The Future Of SEO Is Scary — Transcript

Full transcript

1. 0:00Someone who helps Fortune 500 brands
2. 0:02rank inside ChatGPT and Claude just sat
3. 0:04down with Sequoia and said something
4. 0:06that should terrify every marketer
5. 0:08watching this. He said the internet as
6. 0:10we know it could be dead within 3 years.
7. 0:13That humans are about to stop browsing
8. 0:15websites entirely. That agents are going
9. 0:17to replace consumers as the people
10. 0:19walking through the front door of every
11. 0:21business on Earth. And the data he's
12. 0:23showing backs it up. But here's what
13. 0:25nobody's in the comments is talking
14. 0:26about. Yeah? Are you upset about it? You
15. 0:28keep wanting to be seen. There's one
16. 0:30category of business, one that is
17. 0:32structurally immune to what he's
18. 0:34describing. And Google has quietly
19. 0:36launched a product that confirms they
20. 0:38know it, too. I've been working in this
21. 0:40category for 10 years. Let me show you
22. 0:42what I see.
23. 0:43>> We've now reached a point in marketing
24. 0:45where if your marketing team is not
25. 0:46using agents,
26. 0:48and in particular profound agents, to do
27. 0:50marketing, then you are failing.
28. 0:53>> Hm.
29. 0:53>> It's gone from a nice-to-have to a
30. 0:55must-have. And I think the big
31. 0:56misconception with using agents to build
32. 0:58marketing is that it's just a way to
33. 1:01automate
34. 1:02the work that we've been doing in the
35. 1:03>> until now.
36. 1:04>> Yeah. Um well, thanks for having me on,
37. 1:06Sonia. Um
38. 1:08>> [snorts]
39. 1:08>> I mean, I think
40. 1:10you know, what we're
41. 1:12what we're witnessing is the biggest
42. 1:14platform shift in the history of
43. 1:15marketing
44. 1:16as the world turns from blue link
45. 1:20search, like predetermined blue link
46. 1:21search, to probabilistic AI responses.
47. 1:25Um
48. 1:26and it's and it's more than that as
49. 1:28well. It's not just kind of a case of do
50. 1:30you show up when
51. 1:32someone asks ChatGPT about your
52. 1:35category? Um it's you know, what does
53. 1:38ChatGPT say?
54. 1:39Or how does Claude, you know, recommend
55. 1:42your software, for example, if you're
56. 1:44talking about coding tools? Um
57. 1:46It's it's it's the these models are
58. 1:48really replacing, you know, it's the
59. 1:50agents and superintelligence is
60. 1:52replacing the role of the consumer. It's
61. 1:55not so much that the front door of the
62. 1:58internet has changed. It's actually the
63. 2:00person that's going through the door has
64. 2:01changed. It's gone from being a consumer
65. 2:03that is
66. 2:05using a list of blue links to
67. 2:08discover your website and click into it
68. 2:10to an agent is now using, you know, a
69. 2:12similar index and discovering your brand
70. 2:15products and services and then coming
71. 2:16back through the door and, you know,
72. 2:19maintaining that relationship with the
73. 2:20consumer.
74. 2:21>> So, it's interesting. So, the point he's
75. 2:23making is something that I've heard from
76. 2:25a lot of other people, including Sundar
77. 2:27Pichai, the CEO of Google himself. The
78. 2:30way he phrased this when asked about the
79. 2:32future of Google Search is that he sees
80. 2:34Google Search becoming an agentic
81. 2:37manager, where you would go to Google,
82. 2:39you'd type in what you're searching for,
83. 2:41what you're interested in learning, and
84. 2:42then instead of Google's algorithm
85. 2:45giving you links or an AI overview
86. 2:47giving you an answer, instead Google
87. 2:49will dispatch a series of AI agents to
88. 2:52go find exactly what you want. This is
89. 2:54exactly what he's saying. Marketers are
90. 2:57continuing to pattern match to the
91. 2:59latest platform shift. This is not what
92. 3:02we're talking about. The channel itself
93. 3:04isn't changing, the user is. Where we
94. 3:07used to think about writing so that we
95. 3:09could rank on Google Search, and then
96. 3:12once you're ranked, optimizing for the
97. 3:14humans. That's now not very important.
98. 3:17Most everything is writing to rank on
99. 3:20Google Search and to be recommended by
100. 3:22AI, and then that's kind of it.
101. 3:24>> That's so fascinating. Okay, so your
102. 3:26your point is, you know, ChatGPT may be
103. 3:28the new front door, but the thing that's
104. 3:29important for marketers is, you know,
105. 3:31you now have an agent walking through
106. 3:32that door and making a big part of that
107. 3:35buying decision for you.
108. 3:36>> Yeah, it's this I think that's the
109. 3:38that's the fundamental difference. It's
110. 3:40that the internet has remained the same.
111. 3:42It's just who is using the internet.
112. 3:44When you ask ChatGPT a question or
113. 3:46Gemini a question,
114. 3:48what you're essentially doing is asking
115. 3:50an agent to go and crawl the internet on
116. 3:53your behalf.
117. 3:53>> Mhm.
118. 3:54>> And that is an agent that's going to
119. 3:56visit all of those websites that you
120. 3:58used to visit. And it's an agent that's
121. 4:00going to determine if that's useful or
122. 4:02this is useful and
123. 4:04remix all of that information into an
124. 4:06answer that it spits out to you as uh
125. 4:09you know, here you go.
126. 4:11>> Yeah.
127. 4:12What do you think is the biggest
128. 4:14misconception people have about what it
129. 4:16takes to
130. 4:17to show up well for this new agentic
131. 4:19paradigm?
132. 4:20>> I think the biggest misconception is
133. 4:21that it's just SEO.
134. 4:23>> Mhm.
135. 4:24>> You know, there there's
136. 4:26a reason why that misconception exists
137. 4:28because yeah, of course, you still you
138. 4:30know, in in any world of marketing
139. 4:34and discovery, there is an impetus on a
140. 4:37brand or you know, the yeah, you said we
141. 4:38work with this is close to 12% of the
142. 4:40Fortune 500 now. Their marketing teams
143. 4:42use profound.
144. 4:44And in any world, the solution to a
145. 4:46problem is to create content, distribute
146. 4:49content on your own own channels or you
147. 4:52know, earned media or social channels
148. 4:53etc. So, they're very similar levers to
149. 4:56what we've seen with SEO and yes,
150. 4:58ranking on the index still matters. It's
151. 5:00just that
152. 5:02the human consumer is no longer using
153. 5:05the web.
154. 5:06And you are building content that may
155. 5:08quite literally never be consumed
156. 5:12by a human.
157. 5:13>> So, I can agree with what he's saying.
158. 5:15It's interesting because he almost
159. 5:16immediately went against himself. He
160. 5:18said it's not just SEO, but then he went
161. 5:21to talk about how ranking on Google
162. 5:24which is obviously SEO is still very
163. 5:26important. The difference he's talking
164. 5:28about is what I literally just said that
165. 5:31instead of writing content to rank on
166. 5:33Google and then convert humans, it's
167. 5:36writing content to rank on Google and
168. 5:38then talk to AI agents. For enterprise,
169. 5:41absolutely, there's a lot there. If we
170. 5:44think about local search, that
171. 5:46discipline is basically already there,
172. 5:48right? We know that pretty much every AI
173. 5:51agent is using the Google Business
174. 5:54Profile as the center of a business
175. 5:57entity. Uh Chat GPT uses Bing for
176. 5:59business, but it's basically the same
177. 6:01thing. You can actually even copy them
178. 6:02over. So, when we think about like doing
179. 6:05that for an AI agent in the local space,
180. 6:08it's really I don't know, GBPO?
181. 6:11So, the misconception here isn't local
182. 6:13marketers thinking that it's just SEO.
183. 6:17It's enterprise marketers thinking that
184. 6:19their playbook transfers one-to-one.
185. 6:22>> And I think the fundamental difference
186. 6:24between SEO and this new world is that
187. 6:27in SEO you were building content
188. 6:30that was designed to be, you know,
189. 6:32picked up by an algorithm,
190. 6:34but fundamentally consumed by a human.
191. 6:37Whereas in this new world, you are
192. 6:38building content that is, frankly,
193. 6:40designed entirely to be both discovered
194. 6:43and then consumed by an agent.
195. 6:46>> He's not wrong. And honestly, for local,
196. 6:49we've been in the situation for many,
197. 6:51many years, right? In local search, what
198. 6:54we're focused on doing is ranking the
199. 6:55GBP farther and farther away from the
200. 6:58searcher's location, okay?
201. 7:01We're not trying to rank the individual
202. 7:02website. Very few people who are
203. 7:05searching for a local business will
204. 7:07spend significant time on that local
205. 7:09business's website. From Google
206. 7:11Analytics data, it's like 20 to 30
207. 7:13seconds at most. So, for local search,
208. 7:16this isn't actually a huge change. We're
209. 7:19already building content to rank the GBP
210. 7:22so that the GBP can rank farther and
211. 7:25farther for more and more different
212. 7:26queries, knowing that the likelihood of
213. 7:28a human ever reading that content is
214. 7:31exceptionally low.
215. 7:32>> So, I would imagine humans are less
216. 7:35patient than agents. Um I would imagine
217. 7:38humans are more prone to
218. 7:40emotional biases and being swayed by
219. 7:43language than agents. Like, what are the
220. 7:45what are the biggest differences between
221. 7:47how humans and how agents consume the
222. 7:49internet that marketers should keep in
223. 7:51mind?
224. 7:52>> It's understanding that an agent
225. 7:54crawling the web looking for an answer
226. 7:56or providing an answer frankly um will
227. 7:59discover information differently
228. 8:02to a human. And so it
229. 8:05uses the index differently to us.
230. 8:08In the old world of SEO, you know, 95%
231. 8:11of the value was being in that top four
232. 8:13blue links.
233. 8:14>> Yeah.
234. 8:14>> The top five blue links.
235. 8:16And that really is a function of our
236. 8:18scarce cognitive energy and patience and
237. 8:20our
238. 8:21our lack of time.
239. 8:22>> Yeah.
240. 8:23>> Where an agent is using
241. 8:26that index, it's you what we've seen is
242. 8:28ChatGPT or Gemini or Claude are far more
243. 8:31prone to using the long tail of the
244. 8:33internet. And the amount of surface area
245. 8:35that an agent will use to answer a
246. 8:37question
247. 8:38is orders of magnitude
248. 8:41wider than a human.
249. 8:43>> Absolutely. Completely agree with this
250. 8:46point that he's making. And when Google
251. 8:49started to talk about its Ask Maps
252. 8:51feature, which basically integrates
253. 8:52Gemini into Google Maps,
254. 8:54uh Google started to talk about
255. 8:56attribute matching where somebody can
256. 8:58ask something very specific to Ask Maps,
257. 9:01very similar to what they might type
258. 9:02into ChatGPT, and Gemini or ChatGPT, the
259. 9:05agent of your choice, is actually going
260. 9:07to read all of the reviews about this
261. 9:09business on all of the different
262. 9:10platforms and look for evidence on those
263. 9:13reviews that this business it has done
264. 9:15that exact service before that you're
265. 9:17asking about. It's going to look on your
266. 9:19website to see if you talk about doing
267. 9:22that exact service, whereas a human
268. 9:24wouldn't. A human might need a water
269. 9:26heater replacement, so they type in
270. 9:28plumber Phoenix, and they're going to
271. 9:31just call the first or second result.
272. 9:33But, if you go to ChatGPT and say, "Hey,
273. 9:36I have a 50-gallon water heater. It's in
274. 9:39the closet. I need it replaced in the
275. 9:41next day." Then, the AI agent is going
276. 9:44to go and look at a lot more
277. 9:46information. And I believe 10 times, 100
278. 9:49times the amount of information. Look at
279. 9:51reviews. Has this plumber
280. 9:54replaced that size of water tank before?
281. 9:57Has he done that before in this closet?
282. 10:00In this suburb? Is this a specific place
283. 10:02he's he's working at before? Whereas a
284. 10:05human is just going to tap the call
285. 10:06button.
286. 10:07>> Like for instance, probably about three
287. 10:08or four months ago,
288. 10:10I was looking for a shower head
289. 10:12for my apartment in New York City, and I
290. 10:15used ChatGPT to help me find a new
291. 10:17shower head, and it used 65 different
292. 10:19web pages to answer that question.
293. 10:20>> Wow.
294. 10:22I think I probably went through 65 web
295. 10:23pages on my own for that query.
296. 10:26>> [laughter]
297. 10:26>> A very important purchase.
298. 10:27>> Yeah. Right. So, I mean, again, he's
299. 10:29right. But, if we think about someone
300. 10:31who needs a water heater replacement or
301. 10:34trucker who's broken down on I-10,
302. 10:36they're not asking ChatGPT to compare 65
303. 10:39options. They're opening maps, and
304. 10:42they're going to hit the first call
305. 10:43button. The user behavior that he's
306. 10:45describing here very much applies to
307. 10:48what we refer to as high-trust research
308. 10:51purposes. It does not apply to like an
309. 10:54urgent local service. So, what this
310. 10:56means, right, is if you're a local
311. 10:57service like an LASIK eye surgeon or
312. 11:00cosmetic surgeon, then we absolutely in
313. 11:03Google Analytics data, we see people
314. 11:05spend 10, 15 minutes on websites for
315. 11:09those high-trust services. Most local
316. 11:12businesses aren't in those spaces. Most
317. 11:14local businesses Google Analytics,
318. 11:16people are spending less than 20 seconds
319. 11:18there.
320. 11:19>> And and and if if if so, what what's the
321. 11:21root cause?
322. 11:22>> Yeah, I mean, we see huge differences
323. 11:24between the platforms.
324. 11:25>> Okay.
325. 11:25>> Um and frankly, you know, as a shameless
326. 11:27plug, that's why marketers are using
327. 11:31software like profound because what
328. 11:33we're able to do is help you understand
329. 11:35not just how your brand or product shows
330. 11:38up across different platforms. So, you
331. 11:40know, okay, do you show up more
332. 11:41frequently in Gemini responses or Claude
333. 11:43responses or ChatGPT responses?
334. 11:45>> Mhm.
335. 11:46>> Also, we extract the sentiment and the
336. 11:47themes around you know, so when Claude
337. 11:49surfaces your brand or product,
338. 11:51what are the other things that it says
339. 11:52alongside the answer?
340. 11:54>> Yeah.
341. 11:54>> But then we also get to the root cause.
342. 11:56So, we show we expose to marketers,
343. 11:59okay, these are the citations and
344. 12:01sources that the different models are
345. 12:02using
346. 12:04to answer questions about your brand,
347. 12:06your products, your category, or your
348. 12:07competitors.
349. 12:08>> Yeah.
350. 12:08>> So, once you understand the what and the
351. 12:10why,
352. 12:11then in profound what you're the next
353. 12:14step is you're building and deploying
354. 12:15your own agents
355. 12:18to sort through all of that data and
356. 12:20build a new type of marketing that
357. 12:22>> And is the primary root cause that they
358. 12:24have, you know, different harnesses is
359. 12:26the primary primary root cause that they
360. 12:28have different training data mix?
361. 12:29Primary root cause some of them, you
362. 12:30know,
363. 12:31bias for, let's say, Reddit data over,
364. 12:34you know, company-owned
365. 12:36platforms? Like, what's the root cause
366. 12:38for why these platforms are so different
367. 12:39in terms of how companies show up?
368. 12:41>> Think of them this sounds either very
369. 12:43reductive, but think of them as just
370. 12:44different
371. 12:46and you know, different species.
372. 12:48>> Mhm.
373. 12:49>> If the, you know,
374. 12:50what we've seen is Gemini will lean on
375. 12:53YouTube content a ton, which makes sense
376. 12:55cuz Google owns YouTube. So, that we see
377. 12:57we see
378. 12:58you know, YouTube being a huge lever for
379. 13:00brands or marketers that want to appear
380. 13:02in Gemini responses.
381. 13:03>> Mhm.
382. 13:04>> Whereas ChatGPT, yeah, we see typically
383. 13:06pulls from Reddit if it's consumer or if
384. 13:09it's B2B, it will typically pull from
385. 13:11LinkedIn. We see as a huge source of
386. 13:13truth.
387. 13:13>> Mhm.
388. 13:14>> Um
389. 13:15Claude is
390. 13:17Claude has has been changing a lot
391. 13:19recently. You know, I think what
392. 13:20>> And why is that? I know there's this It
393. 13:22seems like between 4.5 and 4.6 even like
394. 13:24what it's recommending has changed a
395. 13:25lot. Why do you think that is?
396. 13:27>> I think Claude has typically or Claude
397. 13:30has historically relied more on the
398. 13:32pre-trained LLM to answer questions
399. 13:35and is now I think they've updated their
400. 13:38classifiers or something that's you know
401. 13:40basically that the classifier seems to
402. 13:42have become a bit more sensitive to you
403. 13:44know real-time information. So Claude
404. 13:47will use the web more to answer
405. 13:48questions is what we're finding.
406. 13:50>> Okay.
407. 13:51>> But I think you know the the next
408. 13:52paradigm here we're so we're so we're
409. 13:55such
410. 13:56>> [sighs and gasps]
411. 13:56>> humans are such creatures of heuristics.
412. 13:59>> So this is interesting and this is
413. 14:01something that we've known and we've
414. 14:02honestly been taking advantage of again
415. 14:04in the local space. Right? We know that
416. 14:07YouTube presents feeds Gemini. We know
417. 14:09that reviews on your GBP on Yelp on
418. 14:13Trustpilot feed all of these different
419. 14:15AI tools. This is one of the reasons
420. 14:17that at our local SEO agency we will
421. 14:20often produce YouTube videos for every
422. 14:23third to every fourth article we
423. 14:25publish. One that helps rank on Google
424. 14:28but even in a world where Google is
425. 14:30going away
426. 14:31having a presence on YouTube also helps
427. 14:34feed Gemini which is incredibly helpful
428. 14:37for the ask maps feature which we see as
429. 14:40becoming the future of local search. So
430. 14:43again this GBP and model differences
431. 14:47this multi-platform play has very
432. 14:49similar equivalents for local space.
433. 14:52>> Two articles one written by a human like
434. 14:54a journalist and the second written by a
435. 14:56by AI
436. 14:58and it was like 53% of readers voted
437. 15:01afterwards but on a blind test that they
438. 15:03preferred the AI written content.
439. 15:04>> Wow.
440. 15:05>> I think slop is a red herring that
441. 15:09is going to be quite quickly disproven.
442. 15:13You you can you know it's it's it's a
443. 15:14really
444. 15:15I'm not saying you're suggesting this
445. 15:17but I do think that this idea of
446. 15:19if it's written by AI equals slop
447. 15:23uh is a stupid one.
448. 15:25Um
449. 15:26I think AI is more than capable of
450. 15:28writing high-quality content,
451. 15:30high-quality marketing. It's just that
452. 15:33the way to think about it is that the
453. 15:34consumer is super intelligent now.
454. 15:36So, you as a brand
455. 15:39or a marketer, you need to tell Claude
456. 15:43something it doesn't know.
457. 15:44>> Mhm.
458. 15:45>> How do you tell a super intelligent
459. 15:46being something it doesn't know already
460. 15:49when it's been trained on the entire
461. 15:50internet?
462. 15:51>> Mhm.
463. 15:52How do you?
464. 15:53>> [laughter]
465. 15:53>> What What do you say?
466. 15:54>> Have original insight. You know, really
467. 15:56humans are this kind of fleshy API
468. 16:00between reality and
469. 16:02the internet at this point, right?
470. 16:04>> agree. And if we're talking about
471. 16:06writing content for SEO, writing content
472. 16:09for AI systems, what we want to do is
473. 16:11take a look at the content that already
474. 16:13exists about that topic, and then write
475. 16:15something similar to it, but add unique
476. 16:18insights, add more information that
477. 16:20isn't already on the other results that
478. 16:23are ranking nearby. So, every time we're
479. 16:25writing content for local SEO, we're
480. 16:27always starting our step one is going to
481. 16:30be to do research about that exact
482. 16:33service we're writing content about. Not
483. 16:35just the service, that exact service in
484. 16:37that exact city, okay? So, we're going
485. 16:39to look at Google's people also ask
486. 16:41questions. We're going to look at
487. 16:42Reddit. We're going to look at Facebook
488. 16:43forms. We're going to look at the
489. 16:44competitor websites. We're going to look
490. 16:46at other local forms. We're going to
491. 16:48find out what are people actually asking
492. 16:50about for that service in that local
493. 16:53city, and then we're going to write
494. 16:54about that. We're going to make sure
495. 16:56that is included in our local articles,
496. 16:59okay? This point that he's making,
497. 17:01original insight, is the moat. Original
498. 17:04insight is what gets you cited, what
499. 17:06gets you ranked. So, that means the
500. 17:09original insight, it's first-party
501. 17:12operational truth, right? That means
502. 17:14your service area. That means your
503. 17:15photos, your reviews, your actual
504. 17:17problems you can solve, your properly
505. 17:19categorized service list. This is
506. 17:21exactly why the system that we use, that
507. 17:24we call the core 30, this is exactly why
508. 17:26it works. It's not clevered content, it
509. 17:29is structured truth that is designed to
510. 17:32build this trust with these AI systems
511. 17:35that exactly match your website to your
512. 17:38GBP, to what people are actually asking
513. 17:40about in your area.
514. 17:42>> I'd love to chat about
515. 17:45dead internet theory.
516. 17:47Like,
517. 17:48at what point do you think the internet
518. 17:49is just primarily being browsed by
519. 17:51agents? And at that point,
520. 17:53you know, does a company's
521. 17:55marketing website matter at all? Does a
522. 17:57company's, you know, ranking in the
523. 17:59search engine, in the traditional search
524. 18:01engines, matter at all? Or should we all
525. 18:03just have a read me file for the agents
526. 18:05to crawl?
527. 18:07>> I mean, at the risk of sounding a little
528. 18:08dramatic, I think that we could
529. 18:10experience a dead internet outcome in
530. 18:12the next 3 years.
531. 18:13Um, I do think it's possible. Um, you
532. 18:15know, maybe not likely, um, but, you
533. 18:17know, what does that mean? It means that
534. 18:19in a world where humans just speak to AI
535. 18:22to get the responses that they need,
536. 18:25the incentive to publish content
537. 18:29diminishes
538. 18:31to the point of zero.
539. 18:32Meaning, you know, we rely, as humans
540. 18:34today,
541. 18:35>> This is a strange dichotomy from his
542. 18:38earlier stance about writing content for
543. 18:41the AI systems. Now he's saying there's
544. 18:43no benefit to writing content, but if
545. 18:45the AI systems are reading it, absorbing
546. 18:48it, and providing advice to their users
547. 18:50based on the content, you still have the
548. 18:53advantage, you still have the incentive
549. 18:56to write content, even if it's not for
550. 18:58humans.
551. 18:58>> and AI rely, we we underestimate just
552. 19:02how much we rely on first-party
553. 19:05reporting to to feed the information
554. 19:08that AI uses to answer questions or to
555. 19:10feed the information that appears in a
556. 19:11search engine that we rely on every day.
557. 19:14And I think in a world where humans no
558. 19:16longer click into websites, the majority
559. 19:19of the internet is still funded by
560. 19:20advertising
561. 19:21for a start. So, you know, most
562. 19:23publishers rely heavily on advertising
563. 19:25revenue
564. 19:26to to fuel all of the content that is
565. 19:30being more and more consumed by AI.
566. 19:32>> Mhm.
567. 19:32>> In a world where consumers, humans
568. 19:35aren't visiting those web pages anymore,
569. 19:38well, what's the you know, what's the
570. 19:39point in advertising on a web page that
571. 19:41a human isn't visiting?
572. 19:42>> Mhm.
573. 19:42>> Then their business model breaks, the
574. 19:43economics of the internet break, and the
575. 19:46incentives to create editorial content
576. 19:51are removed.
577. 19:52>> Yep.
578. 19:52>> And then in that world, you start to ask
579. 19:55the question, well, where does AI go to
580. 19:57answer these questions?
581. 19:59>> To the readme files.
582. 20:01>> But then who Why would you publish a
583. 20:02readme file?
584. 20:05>> Cuz the agent needs to go somewhere to
585. 20:06answer a question, right?
586. 20:07>> It works if you're a brand or a
587. 20:08marketer, but why would you you know,
588. 20:10would you publish a blog? What's the
589. 20:11point?
590. 20:12>> I publish a hi, I'm Sonya the the VC,
591. 20:15and these are the companies I've worked
592. 20:17with. And in in like as a you know,
593. 20:19founders trying to reason through which
594. 20:20venture capitalist I work with, I get
595. 20:23finds my readme file.
596. 20:24>> I think that's correct for
597. 20:26>> Yeah.
598. 20:28Measured agreement with what he's saying
599. 20:29here. The incentive structures are
600. 20:31breaking, and this is challenging, it's
601. 20:34problematic, it's a problem to figure
602. 20:36out. But honestly, the one category that
603. 20:39at least for now is essentially
604. 20:41structurally immune to this is local
605. 20:43service businesses. Because at the end
606. 20:45of the query, there still has to be a
607. 20:47human with a wrench who can fix the
608. 20:51thing, okay? Agents can research, they
609. 20:53can compare, they can recommend. As of
610. 20:55yet, they can't actually unclog a drain.
611. 20:58Now, Dario Amodei, the CEO of Anthropic,
612. 21:01thinks that's coming in the next 5 to 10
613. 21:03years, but that seems soon.
614. 21:05Google knows this. That's why they
615. 21:07rolled out Ask Maps. Now, Google has
616. 21:10publicly described the future of search
617. 21:12as basically an agentic search manager.
618. 21:16Ask Maps launched as Google's first
619. 21:18agentic consumer project for Ask Maps
620. 21:21launched as Google's first agentic
621. 21:23consumer product for local discovery.
622. 21:26The signal to this is that Google is
623. 21:27conceding discovery to AI across the
624. 21:30board, except they're building their own
625. 21:32agent specifically for local because GBP
626. 21:35is a data moat that no one else has.
627. 21:38Lily Ray did analysis on this and that
628. 21:40showed that AI mode sites GBPs more than
629. 21:43any other external websites for local
630. 21:45queries. That's what I mean when I talk
631. 21:47about how the GBP is the center of your
632. 21:50business's entity. So, he's right that
633. 21:53the internet very likely could break,
634. 21:56but what he's missing is that Google
635. 21:58already knows where the next moat is and
636. 22:01it's the category with a human at the
637. 22:03end of every single query. Local. GBP is
638. 22:06their data moat. GBPs will keep people
639. 22:09and AI systems coming back. The core 30
640. 22:13is how you defend it. Now, this isn't
641. 22:15the end of local SEO, it's the beginning
642. 22:17of local AEO to borrow a silly acronym,
643. 22:22but AEO is fundamentally 90-95% the same
644. 22:26thing as local SEO. It looks so much
645. 22:30like the local SEO that you've already
646. 22:31been doing. There's just more at stake
647. 22:34because of the podium is getting
648. 22:36smaller. So, I want to come back to
649. 22:37something he said about the dead
650. 22:38internet theory. I've been thinking
651. 22:40about this. I filmed that original video
652. 22:41a couple days ago and I just couldn't
653. 22:43get this out of my head, so I felt
654. 22:44compelled to come back and add this. I
655. 22:47think he's wrong about what happens
656. 22:49next. His argument is that the internet
657. 22:51dies because the ad-funded content goes
658. 22:53away. Publishers stop publishing,
659. 22:55bloggers stop blogging. No humans
660. 22:57clicking means no ad revenue means no
661. 22:59reason to create. Fine. That part
662. 23:01probably does collapse or at least
663. 23:03suffers a lot. But look, a huge chunk of
664. 23:06the internet is not ad-funded. It's
665. 23:08people trying to sell you something.
666. 23:10It's people trying to convince you of
667. 23:12something. It's brands, it's affiliates,
668. 23:13it's consultants posting thought
669. 23:15leadership, it's local businesses
670. 23:17writing about their services. None of
671. 23:19that goes away when the ad tier dies.
672. 23:21They're not writing for ad revenue.
673. 23:23They're writing to influence. And in a
674. 23:25world where LLMs are the new front door,
675. 23:28what do you think that group people
676. 23:30does? They don't stop publishing. They
677. 23:32publish more. They publish specifically
678. 23:34to feed the models. They publish to get
679. 23:36cited. They publish so their brand shows
680. 23:38up when an agent is making a buying or
681. 23:41influencing decision on behalf of a
682. 23:42customer. So the internet doesn't die.
683. 23:45It gets worse than dead. The
684. 23:47disinterested content, the hobbyist
685. 23:50writing about their niche, the
686. 23:51journalist covering something because it
687. 23:53matters, the person just sharing what
688. 23:55they learned, that's the stuff with the
689. 23:57weakest incentive to keep going. That
690. 24:00dies first. What survives is content
691. 24:03with an agenda attached to it.
692. 24:05Commercial content, persuasion content,
693. 24:07content written by people who want
694. 24:09something from you. And all of that is
695. 24:12now feeding the LLMs, which means that
696. 24:15the models are being trained and
697. 24:16grounded on an internet where every
698. 24:18remaining voice has a motive. That's not
699. 24:20dead, that's worse than dead. That's a
700. 24:23propaganda layer dressed up as an answer
701. 24:26engine.
702. 24:28Okay, wow. That feels negative to say.
703. 24:30Now, if you want to go deeper on where
704. 24:32this is all heading, the Google CEO did
705. 24:34an interview recently where he walked
706. 24:36through exactly this. The agentic search
707. 24:38manager, the handoff layer, the role of
708. 24:40maps in all of this. I have a full
709. 24:42breakdown to that interview linked up
710. 24:44top. It's the perfect companion if you
711. 24:46thought this was an interesting video.
712. 24:48And honestly, the two together give you
713. 24:50the complete picture of where search is
714. 24:52going in the next 12 months. Watch that
715. 24:55next.