Ex-YouTube Employee Explains the Algorithm — Transcript

Full transcript

1. 0:00I want to talk about your time as a
2. 0:02YouTube partner manager.
3. 0:03>> The algorithm is designed to read your
4. 0:05mind.
5. 0:05>> Is the community posts underused?
6. 0:08>> People will get freaked out, but this is
7. 0:09[music] true. This is the absolute worst
8. 0:11way to promote anything. Just don't do
9. 0:12it ever. Hi, I'm Nastia. I'm a content
10. 0:15strategist and the place I learned that
11. 0:18was YouTube headquarters because I
12. 0:20worked at YouTube working with largest
13. 0:22YouTube channels and companies on the
14. 0:23platform and helping them grow.
15. 0:25>> Okay, quick disclaimer with this
16. 0:26episode. We get really deep into inside
17. 0:29baseball on how to win with YouTube. So
18. 0:31if you're new here, it might be a little
19. 0:32bit overwhelming, but if you've been in
20. 0:34the game for a while, you might find it
21. 0:37very, very fruitful because I definitely
22. 0:38did. Nastia, one thing that really
23. 0:41drives me crazy is when I upload a video
24. 0:43thinking it's going to perform well and
25. 0:44it doesn't. My head always goes to why
26. 0:47didn't this perform? What do you do when
27. 0:50you hear that from one of your clients?
28. 0:53>> So what I do, like first calm people
29. 0:55down. It's stressful. I acknowledge it
30. 0:58and it also happened to me. It's like
31. 1:00very, very bad feeling. I would actually
32. 1:02go into YouTube analytics. And I would
33. 1:03do a few things. Thing number one, and
34. 1:05people going to hate me saying this, I
35. 1:07will not be checking 24-hour performance
36. 1:09of the video and I will not be checking
37. 1:1148-hour performance of the video. So I
38. 1:13will not be looking at real-time data.
39. 1:15Some people do that. We all want to
40. 1:17check real-time because we just
41. 1:18published and we care, but I'm not sure
42. 1:21everyone knows that like the real-time
43. 1:23data is there, but even YouTube
44. 1:24officially says this is the estimate.
45. 1:26>> Okay, so it's not completely accurate.
46. 1:28>> Yes. So okay, this is a bit technical,
47. 1:30but there is basically a data pipeline.
48. 1:32So the calculating views on YouTube is a
49. 1:34technical miracle. We all do not know,
50. 1:36but like technically calculating views
51. 1:38is like it's a very complex technical
52. 1:40challenge. So whenever a creator calls
53. 1:41me like tells the video is not
54. 1:43performing, unless it's news, you know,
55. 1:45unless it's something that you know
56. 1:46going to die in like an hour or like
57. 1:48tomorrow, I try to not look at it like
58. 1:50for the first two days. I'm like, let
59. 1:52just let it sit. Secondly, I would look
60. 1:53into data, the retention, how it is. Is
61. 1:55it lower? Is it higher? Do people click
62. 1:57on it more often? Do people click on it
63. 1:59less often? So, for example, I'll be
64. 2:00using my mom's channel. So, my mom my
65. 2:02mom like we're building channel she runs
66. 2:04a channel that is close to 700,000
67. 2:06subscribers. So, she's watched by women
68. 2:08like mostly 50 plus. And imagine you
69. 2:11know you have no idea about face yoga,
70. 2:13but you came in onto the channel and you
71. 2:14see two videos. One is doing great, the
72. 2:16other is doing bad. The impressions are
73. 2:18like instead of 10 million, it's like
74. 2:20maybe a million.
75. 2:20>> This is a problem that I run into all
76. 2:22the time. I'm looking at my impressions
77. 2:24saying why are my not getting the
78. 2:26impressions I think I deserve?
79. 2:27>> So, backstory. YouTube is a system. It's
80. 2:31a very smart AI. Now I can say this, but
81. 2:33trust me when I was saying this 4 years
82. 2:34ago, people will get freaked out, but
83. 2:36this is true. Like Google created one of
84. 2:38the oldest AIs in the world. Like a lot
85. 2:40of tech that we all use now is actually
86. 2:41created by Google. And YouTube is a
87. 2:43place that is basically the algorithm is
88. 2:45designed to read your mind, know what
89. 2:48you love the most, and like pull the
90. 2:49videos that you want to watch. The
91. 2:51reason the algorithm is there is because
92. 2:53YouTube is a commercial company. The
93. 2:55main reason Google makes money is
94. 2:56because it has ecosystem. So, it just
95. 2:58has a bunch of products that everyone
96. 2:59uses every day. And then because so many
97. 3:02people use it, you can like squeeze a
98. 3:04bit of money from like different parts,
99. 3:05different products. If people stop
100. 3:07liking like YouTube or YouTube videos,
101. 3:09they would leave to watch Netflix and
102. 3:11TikTok and like reels and everything.
103. 3:13And YouTube will be like in deep deep
104. 3:15trouble. So, with that when they you
105. 3:17know when people like the engineers,
106. 3:19they would sit into the room and think
107. 3:21about the ways to improve the algorithm,
108. 3:23they would ultimately try to improve it
109. 3:24to the way that Devin likes it the most.
110. 3:27So, the impressions is basically when
111. 3:29YouTube was accomplishing this task of
112. 3:31pulling specific videos for Devin to
113. 3:32love and like watch at 11 p.m. And and
114. 3:35then if YouTube does not give you
115. 3:36impressions, it's not because it doesn't
116. 3:38want to. It's because this is where it
117. 3:40thinks it's reasonable amount to give
118. 3:42and for some reason it does not have a
119. 3:44signal to get to move further. This is
120. 3:46something that feels out of our control.
121. 3:48But now think about my mom's video. You
122. 3:49know, imagine there is a video like
123. 3:51YouTube shows it to a bunch of people
124. 3:53and everyone clicks. Everyone is happy.
125. 3:55So YouTube is like, "Okay, we can get
126. 3:57riskier. We can throw it to the people
127. 3:58who like maybe care about the belly fat
128. 4:00but do not know my mom." And all the
129. 4:02people click. YouTube is like, "That's
130. 4:03fun." So it's like, "Let's get it even
131. 4:05riskier." So like imagine waves, it
132. 4:07tries to get as risk risk risk risk
133. 4:09riskier and riskier, getting this video
134. 4:11to a lot more people. But as much as
135. 4:13like it gets to a lot more people, it
136. 4:15comes to a lot less loyal people, right?
137. 4:18So at some point it will run out of
138. 4:19people who know my mom. At some point it
139. 4:21may run out of people who are actually
140. 4:23trying to lose belly fat.
141. 4:24>> So if I have a a piece of content and
142. 4:26the impressions are underperforming my
143. 4:28expectations, are you saying, "Oh, well
144. 4:30actually YouTube thinks your video sucks
145. 4:32when you think it's great." Is that the
146. 4:34crux of it?
147. 4:35>> So as long as YouTube pushes it further
148. 4:37and as long as like the next person
149. 4:38loves it, it will push it further. So
150. 4:40what it means if it does not get as many
151. 4:42impressions, it means that like pushed
152. 4:43it to a certain point and then the
153. 4:45people actually stopped reacting to it.
154. 4:47>> It's interesting though because there's
155. 4:49a specific case study that I'm thinking
156. 4:50of with a client of mine where we're
157. 4:53seeing a CTR of, you know, 7% or so,
158. 4:56which I would deem is pretty good for
159. 4:58educational content. Like I'm I'm
160. 5:00pleased with that. I mean, but the
161. 5:01impressions are quite low. So it's not
162. 5:02surprising that the the CTR is higher,
163. 5:04right? Your the core audience is
164. 5:06reacting more positively. So then the
165. 5:08average view duration for say a 15 to
166. 5:1020-minute video is roughly 7 minutes,
167. 5:12which I would deem is also pretty good.
168. 5:14I'm looking at those two core metrics
169. 5:15saying, "Okay, these meet my
170. 5:17expectations. This is a good video for
171. 5:19the existing audiences that loves this
172. 5:21content." Then I look at the impressions
173. 5:23and it's like, "Well, 10,000
174. 5:25impressions." And I'm like, "No, no, no,
175. 5:26the average view duration and the CTR
176. 5:28are suggesting this is worthy of many
177. 5:30many more than 10,000 impressions. So
178. 5:32why is it not doing it?"
179. 5:34>> That's a great question. That's a great
180. 5:35question. Now with Now we see a few
181. 5:36things. We look at the new viewers on
182. 5:38the video and the returning and like
183. 5:40casual. So basically YouTube allows you
184. 5:41to break down viewers into like few
185. 5:44groups, right? New and and like old
186. 5:46viewers. And when I was at YouTube, we
187. 5:48had like this feature available before
188. 5:50it was in the analytics. So we had a few
189. 5:52like, you know, metrics available. They
190. 5:54only later came to analytics. And I
191. 5:56could like have a look at those and you
192. 5:58you would have in in these cases, you
193. 6:01would have a very different like just a
194. 6:03lot smaller velocity among non-loyal
195. 6:05audience. In other words, the video can
196. 6:07get great, but YouTube just found like
197. 6:09great traction like very very loyal
198. 6:11people and like close to zero or like
199. 6:13very small traction in like very new
200. 6:15people. And I think the best way to
201. 6:16check it on the analytics would be just
202. 6:18to show the breakdown of new and
203. 6:19returning people on per video. And if we
204. 6:22saw like a very very small number of
205. 6:23new, it would just mean like there was
206. 6:24no traction there. And if you ask me, it
207. 6:26could be one of the reasons you just
208. 6:28just not have this traction beyond the
209. 6:29score people. But then, you know, like
210. 6:31there are few other things you could do.
211. 6:33First, you like I would still try to
212. 6:35assess if anything can be done. Like,
213. 6:37you know, the video is great, but maybe
214. 6:38we did not do the best job with the
215. 6:39packaging. Maybe we can change it and
216. 6:41give it a try. I would try to go and
217. 6:42compare it not only to the channel
218. 6:44average, but to a similar video on your
219. 6:46channel. So imagine you have like a few
220. 6:47content pillars. One of your content
221. 6:49pillar is like reviewing mascara. So
222. 6:52when your mascara video doesn't work,
223. 6:54find another mascara video that you've
224. 6:56done and try to compare those. And
225. 6:58what's important, YouTube allows you to
226. 6:59manually set the time frame. So that set
227. 7:02the time frame of comparing both of
228. 7:04these videos for each like first week of
229. 7:06their life, which will help me
230. 7:07understand how is the velocity of this
231. 7:09video. Bottom line, quite often I would
232. 7:11find out they're just like not as bad as
233. 7:12creator thinks.
234. 7:14>> I like that you use the word velocity.
235. 7:15That's a word that I haven't really
236. 7:17heard on YouTube a lot, but it's a good
237. 7:18way of measuring performance, I think.
238. 7:21Like how much velocity does it have in a
239. 7:23certain window of time? That's really
240. 7:24interesting to me.
241. 7:25>> It It's like sometimes, you know, we
242. 7:27would I don't know. It's like very
243. 7:28commonly that you would look at someone
244. 7:30else's channel. Oh, they have 700
245. 7:32700,000 like a million views in each
246. 7:35episode. And my episode is like, I don't
247. 7:36know, 50K. But then like maybe they had
248. 7:39like a year to get to this million
249. 7:40views. Or maybe they publish once a
250. 7:42quarter and you publish daily. So I just
251. 7:45like all also want to make sure all the
252. 7:47metrics are in the place to compare and
253. 7:49then I would compare and I would just
254. 7:50dive in like what actually lagged. And
255. 7:52quite often you would find what was
256. 7:54there. The audience maybe reacted to not
257. 7:56as much as you wanted it them to react
258. 7:57to.
259. 7:58>> Maybe you want to do a few community
260. 7:59posts.
261. 8:00>> Oh, interesting. To promote the video?
262. 8:02>> Not to don't promote it. Don't promote
263. 8:03it. So the way you should be doing
264. 8:05community posts is not to promote it in
265. 8:07the sense like here is my video watch
266. 8:08it. Like this is the absolute worst way
267. 8:10to promote anything. Just don't do it
268. 8:11ever. Because if you will be like I saw
269. 8:13it anyways. Don't don't like throw it
270. 8:15the video at me in two places. But
271. 8:17usually what happens, you know, your
272. 8:18video is like you said 7-minute
273. 8:20retention. So probably was like I don't
274. 8:22know 10 13 minutes. So there is a ton of
275. 8:24value in this video that you were unable
276. 8:26to put in a thumbnail because the
277. 8:27thumbnail has that much real estate. But
278. 8:29this video probably has something else.
279. 8:31So just try to get this extra content in
280. 8:33like a carousel. I think a really smart
281. 8:35way is to make like an Instagram like
282. 8:37carousel about the subject.
283. 8:38>> Is the community post underused or not
284. 8:42really worth a lot of energy?
285. 8:44>> So underused and like you get them you
286. 8:46get them in a very fresh position.
287. 8:48Literally today I sent a message to the
288. 8:50client that out of let's say there is
289. 8:53like a time frame
290. 8:55certain time frame we have 10 million
291. 8:57impressions on the channel 1 million
292. 8:59comes from
293. 9:00community posts.
294. 9:01>> Really?
295. 9:02>> 10%. Yeah. I think this is the most
296. 9:06underused real estate in all YouTube.
297. 9:08This is something I'm learning really
298. 9:10recently.
299. 9:11>> I'm fascinated about community posts.
300. 9:13It's something that I think is
301. 9:14definitely worth doing. There is a term
302. 9:16that has become commonplace in YouTube
303. 9:19education called water falling and it's
304. 9:21when you take an existing piece of
305. 9:22long-form media and you take portions of
306. 9:25it and you turn them into carousels and
307. 9:27Twitter posts and LinkedIn posts and
308. 9:28suddenly this one piece of work that
309. 9:30you've put a lot of blood sweat and
310. 9:31tears into has become 15 different
311. 9:34assets. And the team that I work with
312. 9:36we've gotten to the point of turning
313. 9:37them into carousels and posting them on
314. 9:39Instagram, but we forget about YouTube
315. 9:41community posts. So, it's a shame not to
316. 9:43just simply move it there.
317. 9:45>> I agree. Yes, a community post is like
318. 9:48the unit of content in itself. YouTube
319. 9:50does promote it. You do see it on your
320. 9:52home page. So, basically YouTube
321. 9:54distributes it. And this is something
322. 9:56very new because I discovered it like 1
323. 9:58hour ago. I was going through my feed
324. 10:00and I found a bunch of community posts
325. 10:02that are literally viral. They had a few
326. 10:05million uh replies to um quiz. So, I
327. 10:07think one, we should scope it to people
328. 10:10who watch us on a home page and like
329. 10:12maybe random as well. Second, yes, like
330. 10:14making a carousel. So, basically getting
331. 10:16value out of the post. You know, if the
332. 10:18post is about this is the finance of AI.
333. 10:21So, this is the example. Did you know
334. 10:23that like Entropic will cost $1 trillion
335. 10:25once it's on IPO, which is basically
336. 10:27close to Google. If you a bunch of
337. 10:29things, maybe something simpler. So, you
338. 10:31put it out as like a carousel, as like a
339. 10:33post. What works really great, which is
340. 10:35really simple, it's not really a
341. 10:36carousel, but just one slide. So, photo
342. 10:38and text. This is like super basic
343. 10:40thing. We just try not to crowd the text
344. 10:42as much as like have a photo like a hook
345. 10:45on it.
346. 10:46>> What what kind of conversion is it
347. 10:47bringing though? Is it bringing
348. 10:49subscribers? Is it bringing views?
349. 10:51>> It does. So, it all depends on
350. 10:53subscribers. So, it does bring and I saw
351. 10:56a channel, though I have to make a
352. 10:58disclaimer. It was a news channel. So, I
353. 11:00think if you're a news channel, this is
354. 11:01literally your separate news feed like
355. 11:03Reuters. And I saw a news channel that
356. 11:04would get like certain million
357. 11:06impressions in like a month from there,
358. 11:07which was sometimes close to their
359. 11:09actual content.
360. 11:10>> One thing that I have been thinking
361. 11:11about lately has been what kind of
362. 11:13subscriber do I want for my clients?
363. 11:16People who subscribe because of your
364. 11:18shorts, I don't know if they would cross
365. 11:20over and watch your young your your long
366. 11:22forms as religiously as someone who
367. 11:24subscribed watching your long forms. And
368. 11:26so, when we're talking about community
369. 11:27posts, my question now becomes, okay,
370. 11:29let's say someone actually does like
371. 11:30your community post and they subscribe,
372. 11:32are they going to actually watch the
373. 11:33long forms or are they really just
374. 11:35looking for more community posts?
375. 11:37>> My hot take is that you don't like
376. 11:39splitting the subscribers into good and
377. 11:42bad ones. It's just something that I
378. 11:43want and someone I don't want. I want a
379. 11:45subscriber.
380. 11:46>> No matter what.
381. 11:47>> And I'll tell you why. So on YouTube, a
382. 11:49subscriber who is here but doesn't watch
383. 11:51you cost nothing. It does not harm your
384. 11:53channel. So I think there is like a myth
385. 11:56that comes from Instagram and more like
386. 11:58from feed. You know, if you have like
387. 11:59people who don't watch you kind of harms
388. 12:01you, not the case. So why not have them?
389. 12:03I mean, why why do you want to close the
390. 12:04door?
391. 12:05>> I mean, that's fair. That's fair. I do
392. 12:07see some channels that have like 750,000
393. 12:10subscribers and then you look at their
394. 12:12latest long form video and it's like
395. 12:145,000 views and you're like, oh this
396. 12:16person is a shorts creator or oh this
397. 12:18person and maybe it's more of a social
398. 12:20thing, you know, maybe it's just like
399. 12:22social judgment than anything that's
400. 12:23actually important.
401. 12:24>> Or, you know, what? How many channels
402. 12:26you roughly subscribe on YouTube?
403. 12:29>> I don't know. Oof. Probably 500.
404. 12:32>> Yeah, same. Do you watch all of those?
405. 12:34>> No, no, no, of course not.
406. 12:35>> So I think this is where it's like
407. 12:36absolutely okay when YouTube says that.
408. 12:38It's just like the nature of how YouTube
409. 12:40built the subscription function. YouTube
410. 12:42is great for like helping you subscribe
411. 12:45and it's quite bad, which is good for
412. 12:46us, for helping you unsubscribe.
413. 12:48>> [laughter]
414. 12:49>> Okay.
415. 12:50>> So with that, it's like a very natural
416. 12:52feature of the platform to accumulate
417. 12:53subscribers who don't watch you, if you
418. 12:55think about it. I don't think it was a
419. 12:56goal of the platform. I just think it's
420. 12:57just like, you know, then then
421. 12:59consequence. So every channel would have
422. 13:00like a ton of subscribers who don't
423. 13:02watch it and if you think about the age
424. 13:04of YouTube, you know, it's like old.
425. 13:05>> Yeah, channels that have been around for
426. 13:0720 years may have a million inactive
427. 13:10subscribers, which is crazy.
428. 13:12>> But also validly so, you know, maybe
429. 13:13they became big doing I know they were
430. 13:15teenagers, they were doing pranks and
431. 13:17like today they're like serious people.
432. 13:20I know it's like I saw such a person
433. 13:21like recently. They did very actually
434. 13:24funny hilarious videos. Like 20 years
435. 13:26ago this was like super big channel, but
436. 13:27now this person like very serious
437. 13:29creative director for an eye company.
438. 13:31You know, they do certain things, but
439. 13:32just like they grew and their content
440. 13:35grew and they don't necessarily want to
441. 13:36be that person back in time. And
442. 13:38actually might a lot of creators at
443. 13:39YouTube like those OG's, you know, who
444. 13:41just grew and developed and they want to
445. 13:43do something different.
446. 13:44>> What do you think about people who are
447. 13:46hot swapping packaging in the first day,
448. 13:49in the first few hours? What's the best
449. 13:51practice?
450. 13:52>> If your content is super high stakes,
451. 13:54like super expensive, you know, you
452. 13:55publish once a month, you put like a ton
453. 13:57of money in there, I do think it
454. 13:58actually can make sense. Because like
455. 14:00every second is there, given that you
456. 14:02like have the data and like the
457. 14:04intuition and I think every listener of
458. 14:06you would have it. So, I don't say no, I
459. 14:08actually allow it and like some people
460. 14:10working in my team, they do it. At the
461. 14:12same time, if you're like publishing
462. 14:14more frequently, you basically do it
463. 14:16like weekly and it's like lower stakes,
464. 14:18it's not like the big movie. It's a bit
465. 14:20of like lower stake. I think it's also
466. 14:22hot like a very hot take. People won't
467. 14:24like it. I would rather analyze and
468. 14:26focus on my next episode.
469. 14:28>> I agree with that actually. Yeah, yeah.
470. 14:30>> I actually became more flexible when I
471. 14:32left YouTube on this. But I when I was
472. 14:34at YouTube, that was like the only just
473. 14:35focus on the next episode, just focus on
474. 14:37the next. I think what always is putting
475. 14:38a manager as we look at like a catalog,
476. 14:40this is what we learned. So now we take
477. 14:42this learning and apply it further
478. 14:44instead of like trying to obsessively
479. 14:46rescue the past stuff. I think there is
480. 14:48a lot of money I would put into the
481. 14:49episode and then I would obsessively
482. 14:51change the thumbnail every minute. I do
483. 14:52think that such amount of money exists.
484. 14:54>> Yeah. I hear you. I want to talk about
485. 14:56your time as a YouTube partner manager
486. 14:59and I want to get kind of like the
487. 15:01inside scoop on how you were trained.
488. 15:04What were the things that YouTube said
489. 15:06are really important to you as a YouTube
490. 15:08partner manager?
491. 15:09>> Good question. So first of all, my
492. 15:10training was mostly going through actual
493. 15:12cases and speaking to colleagues and
494. 15:14like just I was hired, I did not really
495. 15:16have a content strategy expertise and I
496. 15:17do have a finance background. A finance
497. 15:19and a content person. But I just left
498. 15:21YouTube. So they asked me a few
499. 15:22questions and like you don't have to be
500. 15:24confident strategies to get this job, at
501. 15:26least when I was getting it. Then, I got
502. 15:27like a a ton of decks to read. This is a
503. 15:29must-read and there are your creators
504. 15:31and you talk to them. That's it. Very
505. 15:33scary because normally I would receive a
506. 15:35question from a creator like yours,
507. 15:37"Hey, why this feature doesn't work?"
508. 15:38And it could be like a very simple
509. 15:39question, but I was like, "You know
510. 15:41what? This is a very hard question. I'll
511. 15:42get back to you." I would literally
512. 15:43call, get to my colleague um
513. 15:46Olga Vadim. Most of my training came
514. 15:48through this. However, we did have a few
515. 15:50kind of moments and ideas. One of those
516. 15:53was like data. So, I think you saw like
517. 15:55a process call. I know that a lot of
518. 15:57like YouTube is a creative place. And I
519. 15:59would say as much as a creative place,
520. 16:00we're looking at data. So, like a
521. 16:02creator comes to me, they like the
522. 16:03episode like you mentioned, "You put
523. 16:04this episode, 7% CTR." But like to me
524. 16:07the data shows there are no impressions.
525. 16:09So, it's like, "I know you care, but it
526. 16:11just like it was not audience did not
527. 16:13like it." So, I would be trained to look
528. 16:15at like pure data. And I think this is
529. 16:18really kind of um helpful to just
530. 16:20analyze video more adequately. I think
531. 16:23most of my job, finding patterns on this
532. 16:25channel. So, instead of like, "This is
533. 16:27the blueprint for everyone." You don't
534. 16:29have this blueprint. So, YouTube is not
535. 16:31trying to find a blueprint. Instead,
536. 16:33YouTube, you know, has teams who like
537. 16:35obsessively study what already happened
538. 16:36on YouTube in the past and they're just
539. 16:38synthesize the data. And I think this is
540. 16:40the only way to learn internally as
541. 16:41well. So, if I would work with the
542. 16:42individual channel, I would go into the
543. 16:44channel like, "This worked for you. This
544. 16:45does not." So, basically what worked, do
545. 16:47more. Which is like
546. 16:48>> Sure.
547. 16:49>> such a simple formula, but you know, you
548. 16:51can branch it out into so many
549. 16:52approaches. But, if you ask me which is
550. 16:54one one formula, it's just like always
551. 16:55trying to find this Pareto principle of
552. 16:57content that does the most job and just
553. 17:00doubling down on it on your channel and
554. 17:02other channels. And if something doesn't
555. 17:03work, well, let's see. Do we want to fix
556. 17:05it or maybe we just don't want to like
557. 17:07spend time and money there and just
558. 17:09let's cut it out and just reinvest the
559. 17:11energy. And I would say that most of my
560. 17:13job was like actually trying just to
561. 17:14find the spot, this area to invest
562. 17:16energy into.
563. 17:17>> Yeah, I mean that's a great point
564. 17:18because you you do have limited energy.
565. 17:20And so you can spend your energy, you
566. 17:22know, improving your back catalog by
567. 17:24updating the thumbnail every 24 hours,
568. 17:26or you can focus on the next video. And
569. 17:28I do think it's helpful to have somebody
570. 17:30who's third-party stepping in, analyzing
571. 17:32the channel as a whole, and kind of
572. 17:34making that guidance. I do find that to
573. 17:36be pretty valuable. What what kind of
574. 17:37partner manager actually do for a
575. 17:39channel, and what is off-limits?
576. 17:42>> What a partner manager does for a
577. 17:43channel is like first of all it's your
578. 17:44POC, it's YouTube. So it's basically its
579. 17:46job to make you a happy creator who is
580. 17:48loved by its platform. So this is what I
581. 17:50would say. So I think that's number one.
582. 17:52You're basically here like a real human
583. 17:54face, which I think a real big strength
584. 17:56of YouTube compared to other platforms.
585. 17:57Not sure how it changed, but I think
586. 17:59it's still like it had it has like the
587. 18:01probably like really good human touch.
588. 18:03Second, um when I was at the team, the
589. 18:05focus of my team I would always say like
590. 18:07we are a McKinsey for YouTube channels.
591. 18:09I was saying about this, like a
592. 18:10consulting company. So basically the
593. 18:12channel doesn't work or the revenue
594. 18:13doesn't work for a creator. So my job
595. 18:15was like to make views and revenues for
596. 18:17them. And like tell them what to improve
597. 18:20to make it better. So this is what we
598. 18:21were focused on our job. However, like
599. 18:24you don't have like a target. Like I did
600. 18:25not have like to make you make more
601. 18:27money, or like to make you make more
602. 18:28views, or like actually to like, you
603. 18:30know, cuz it depends on so many factors.
604. 18:32I could advise, but I could not enforce.
605. 18:34This is one thing off-limits. Also
606. 18:36what's off-limits is they get into much
607. 18:37into production process. Because I think
608. 18:39as YouTube, um you want to remain a
609. 18:42platform. With this like a partner
610. 18:43manager is a part of this platform, it
611. 18:45just wants you to use the platform the
612. 18:46best, explain how it works, but it would
613. 18:49they will not necessarily go into your
614. 18:50creative process or change your content,
615. 18:52or give you creative ideation. Because
616. 18:53this is something that creators do not
617. 18:55like the partner managers. So we'd give
618. 18:56you data-driven hints. We would give you
619. 18:58data-driven hints. We would tell you
620. 19:00what works on the platform for other
621. 19:01people.
622. 19:02>> Is there any incentive on YouTube's end
623. 19:05to encourage people to try new features?
624. 19:08>> I do think this is another pillar of
625. 19:10partner manager job as because YouTube
626. 19:13is a product, so they want you to know
627. 19:15about features and when a feature they
628. 19:18think suits you, they want you to use
629. 19:20it, to try it. But then like of course
630. 19:22like something you mentioned, I think I
631. 19:24just want to clarify for everyone, is
632. 19:25that if you use a feature, people think
633. 19:27like YouTube, like any other platform
634. 19:30will kind of give you impressions. It's
635. 19:31like almost if people use it, we'll like
636. 19:34kind of prioritize them, which is not
637. 19:36true.
638. 19:36>> Right.
639. 19:37>> Because if we get back, if we get back
640. 19:39to these people, you know, in the deck
641. 19:41like Todd and everyone who give you
642. 19:43impressions, ultimately they are
643. 19:45incentivized to like just make a YouTube
644. 19:47a lot more watchable and fun place for
645. 19:49you as a viewer. So naturally, you know,
646. 19:51they would not want to be in a position
647. 19:52where like there is like content that is
648. 19:54like
649. 19:54not really good, but uses all the
650. 19:56features. They would not really want to
651. 19:58promote it additionally because they
652. 19:59would like it would be detrimental to
653. 20:01the platform.
654. 20:01>> I think I felt guilty of that. I fell
655. 20:04victim of that when I had a partner
656. 20:05manager. I really truly thought it was
657. 20:07like, oh, great. YouTube is
658. 20:09acknowledging that we're doing really
659. 20:10well and as a result they've now given
660. 20:12us an advisor. I thought it was like,
661. 20:14okay, if I try all these new features,
662. 20:16if I listen to what they say, that's
663. 20:17that's good, but it's actually just
664. 20:18someone advising you based on the data.
665. 20:20>> You know, YouTube tests features, you
666. 20:22know, there is a feature and before it
667. 20:24comes to like very wide use like to
668. 20:26everyone, YouTube would test it out. For
669. 20:28example, they roll it out to like tiny
670. 20:29tiny bit of creators, it may roll it out
671. 20:32to you and like usually if you have a
672. 20:33partner manager, a partner manager will
673. 20:35be like your bridge to have it rolled
674. 20:37out for you quicker because of this they
675. 20:38can collect feedback from you. So I do
676. 20:40think there is a piece of some a bit of
677. 20:42like exclusive access that you have, but
678. 20:44on the other hand, I think what's
679. 20:45important because so many people don't
680. 20:47have partner managers and they cannot
681. 20:48control it because you cannot like apply
682. 20:50for a partner manager. And as a partner
683. 20:52manager, I've like I'm I don't have any
684. 20:54decision rights.
685. 20:55>> Yeah, how do you get selected or how do
686. 20:57you like what if your team wants one?
687. 20:59>> So as a partner manager, I did not have
688. 21:01any like you just receive. You know,
689. 21:03usually these are larger channels than a
690. 21:04smaller channels, but YouTube does not
691. 21:06really disclose all the parameters. But
692. 21:08I think with that like my important
693. 21:10message and I truly think it's important
694. 21:12because I've been on both sides. Having
695. 21:13partner manager does not make you grow
696. 21:15faster.
697. 21:16>> What does it do?
698. 21:17>> I think just people think that if you
699. 21:19know person at YouTube like suddenly
700. 21:20like the unlock for growth, you know,
701. 21:23it's just not like something, you know,
702. 21:24like super secretive, something
703. 21:26inaccessible, something you cannot get
704. 21:28yours. It's just helpful, but I don't
705. 21:30want to say this like a parameter that
706. 21:32decides if you go faster or not. I've
707. 21:34met so many big creators who do not
708. 21:35speak to their partner managers.
709. 21:37>> I mean, I've heard that also. At the end
710. 21:39of the day it just sounds like you're
711. 21:40just there as a third-party advisor
712. 21:41based on the data in your YouTube
713. 21:43studio. What is your sentiment around
714. 21:46YouTube in the year 2026?
715. 21:49>> I started eight years ago. It's a very
716. 21:51different world. Eight years ago YouTube
717. 21:53was having these goals. We want to be
718. 21:55this like, you know, we want to be It
719. 21:57would not say it explicitly, but we
720. 21:58wanted like YouTube wanted to be like
721. 22:00place where every major advertiser come.
722. 22:02YouTube wanted to be the place where
723. 22:03like, you know, all the people come, all
724. 22:05the main events come, and like gosh,
725. 22:07this company got everything it wanted.
726. 22:09Bottom line now when I see YouTube sets
727. 22:11a goal, I'm a lot more
728. 22:13I'm a lot I'm listening to this goal a
729. 22:15lot more attentively cuz I do have
730. 22:16belief
731. 22:17it does have resources to achieve it. So
732. 22:19I'm excited, but I think it's an adult
733. 22:21like it's very grown-up platform with
734. 22:23all the media and like of course in the
735. 22:25marketing it wants to like replace your
736. 22:27TV, you know, we have all the events, we
737. 22:29have all the big creators, we have all
738. 22:30the big shows. If you ask me where I'm
739. 22:32really thinking hard, there's been an AI
740. 22:34announcement of like ask YouTube
741. 22:36feature.
742. 22:37>> This is the AI tool that you can now
743. 22:39engage with across the platform, yeah?
744. 22:41>> Yeah. So I think there's like content
745. 22:42and AI is like a big important part of
746. 22:45boiling water and like everyone is
747. 22:46uncertain. So one thing everyone was
748. 22:48uncertain like YouTube blocked a ton of
749. 22:50AI channels, AI slop. So I think like
750. 22:52one YouTube fights AI slop and we've
751. 22:54seen it before with different channels.
752. 22:55I think it's going to be like a lot more
753. 22:57human-driven and want quality content,
754. 22:59so I don't think AI slop it be an issue.
755. 23:01But where uh my thoughts are and I
756. 23:02didn't know what's going to happen is
757. 23:03like ask AI basically, you know, makes
758. 23:06people talk to YouTube. You know, they
759. 23:07have like this interesting real like
760. 23:10show real where they like how to teach
761. 23:12my 3-year-old ride a bike. And then it
762. 23:14gets you text and like a few time codes
763. 23:17and that's it. And I'm like there are so
764. 23:18many channels who like will answer. But
765. 23:21also watch time is important.
766. 23:23And also you want people just to stay
767. 23:24there like watch for long. And like
768. 23:26basically this feature, it's not yet
769. 23:28there. It's coming summer. But I do
770. 23:30think it breaks so many patterns you and
771. 23:33me are used to. I don't know how much
772. 23:36traffic it's going to have. I don't know
773. 23:37what going to be the parameters. I will
774. 23:39be paying lots of attention to it
775. 23:40because I don't know where it's going to
776. 23:41go.
777. 23:42>> Yeah, there's some serious disruption
778. 23:43happening and about to happen on our
779. 23:46little platform. And it's scary to think
780. 23:49that the platform itself is making such
781. 23:52dramatic changes to impact search in the
782. 23:56the way that you just described. You
783. 23:57know, there are millions of channels
784. 23:59probably are hundreds of thousands of
785. 24:00channels that depend on search. I don't
786. 24:02know if they've been wiped out or if
787. 24:04they're going to seriously need to
788. 24:05modify their strategy because, you know,
789. 24:07YouTube has just decided, "Hey, this is
790. 24:09what we're going to do." So it can be
791. 24:10daunting.
792. 24:11>> Maybe they're going to be elevated. We
793. 24:12don't know. Like, you know, I would
794. 24:13imagine I would imagine like you would
795. 24:15have like a strategy, you know, how a
796. 24:17strategy of like search up to my own
797. 24:18channel, you know, this like show up to
798. 24:20my own channel. Maybe you have a search
799. 24:21strategy of like geo optimized channel
800. 24:23where we will find a balance. I'm sure
801. 24:25people will find a way to like use it
802. 24:27and for their benefit. But if if if
803. 24:29there were times like I would be last
804. 24:32least relaxed as a content strategist.
805. 24:35>> [laughter]
806. 24:35>> I think they are
807. 24:36I think they are today and coming.
808. 24:39>> [laughter]
809. 24:41>> Fair enough. Yeah, I feel very uncertain
810. 24:43as well. And you know, I think the last
811. 24:45couple of years as a strategist myself,
812. 24:48I have a certain degree of confidence
813. 24:50that I know what's going on. And that
814. 24:51has really been flipped on its head this
815. 24:52year. So I am kind of uneasy about it.
816. 24:55I'm very curious to see what what what
817. 24:56happens. But the one takeaway that I
818. 24:58learned from your most recent response
819. 24:59was you've seen YouTube 8 years ago
820. 25:02forecast what it wants to be today, and
821. 25:04because of that, you have confidence
822. 25:06that when they say they're going to do
823. 25:07something, they're going to achieve it.
824. 25:09And so, I am also going to adopt that
825. 25:11mentality because if they say something,
826. 25:13the likelihood of them achieving it
827. 25:14sounds very very high, and it just means
828. 25:16we have to change with it, I guess.
829. 25:18>> I think one last takeaway that I really
830. 25:20want people to understand, and I think
831. 25:22that was a pain for me inside YouTube,
832. 25:25is I don't want to like sometimes as a
833. 25:27creator we want to see ourselves like we
834. 25:29are the creator, here is the YouTube, we
835. 25:31are like all we are not animus, but we
836. 25:33would disagree in things, and sometimes
837. 25:34we would have an argument, which would
838. 25:36sometimes make you feel like you are
839. 25:38against YouTube cuz YouTube isn't
840. 25:40getting you impressions or something.
841. 25:41Like we all been in this emotional, at
842. 25:43least for at least a minute everyone
843. 25:45felt it. But, I think my perspective as
844. 25:47a product manager, and when I see you
845. 25:48this was also happening with Shorts.
846. 25:50When Shorts were introduced, a few
847. 25:52people were happy, but ex- so many
848. 25:54long-form creators were unhappy because
849. 25:56they felt like, you know, YouTube is
850. 25:57taking. This was ours, we've put so much
851. 26:00work into it, now you want us to stop
852. 26:02doing this. But, if you think of it from
853. 26:03a business perspective, think what would
854. 26:05happen to YouTube if they would not
855. 26:06introduce Shorts. They would be eaten.
856. 26:08So, as a big business, as a big tech, as
857. 26:11Google, um you're always in this
858. 26:13position where you want to make creators
859. 26:14happy because they're like the most
860. 26:15important part of your ecosystem, but
861. 26:17you do have like what called market
862. 26:18pressure. So, if people is like using
863. 26:20AI, even now already had creators, I'll
864. 26:22tell you, who lost all their traffic
865. 26:25because they were like their videos were
866. 26:26based like very simple questions. If you
867. 26:29like, what to do if I today have a high
868. 26:31fever? Like YouTube has this way to like
869. 26:34house, but you would go to ChatGPT now.
870. 26:36So, I already seen the wave of typical
871. 26:38problem-solving videos to have like no
872. 26:40traffic because like you would go away.
873. 26:43So, with that what I want to say, like
874. 26:44sometimes YouTube has to make these
875. 26:45unpopular decisions, but because like it
876. 26:47forecasts what happening in the market,
877. 26:48if it doesn't do it, it would just like
878. 26:50lose it. I know it's very hard to have
879. 26:51empathy for like a big company.
880. 26:53>> [laughter]
881. 26:54>> But at least from this perspective, I
882. 26:55truly seen that people who make features
883. 26:57and all all of the product managers,
884. 26:59they put a lot of thought into like
885. 27:00creators, but also how this going to
886. 27:02impact the market. You know, the world
887. 27:04is changing like YouTube is trying to
888. 27:05get trap and to become to be still like
889. 27:08this big rich platform that gives us an
890. 27:10opportunity and like gives us
891. 27:11monetization. Maybe it's just like what
892. 27:13it has to do.
893. 27:14>> That perspective kind of gets forgotten
894. 27:16about when YouTube changes and impacts
895. 27:19your entire livelihood. So, it's
896. 27:21refreshing to hear it.
897. 27:22>> But it's it's very aware like managers
898. 27:24are very aware it's impacting the
899. 27:25livelihood of people like
900. 27:27>> Yeah, okay. That's a tough decision
901. 27:29then. It's a tough call to be making
902. 27:30such such impactful changes to a
903. 27:33platform that has provided money
904. 27:35lifestyle for people. So, I'm glad I'm
905. 27:37not one of those developers.
906. 27:39>> [laughter]
907. 27:40>> That's it for this episode. Thank you so
908. 27:42much for watching or listening. If
909. 27:43you're new here, please consider
910. 27:44subscribing or leaving us a review on
911. 27:46the audio platforms. It really helps us
912. 27:48out and reach more people. If you are a
913. 27:50producer, a head of content, or creative
914. 27:52director for a channel that has over a
915. 27:54million subscribers, please reach out.
916. 27:56You have knowledge that I would love to
917. 27:57ask you about. We would love to get you
918. 27:59on the show. We'll see you next week
919. 28:00with another episode. Thank you [music]
920. 28:02so much for joining.