you know it probably takes 3 to 5
6:19minutes to read through the session
6:21summary. And so I think that's kind of
6:24the obviously the most highest intent
6:26and highest activity period. Um it's I
6:30mean it's the biggest value delivery
6:31that exists in the app right now is is
6:34that's kind of the payoff.
6:35>> So this introduces another question.
6:37What do you actually do with a moment
6:40like that? This is the one time people
6:42show up on their own every week right
6:45after the most emotional hour of their
6:47week. And what was waiting for them in
6:50the old app was a block of text and a
6:52see full recap button. So the best
6:54moment in the product was also the
6:56weakest screen in it and it only pays
6:59off for people who are still around to
7:01see it. Half of them weren't. So we
7:04started at the other end and the first
7:06thing we changed was something most
7:08people wouldn't call a retention fix at
7:11all. Right, before we dig into that, if
7:13you're watching this thinking, "Hey,
7:15this is the kind of product approach you
7:17need for your app." I personally run a
7:19few free design strategy calls every
7:21month at Sips App. Links down below if
7:23it sounds interesting.
7:27All right, so the first thing we changed
7:28was the brand. That sounds like a detour
7:31when the problem is retention until you
7:34look at what it was doing to everything
7:36else. Every fix we were about to make,
7:39the onboarding, the record flow, the
7:41recap, all of it had to come out feeling
7:44like the same product. And you can't
7:46design how something feels until you
7:48know what it's supposed to look like
7:50when it's right. So, we settled the
7:52visual language first before touching a
7:54single flow. The Undertone team already
7:56knew where they wanted the brand to go.
7:59We we've really tried to keep things
8:03somewhere right on the the border of
8:06clinical and minimal. Um, you know, we
8:09don't want it to be too sterile that
8:10it's got no personality. But I think
8:13that having these sort of simple lines
8:15really really speaks to the brand itself
8:18and lets the brand shine.
8:19>> So that was the direction. What it meant
8:22in practice was a short list of rules.
8:24Pistachio and Sage against off-whites
8:27with a charcoal that never goes fully
8:29black. But a color palette isn't a look.
8:31They've been circling what they wanted
8:33for a while without being able to hand
8:36anyone a picture of it.
8:38>> I think that's the hard part that I
8:39can't do because, you know, I just don't
8:42know what that means. To be
8:43intentionally minimal, I think is like
8:45it takes some some serious skill. This
8:48is admittedly the hardest part about
8:50branding, helping a company find a
8:52visual representation that matches what
8:55they're seeing in their head while also
8:57making sense for their customer. In this
9:00case, that intentional minimalism they
9:02were seeking and couldn't put into
9:04words. So, we defined it, drew it, and
9:07shipped it as a solid brand refresh and
9:09design system. Think type, color, icons,
9:12components, button states, the whole
9:14foundation, which in turn frees you up
9:17to focus on the experience of the
9:19product. Starting with a leak at the
9:22record button. We knew the mic test was
9:24a clear friction in the app. We also
9:26knew that onboarding was the opposite.
9:29People flew through it. So, what if we
9:32move the mic test into the onboarding
9:34instead? you test your mic while you're
9:36still inside the flow that you're
9:38already highly motivated to finish. And
9:41when the founders saw it, they agreed.
9:43>> And particularly, I think moving the
9:46recording into the onboarding makes
9:48sense because it we'll see. Obviously,
9:52this is, you know, speculative, but the
9:55idea that someone can get that itch out
9:57of their mind to see what it does to
10:00record anything in the onboarding, um, I
10:04think that will go a long way to sort of
10:06proving the concept of, you know, okay,
10:09yes, it's going to record me and, you
10:11know, it's going to create some output
10:13based thereupon. Then of course the dead
10:16end itself. Instead of having a brick
10:18wall that punishes you for trying, you
10:21get shown a real recap before you've
10:23even gotten into the app. This is how
10:26the team saw it.
10:27>> And so I think having this sort of
10:30interactive demo that will quite
10:32literally show what it would look like
10:34if a therapy session passed through it
10:37really helps sort of underscore
10:39undertone
10:41the idea of what [laughter]
10:43what's going to happen. So by the time
10:45you land in the app, the two big
10:47questions for a user, does this work and
10:49what will it give me are both already
10:52answered. The leak at the record button
10:54closes here by moving it into the flow
10:56that already overperformed. So that's
10:58the first run fixed all the way from
11:00opening the app to a recording that
11:03actually works. But then therapy is
11:05over. The next one is a week from now
11:07and there's nothing to record and
11:09nothing to really come back for.
11:14Undertone's answer for those days was
11:16the classic notifications. But phones
11:19give you a lot more room than that
11:20nowadays, especially iPhones. Your
11:23product can be present without anyone
11:25opening it, and it can keep doing its
11:27job in the corner of someone's eye while
11:30they walk away from it. So, we looked at
11:33the dynamic island. While a session
11:35processes, the app sits at the top of
11:38your phone. If you walk to the car after
11:40your session, it's still with you. And
11:42in a case of a recording app, it also
11:44serves as a nice bug prevent.
11:46>> There has been a a few users now who
11:50have forgotten to come back into the app
11:53after their session and so they've ended
11:56up running four, five, 6hour long
11:59recordings. Um, and and this actually
12:03ended up glitching our system. It's not
12:06built to transcribe 6 hours of audio.
12:10Um, and so we unfortunately actually
12:13lost a couple sessions uh at the outset.
12:16>> Now, let's stop for a moment and think
12:18about what you just handed the app. An
12:20hour of the most private conversation
12:23you have all week. The user is typically
12:25in a pretty emotionally loaded state
12:28where privacy, handholding, and
12:30information disclosure is key. So, the
12:33island names each step while it works.
12:35Transcribing, anonymizing, summarizing.
12:38You can watch your own session being
12:40handled which makes the weight calming.
12:43We show the team a mockup of this idea.
12:45>> So when you first introduced it the
12:47immediate concern was I think one of the
12:49mockups you had given showed like some
12:52progress of the processing and we didn't
12:56have that capability built yet. as it
12:59moves through the stages of transcribing
13:02uh and and anonymizing and and
13:04summarizing, we've kind of created a bit
13:06of a progress bar. So, that actually now
13:08does fall into the realm of possibility.
13:11So, they built it and that's the whole
13:13path fixed. The way in, the first
13:15recording, the wait, the days in
13:17between. Every one of those was a place
13:20people were falling out. But none of
13:22them are the reason anybody comes back.
13:25That's one moment, and it's the one
13:27we've been building up towards since the
13:29start of this video. One moment in this
13:32app is worth more than all the others
13:34combined. It's the one where you walk
13:36out of that room, still carrying
13:38whatever just came up in your therapy
13:40session, and the app has to give you
13:42that hour back to you in a form that
13:44makes you want to do it again and again.
13:47That's a critical retention moment of
13:49this product.
13:52And to get this right, we had to go back
13:54to where the project started. On one of
13:56the first calls, I asked the team which
13:59companies they admire and why. Their
14:01answer came to shape the final design of
14:03the retention moment. The feeling that I
14:06get from Apple, that sort of neat, clean
14:08sophistication is actually probably
14:11found not even necessarily in their
14:14hardware, but like in their packaging,
14:17>> you know, when you get a laptop from
14:18them, how the lid comes off, how the,
14:21you know, uh, the the the plastic that
14:24it's wrapped in.
14:25>> Nobody asked about packaging. We were
14:28talking about an app. But there's a rule
14:30behind why that box works. The peak end
14:33rule. You remember an experience by its
14:35biggest moment and by how it ends.
14:38Apple's finish line is a cardboard box.
14:40Undertones is 10 minutes after walking
14:42out every week without anyone asking.
14:45That's the 30 seconds we had to win. So
14:48here are the seven things we built into
14:50it. One, hold the thread while they
14:53leave. Processing lives in the dynamic
14:55island. Two, bring them back
14:57deliberately with a notification when
14:59the recap is ready. Three, give it a
15:01surface worth opening. The recap arrives
15:04wrapped with a subtle bounce so the
15:06gesture is obvious before any
15:08instruction appears. Four, make them do
15:10it themselves. They pull the tape from
15:13top to bottom and it peels away. Instead
15:15of watching an animation, they're
15:17actually performing the reveal on their
15:20own, like opening a gift.
15:22>> But the idea of opening something and
15:24you know there's there's something here
15:26for you. open it up and and see what
15:28we've got for you. I think makes sense.
15:31Um I I think, you know, there's there's
15:35I don't want to say like it's a a gift
15:37cuz I think that's a little bit tacky,
15:39but it's like there's something there's
15:41something good in there. I'll say
15:42>> after that gift moment, we have five.
15:45Pay it off on two sensory channels.
15:47Confetti and a subtle haptic hit. Six.
15:50Frame what they just got. This is the
15:52first entry in a mental health record
15:54that nobody had before. So it gets
15:57treated like a real magical event. And
15:59seven, we then put the ask and the real
16:02retention lever inside the peak of this
16:05experience. The prompt to book the next
16:07session lives in the same moment as the
16:10engineered end. Here's the first time
16:12the team saw the unboxing event live on
16:14a call.
16:15>> I love this. I think this is really,
16:17really cool. I'm having an emotional
16:19design moment.
16:20>> I love that. I love that.
16:22>> This is very, very cool. I like how it
16:24kind of peels and falls off.
16:26>> I love that he called it emotional
16:28design because that's exactly what it
16:31is. Don Norman wrote a book about it. I
16:34got a full video on it on this channel
16:36and it's exactly what we try to build
16:38into moments like this. But it only
16:40lands if it fits the person using it.
16:42The brief had two edges. Nothing
16:45clinical because this is therapy.
16:47Nothing cute because this is therapy.
16:49That's why the recap can arrive wrapped
16:51and still feel serious and more
16:53importantly why it will always feel more
16:55right to their user than a non-tailored
16:58alternative. adding these layers and
17:01these polishes and these small
17:04interactive touches I think are the
17:07difference between I want to use
17:09undertone and I could use I mean I could
17:13audio record and put it in GPT or I
17:16could use granola or I could use any of
17:18these adjacent you know just at its core
17:21transcription tools but I want to use
17:23undertone because when I use undertone
17:26every session I get to open it up
17:27>> so here's the part I really want you to
17:29take away. Decent UX, decent UI, decent
17:32branding, those are all table stakes
17:34today. Anyone can ship a clean app. The
17:37components are free. Design patterns to
17:39copy are everywhere. And a model will
17:41hand you a pretty good flow in a couple
17:43of minutes. What none of that gives you
17:46though is connection, a personality, a
17:48thought through custommade experience.
17:51If Undertone sits on a shelf next to a
17:54bunch of other tools that turn audio
17:55into text, the reason someone picks them
17:58is going to be for those reasons. The
18:00sleeve, the raft recap, the 10 seconds
18:03after a session where the app seems to
18:05really understand them. That's
18:07deliberate craft, which is much harder
18:10to copy. we had something so simplistic
18:13to begin with that pretty much
18:15everything that you guys delivered was
18:17so far beyond what we were imagining
18:19that it it landed really well. Um I
18:23think really like if I were to say one
18:25thing and again it's a bit broad but
18:27it's the subtle touches the idea of this
18:30vinyl in the sleeve. I mean it's a
18:33record and our what we're sort of
18:36pitching undertone is your personal
18:38mental health record. the small touches
18:40that we didn't expect or know to
18:43describe to you. I think that's really
18:45those are the pieces that when we looked
18:47at it, it was like, yeah, this this
18:49feels cohesive. This has life. This has
18:52a personality.
18:53>> Now, the app is going to be out on the
18:55app stores anytime soon. Might even be
18:58when you're watching this. Just search
18:59Undertone Therapy Notes, and it's going
19:01to be the top results on iOS and
19:03Android. They're also raising money at
19:05the moment. So, if you invest in this
19:06kind of thing, there is a contact form
19:08on undertone.co that goes straight to
19:11the founders. Now, if you found this
19:13useful and you're about to put real
19:15money into a redesign, again, we open up
19:18a couple of free design strategy calls
19:20each month at SIPs, links down below.
19:23Also, if you want another video like
19:24this, I'm pretty sure you'll enjoy this
19:26one here somewhere about how the top 1%
19:29of apps keep users hooked. With that,
19:32until the next one, have a great

About this transcript

This page contains the full transcript of How Top 1% Apps Earn Your Second Visit by Tim Gabe, generated from the public captions YouTube serves with the video. The transcript has 3,478 words across 500 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.