The Psychology of Premium Apps — Transcript

Full transcript

1. 0:00Right now, most founders building with
2. 0:02AI have the same problem. Their product
3. 0:04looks decent, and so does everyone
4. 0:06else's. And that ultimately caps how
5. 0:09much you can charge for it. You see, the
6. 0:11brain has no premium detector, but it
7. 0:14does keep score quietly of every piece
8. 0:17of evidence that a product is good. And
9. 0:19there is one line every product has to
10. 0:21clear to feel premium. I call it the
11. 0:24commodity ceiling. You see, the problem
12. 0:26today is when you try a new product,
13. 0:28odds are high it's going to feel
14. 0:30familiar. You know, the AI look. And
15. 0:33honestly, a lot of it is pretty decent,
16. 0:35but also all of it is built with a tool
17. 0:38that everyone else has access to. So, in
18. 0:41this video, we'll explore how the brain
19. 0:42decides something is worth more, and how
20. 0:45to get your product past that ceiling in
21. 0:47an age where AI slop is everywhere. By
22. 0:50the way, I'm Tim. The last decade, I've
23. 0:52been designing interfaces for unicorns
24. 0:54like Spotify and dozens of other
25. 0:57companies. So, this is what I think
26. 0:59about daily. Today, a founder can plug
27. 1:02Marvin's MCP into their agent, pull from
28. 1:05every great design out there, and have
29. 1:07something decent in a day. That's why I
30. 1:09usually say that we're at that
31. 1:11minimalistic Bauhaus stage of user
32. 1:13interfaces. You know, those houses you
33. 1:16see everywhere now that all look the
34. 1:18same. Which is good for price-optimizing
35. 1:21consumers, but it's a problem if you
36. 1:23want to be the one they pay more for.
37. 1:25Because pretty and premium were never
38. 1:28the same. Packaging studies from 2010
39. 1:30and 2021 found that the look gets you
40. 1:33picked, but evidence of a good product
41. 1:35is what gets you called premium. Which
42. 1:37leaves one question. Where does that
43. 1:40evidence live in your product? From what
44. 1:42I've seen, it lives in three places, all
45. 1:45above the commodity ceiling. And the
46. 1:47first is the part AI leaves out. Most
47. 1:50founders assume that if the screens look
48. 1:52right, the product behind those screens
49. 1:55is mostly done. And with AI, the screens
50. 1:57really do look kind of right. But in
51. 2:00mid-2026, researchers gave the same
52. 2:03design tasks to five AI building tools,
53. 2:06like ChatGPT and Claude. The tools
54. 2:08explained their design choices really
55. 2:10confidently, but about a quarter of that
56. 2:13reasoning never made it into the
57. 2:15interface. And the parts they skipped
58. 2:17were the unglamorous ones, like showing
59. 2:19you what the system is doing, letting
60. 2:21you undo something, helping you recover
61. 2:24when an error happens. The researchers
62. 2:26even gave the gap its own name, design
63. 2:29theater, which costs your users a lot of
64. 2:32mental effort. Did the tap work? Can I
65. 2:35go back? And the brain keeps score of
66. 2:37exactly that kind of work. Decades of
67. 2:40research suggests that what's easy to
68. 2:42take in tends to get judged more kindly.
69. 2:45We saw this first hand with an app we
70. 2:47redesigned called Undertone. People
71. 2:49would test their mic, hit stop, and get
72. 2:52told it was too short to count as a
73. 2:54session. With nothing really to show for
74. 2:56it. One of the founders summed up how
75. 2:59that felt for their users.
76. 3:00>> I was expecting anything instead of
77. 3:03nothing.
78. 3:03>> And then, he said,
79. 3:04>> That right there is where we lose them.
80. 3:07>> We're going to come back to this app and
81. 3:08a second problem that we helped them
82. 3:11solve. By the way, if your product's
83. 3:13already growing and you want to take
84. 3:14that real step towards premium, like
85. 3:17Undertone did, I do a few free strategy
86. 3:19calls with founders every month. Links
87. 3:21down below. Now, you might be thinking,
88. 3:24won't AI tools just get better at this
89. 3:26stuff? And yes, of course, they will.
90. 3:29But whatever a tool learns, every
91. 3:31competitor gets the same day. So, it
92. 3:33just becomes the new floor right under
93. 3:36the commodity ceiling. What actually
94. 3:38saves yours is thinking through every
95. 3:40state of your product for your users.
96. 3:44That's one of the few ways your product
97. 3:46stays a relevant alternative to an AI
98. 3:48model company whose product can solve
99. 3:50similar jobs as well, or in some cases
100. 3:53better than your product. So, if
101. 3:55possible, start with this. Walk your
102. 3:58core flow and make sure your user always
103. 4:00knows what just happened, that they can
104. 4:03take it back, and that they know what to
105. 4:05do next. Just that hand-holding can get
106. 4:08you really far. Now, let's imagine you
107. 4:10map out the fixes to your core flow,
108. 4:13screen by screen. There's still a catch.
109. 4:15Fixing things screen by screen only gets
110. 4:18you so far because people use your app
111. 4:20for weeks or months or hopefully years
112. 4:23as one product, which collides a bit
113. 4:25with how most teams review their
114. 4:27product. Again, one screen at a time,
115. 4:30each one looking great as a happy case
116. 4:32test build or a clawed prototype. But
117. 4:35your user, that person who uses it
118. 4:37daily, sees screen 80 sitting right next
119. 4:40to screen one. And this is sneakier than
120. 4:43it sounds because people are notoriously
121. 4:45bad at telling you how much a product
122. 4:48slows them down. A 2009 study at the
123. 4:51University of York shows what that looks
124. 4:53like. Partly through a task on a
125. 4:55website, the researchers moved that
126. 4:57navigation bar from the left side to the
127. 5:00right, and nothing else on the page
128. 5:02changed. Time on that page more than
129. 5:04doubled, and even after moving on,
130. 5:06people kept going slower on the next
131. 5:08pages, too. And when the researchers
132. 5:11asked, most people had noticed the
133. 5:13change and still thought it didn't slow
134. 5:15them down much. When it actually did
135. 5:17slow them down a lot. So, if your
136. 5:19product has this kind of problem, it's
137. 5:21probably costing you a lot more than
138. 5:23your users will say out loud, or even
139. 5:25what your data will show. All of this
140. 5:28has been a problem for a while, but AI
141. 5:30supercharges it. Because when an agent
142. 5:32can spin up a new screen in a minute,
143. 5:35it's way easier to end up with 80
144. 5:37screens that don't quite agree. We saw
145. 5:40this in Undertone, too. When you run out
146. 5:42of sessions, the app handed you off to
147. 5:44one of two separate screens, both with
148. 5:47the same generic "You've used all your
149. 5:49sessions" message. No count and no date.
150. 5:52So, we designed one clear pattern for
151. 5:54both. A two out of two sessions used
152. 5:56badge, a line that says "You've used
153. 5:58both sessions this month" and the exact
154. 6:00date they refresh. On the surface, it's
155. 6:02a really simple change. Just a tiny bit
156. 6:04of copywriting and visual structure, but
157. 6:07it makes a world of difference at scale.
158. 6:09If your product has an actual edge, it
159. 6:12should trickle into hundreds of small
160. 6:15decisions like that. And no component
161. 6:17library or AI model can hand you those
162. 6:20out of the box. That's what people end
163. 6:22up paying more for. A wholesome product
164. 6:25that's thought through every state,
165. 6:27where all the connective tissue between
166. 6:29screens hold together, which makes the
167. 6:31whole thing feel like one product
168. 6:34someone cared about. So, if possible,
169. 6:36use your product for an hour like a real
170. 6:38user would and write down every spot
171. 6:41where it stops feeling like the same
172. 6:43product. Now, both places so far are
173. 6:45about the everyday moments in your
174. 6:47product. The last one is the grand
175. 6:49finale. It's where premium stops being
176. 6:52an opinion.
177. 6:54Now, let's rewind the clock a bit. Back
178. 6:56in 2014, Google did something legendary
179. 6:58in their Chrome browser. You lose your
180. 7:00connection and instead of a dead page,
181. 7:03you get a little pixel T-Rex. And if you
182. 7:05hit space, it turns into a game. Four
183. 7:08years after launch, people were playing
184. 7:10it around 270 million times a month.
185. 7:13Google took something that was a pure
186. 7:15failure and turned it into something
187. 7:18people actually enjoyed. This is a
188. 7:20hugely underrated premium driver in a
189. 7:23product. Think about the worst-case
190. 7:25moments in your product. The payment
191. 7:27that fails halfway, the moment someone's
192. 7:30genuinely worried something went wrong.
193. 7:32Error recovery like this is exactly
194. 7:35where AI tools struggled in the reports
195. 7:37I talked about before. And I think
196. 7:39they'll keep struggling here because
197. 7:41getting it right takes a real usage of
198. 7:43your product, testing it, and breaking
199. 7:45things. Getting it right even has a
200. 7:48name, the service recovery paradox. And
201. 7:50a big 2007 review found that when a
202. 7:53failure gets handled really well, people
203. 7:55can end up as happy as if nothing went
204. 7:58wrong, or even happier. In other words,
205. 8:01this has the potential to become the
206. 8:03moment someone goes, "Wow, this thing
207. 8:07went south, but they caught me." And
208. 8:09that, my friends, is exactly what a
209. 8:12premium experience feels like.
210. 8:15We changed that in a workshop with a
211. 8:17crypto card client, and support was a
212. 8:19big part of it. The person who handles
213. 8:21their support pointed out that people
214. 8:23are trusting the app with their own
215. 8:25money. So, they need to feel safe every
216. 8:28time they spend or deposit. And he
217. 8:30described a failed transaction like
218. 8:32this. There's just zero information on
219. 8:35what you should do. You're just left
220. 8:37alone. So, the question became, "How do
221. 8:39we make people feel confident in the
222. 8:41scariest moments?" What we did with them
223. 8:44was simple. When a payment gets
224. 8:46declined, you get an alert telling you
225. 8:48to go check your balance with a contact
226. 8:50support button right there. Tiny detail
227. 8:53that makes a world of difference. That,
228. 8:55along with a few other design moves,
229. 8:58grew their new accounts by 60%, and the
230. 9:00money people loaded onto their cards by
231. 9:0279%. But, it's not just payments.
232. 9:06Whatever your product is, most people
233. 9:08don't walk away from it because of a
234. 9:09home screen. A 2018 survey shows that it
235. 9:12only takes one bad experience for about
236. 9:15a third of people to be like, "Eff
237. 9:17this." So, if possible, open your
238. 9:19product and break it on purpose. Turn
239. 9:22off the Wi-Fi mid-flow. Fail a payment.
240. 9:24Empty every list of data. Then, design
241. 9:28those states as carefully as your
242. 9:30success screens, because a lot of teams
243. 9:32skip it, which makes it one of the
244. 9:34cheapest places to buy that feeling of
245. 9:36premium quality. Again, think of those
246. 9:38box houses. Every one of them looks
247. 9:41decent now. What gets you past the
248. 9:43commodity ceiling is what it's like to
249. 9:45live in it. And if your product's
250. 9:47already growing and you want help
251. 9:49getting it past the commodity ceiling,
252. 9:51we open up a few free design strategy
253. 9:53calls every month at Sisap. Links down
254. 9:55below. And if you want to see the whole
255. 9:57story of the app with a mic test, you'll
256. 9:59probably love this one here where I
257. 10:02break down how we in the agency took a
258. 10:04light up from kind of ordinary into that
259. 10:07premium realm. Now, until the next one,
260. 10:10have a great life.