ChatGPT Ads Don't Convert — I Know Why — Transcript

Full transcript

1. 0:00Do chat GBT ads actually work?
2. 0:03Because
3. 0:04I've had a lot of clients trying them
4. 0:07lately and they certainly don't seem to.
5. 0:10Then
6. 0:12last week Robbie
7. 0:14I type of senior
8. 0:16ad strategist, him and I were sitting on
9. 0:18a call for a local business called Money
10. 0:22Penny. They did a webinar. There was
11. 0:24about 80 pretty good experts on it
12. 0:26actually, but 80 people on this webinar
13. 0:28and they were everybody was discussing
14. 0:31chat GBT ads and how it was going and
15. 0:33everything like that. And one of the
16. 0:35things that came out as consistent
17. 0:37message was starts well and then just
18. 0:39dies flat. So in their case if they had
19. 0:42six conversions, they put a fair bit of
20. 0:44money against it as well. Six
21. 0:45conversions and then just totally flat
22. 0:48and dead and nobody really seemed to
23. 0:50have any type of answer for this. So
24. 0:55as I do I wandered off, did the rest of
25. 0:57my daily work and then started pondering
26. 1:00this question a little bit. And one of
27. 1:02the things that started to annoy me a
28. 1:04little bit was, you know what? When you
29. 1:06start to really think about chat GBT
30. 1:08ads, you realize something so
31. 1:11fundamental. I don't know how these guys
32. 1:13didn't see it.
33. 1:15You know I love this book called Path
34. 1:17Through the Jungle and he talks about
35. 1:19the human side of the brain and he talks
36. 1:21about the chip side of the brain and you
37. 1:24probably know it as like you get a gut
38. 1:25reaction, you know,
39. 1:27fight or flight, that type of response.
40. 1:30Now I started to think about this cuz I
41. 1:32do a lot of big thinking, nothing else
42. 1:34to do these days really. Teams running,
43. 1:36making money for everybody.
44. 1:37Anyway, when you really start to think
45. 1:40about it, chat GBT ads is unlike any
46. 1:43other advertising platform and it's not
47. 1:46because it's AI, far from it.
48. 1:48Just get myself through the gate here.
49. 1:50Everybody
50. 1:51moans.
51. 1:52It's a bit too narrow, but
52. 1:55like a ninja there.
53. 1:57Anyway,
54. 1:58I think Todd's probably in the studio
55. 2:00today, so I might as well stay here and
56. 2:01continue this conversation with you.
57. 2:05If you think about Chat GPT ads,
58. 2:08they're fundamentally different from
59. 2:10Google and Meta ads. So, Meta really
60. 2:14knows everything about you that you
61. 2:16don't disclose. So, you're sitting there
62. 2:18and I've got a phone in my pocket here.
63. 2:20Where is it? Not too sure.
64. 2:22So, me and my phone, I'm busy scrolling
65. 2:24through. Something catches my on
66. 2:26Instagram, I pause, I watch the video,
67. 2:28same on Meta, everything like that. So,
68. 2:31it knows what I don't disclose. It's
69. 2:33like your actual real inner chimp, what
70. 2:36attracts the chimp. And
71. 2:39the same is whereby this phone knows I'm
72. 2:42now at work. So, basically, I've left
73. 2:45home, uh driven to work.
74. 2:47Meta knows exactly where I am. Now, it's
75. 2:51the same thing with Google ads.
76. 2:54Google ads is built on the Google
77. 2:55platform. So, we've got analytics, we've
78. 2:57got Google Maps. So, Google Maps knows
79. 2:59again I'm sitting exactly here. It knows
80. 3:02the journey I do every day. So, whereas
81. 3:04on Meta, I'm subconsciously disclosing
82. 3:07disclosing what I I actually want and
83. 3:10like subconsciously, it also listens to
84. 3:13us, we all know that one. But on Google,
85. 3:15I'm actually purposely going there only
86. 3:18when I'm interested in solving a
87. 3:20purchasing problem. These days, you
88. 3:22don't really go to Google unless I I
89. 3:23don't want some new Ray-Ban sunglasses,
90. 3:26Ray-Ban aviator sunglasses. Type that
91. 3:28in, whole load of advertisers come up,
92. 3:30and then Google will follow you round.
93. 3:32It will show you
94. 3:33what do you call it, Ray-Ban sunglasses
95. 3:35ads on YouTube and everything, all that
96. 3:37top of funnel stuff I keep talking
97. 3:38about. So, we're going to this location
98. 3:41on Google with a buying and purchase
99. 3:44intent. We're on Meta, where they know
100. 3:47what our hidden purchase intents are.
101. 3:50So, if you So, for example, Paddy here
102. 3:53he has just got engaged. Fantastic.
103. 3:56Absolutely brilliant for him. So happy.
104. 3:58But, he would have been looking at
105. 3:59rings, he would have been looking at all
106. 4:01different types of things about
107. 4:03engagements, engagement speeches,
108. 4:05proposals, etc., etc. So, Meta knew
109. 4:08everything about that to do with it. He
110. 4:10[snorts] also then started looking for
111. 4:12rings actually on Google, so they knew
112. 4:15about it and they knew he was looking
113. 4:16for a ring. Now, think about this.
114. 4:19If you consider
115. 4:21Chat GPT for once, I call Chat GPT
116. 4:25disclosure advertising. And the reason
117. 4:27I'm calling it disclosure advertising is
118. 4:30unless you tell it something about what
119. 4:32you're doing, then
120. 4:35it doesn't know anything about it. So,
121. 4:37most of you are going there and you're
122. 4:39trying to solve some technical
123. 4:40advertising problems, you're trying to
124. 4:42solve some legal problems, you've got a
125. 4:43dispute with a tenant, you've got maybe
126. 4:47an issue with the car you've got and
127. 4:49you're looking at how you fit this part.
128. 4:50But, unless you tell it about it, then
129. 4:53it doesn't know anything about it. So,
130. 4:54do this little bit of an experiment. Go
131. 4:56to Chat GPT and type in, "Tell me
132. 4:58everything you know about me and
133. 5:00anything you're, I don't know,
134. 5:03surmising or or guessing about me." Just
135. 5:05type that in and see what comes back.
136. 5:07You're going to be surprised. It's going
137. 5:09to be amazingly accurate about a lot of
138. 5:11things, but much of your life and much
139. 5:14of your activities it knows nothing
140. 5:16about. So, if we come back to the Meta
141. 5:18thing again, on Meta, we know that
142. 5:21virtually every website on the planet
143. 5:24has got that Facebook pixel on it. And
144. 5:26we know virtually every
145. 5:28uh commercial site on the internet has
146. 5:31got Google Analytics on it. So, it
147. 5:33doesn't matter with either of those two
148. 5:35platforms if you've actually told it
149. 5:37something. If If Paddy starts going
150. 5:40looking for engagement rings, starts
151. 5:41going around all these jewelers,
152. 5:43Google automatically knows about it
153. 5:45already, so it can start sticking ads on
154. 5:47YouTube. Then, the next thing that comes
155. 5:49in is that
156. 5:52Facebook and Meta and Instagram, they
157. 5:54know that as well. And TikTok will know
158. 5:56an awful lot as well, cuz they have
159. 5:57pixels on a lot of sites. Now, so you
160. 5:59can see they have far more visibility.
161. 6:02As you travel on their platform or off
162. 6:05their platforms, they know everything
163. 6:08about you.
164. 6:09The only thing ChatGPT knows about you
165. 6:11is what you disclose to it in a chat.
166. 6:15So, it means that that targeting isn't
167. 6:18really as focused and as accurate as
168. 6:20everybody else's. Now, I know the
169. 6:22argument as we go forward and the what's
170. 6:25called the Open AI pixel is put on more
171. 6:27and more websites, they will get to the
172. 6:29point where they can follow us around
173. 6:31and see what's happening. But, at the
174. 6:32moment, that's the problem they've got.
175. 6:35Unless you disclose onto that chat, then
176. 6:39they've got no visibility of you. It
177. 6:41gets a little bit more complex as well,
178. 6:43because if you've already the people who
179. 6:45are probably their ideal candidates,
180. 6:48they're the ones who are paying for the
181. 6:49subscriptions. They don't don't often
182. 6:51get to see ads. It's only the guys who
183. 6:53are paying and not paying on the free
184. 6:55ones. Then, we get to something else.
185. 6:58We get to what I call I I suppose we'd
186. 7:01call it nested disclosure, or we could
187. 7:04call it nested chats. I think of a
188. 7:06better word for it, anyway.
189. 7:09Karen
190. 7:11basically has ChatGPT, I put it on a
191. 7:13phone for her, and all she's doing with
192. 7:16it is using it to do funnies to send to
193. 7:18her mom and her friends on their
194. 7:19WhatsApp groups. So,
195. 7:21ChatGPT knows nothing at all about her
196. 7:25other than she's doing these funnies.
197. 7:28And how many people are doing that every
198. 7:29day? That's not real advertising intent,
199. 7:32is it? In buyer intent, rather. That's
200. 7:34not someone going to Google and saying,
201. 7:36"Hey, I want the sunglasses." It's not
202. 7:39me pausing on a sunglasses ad.
203. 7:43or a video from a UGC video content. You
204. 7:46see the problem that start here? Now, on
205. 7:48my phone, of course, when there's a real
206. 7:51problem that she wants to know about
207. 7:52something, she doesn't really feel that
208. 7:54confident with using the ChatGPT. She
209. 7:56doesn't really like it. It's okay for
210. 7:57images, she's fine with that. So,
211. 7:59Muggins here gets asked to go and find
212. 8:02out about this. So, now I'm typing on my
213. 8:04chat about something she wants to know
214. 8:06about. And then I'll be wandering around
215. 8:08the office and somebody will say
216. 8:09something like, "Oh, what was that song
217. 8:11in the '80s?" And I'm there, "Doo doo
218. 8:12doo doo."
219. 8:13That's not even my my chat, is it? So,
220. 8:16what you're getting in within the
221. 8:18ChatGPT history of all the conversations
222. 8:20you're getting, you're getting queries
223. 8:22that are related to colleagues, friends,
224. 8:24your partner's requirements, not yours.
225. 8:28So, you can see how it starts to add to
226. 8:30the inaccuracy of this whole ChatGPT
227. 8:33advertising.
228. 8:34Now,
229. 8:35so I think that fundamentally is the
230. 8:37problem. Meta knows everything about you
231. 8:40subconsciously.
232. 8:42Face uh rather Google knows about you
233. 8:46when you're actually going making a
234. 8:47query like Ray-Ban sunglasses. It's a
235. 8:50buying signal that it can actually
236. 8:52immediately act on. But you're not going
237. 8:55to ChatGPT and put in "Best place to buy
238. 8:58Ray-Ban sunglasses."
239. 9:00You're just not doing it. You're asking
240. 9:02it more these deeper questions. You're
241. 9:04asking it a problem about let's say you
242. 9:06need a remortgage or you've got some
243. 9:09debt consolidation issues you need to
244. 9:10talk about. That's the questions it's
245. 9:12getting there. Now, so
246. 9:15in effect, anything that's really
247. 9:18complex and really needs a solution to,
248. 9:20ChatGPT's ads may be very good.
249. 9:23But anything to do with fashion,
250. 9:26apparel,
251. 9:28holidays, those types of things, ChatGPT
252. 9:31probably isn't getting that visibility
253. 9:33on you. So, it only He a partial view of
254. 9:36you.
255. 9:37So, that explains, I think, why the
256. 9:39targeting is so far off. So, the guy
257. 9:41from Moneypenny was mentioning he
258. 9:43doesn't like the gym. He's not keen on
259. 9:45it at all. Doesn't really do it. And
260. 9:47yet, he's seeing ads for Under Armour.
261. 9:49And I thought about that, and I just run
262. 9:51back through this idea of nested
263. 9:52conversations. He's probably somebody
264. 9:55has asked him a question, "Oh, for the
265. 9:57Moneypenny telephone answering service,
266. 9:58we want to target uh gym coaches at the
267. 10:00moment and fitness instructors." So,
268. 10:03they've done a little bit of
269. 10:05Googling away on ChatGPT, asking it the
270. 10:07questions. And that's how it that
271. 10:09profile has come to the conclusion,
272. 10:11incorrectly, that he's got an interest
273. 10:13in actually Under Armour and fitness
274. 10:16apparel.
275. 10:17So, here it goes. Now, the next one is
276. 10:20The next one is the conversions when you
277. 10:22start ChatGPT
278. 10:24on advertising, they just die.
279. 10:26They'll start off, they'll go for a
280. 10:28while, and bang. And what a lot of
281. 10:30advertisers are actually doing is
282. 10:32they're pausing that campaign,
283. 10:34rebuilding a new campaign, setting that
284. 10:36live, and off it goes again doing that.
285. 10:38So, they're they're in effect having to
286. 10:40do
287. 10:41continuously reinvent the campaign to
288. 10:43get conversions.
289. 10:45And
290. 10:46I began to wonder about that. Why is
291. 10:48that the problem that's occurring here?
292. 10:50And you know, when you're doing Google
293. 10:51Ads and you come across a campaign where
294. 10:53somebody's just gone in there and they
295. 10:54put broad match on anything, and they've
296. 10:56got a really low budget. What they're
297. 10:57doing is they've got that bang bang
298. 10:59bang, a shotgun approach. So, they're
299. 11:00just spraying the ads everywhere on
300. 11:03really cheap bids. And some of it
301. 11:04actually converts. So, the next thing
302. 11:06you do as a Google Ads advertiser is you
303. 11:09think about this and go, "Aha! I know
304. 11:10what I'm going to do is I'm [snorts]
305. 11:12going to add a whole lot of negative
306. 11:13keywords in." Now, remember, ChatGPT
307. 11:15hasn't got any negative targeting, which
308. 11:17is a real flaw in it. And then, the next
309. 11:20thing is Well, it's not only the
310. 11:21negative keywords it hasn't got at all,
311. 11:23it's got real problems with its
312. 11:24geo-targeting as well. You can only do
313. 11:26the whole of the United States, United
314. 11:28Kingdom. You can't really get granular
315. 11:30on where you want to go. So, there's a
316. 11:31whole range of issues with targeting on
317. 11:34Chat GPT that I'll cover off.
318. 11:36Anyway, back to our problem. So, in
319. 11:39Google Ads, what you do is you'd add a
320. 11:41whole lot of negative keywords in, you
321. 11:42choose the winning ads, and then you
322. 11:44start pushing the bids on the keywords
323. 11:46that actually do convert. And what you
324. 11:48find out is you find out something
325. 11:50really strange. The conversion costs
326. 11:52either go through the roof because
327. 11:54you're now competing solely on the terms
328. 11:56that actually convert, or you don't get
329. 11:58any at all because you were randomly
330. 12:01these search terms and Google will throw
331. 12:02you them at you on the cheap trying to
332. 12:04bait you, but now you're having to
333. 12:05actually compete on those terms very
334. 12:08specifically, you can't afford them
335. 12:10anymore, and the campaign just dies.
336. 12:12And I'm beginning to wonder if that is
337. 12:14the same thing on Chat GPT Ads. What
338. 12:18it's doing is it's type of targeting in
339. 12:21on the actual terms or the chats that
340. 12:24are actually going to convert and the
341. 12:25profiles that are going to convert, but
342. 12:27once you get into there, the actual cost
343. 12:29of those CPCs CPAs to acquire those
344. 12:32conversions is actually a lot higher
345. 12:34than your budget can support. So,
346. 12:36literally, it just dies.
347. 12:38You could try throwing a lot more budget
348. 12:40about
349. 12:41at it, and I know some people have done
350. 12:43that as well, and but the same thing
351. 12:45tends to happen again, and I'm wondering
352. 12:47when you actually start getting into
353. 12:48this, is the actual pool incredibly
354. 12:52small to get to? Cuz if you think about
355. 12:55it, when you go to Google, there's ads
356. 12:57everywhere. There's plenty of inventory
357. 12:59space. When you're on
358. 13:01Instagram, you're on Google or Google
359. 13:04Facebook, there's huge amounts of
360. 13:06inventory there as well. You can scroll
361. 13:07there, never-ending inventory. On Chat
362. 13:10GPT, you just see that occasional little
363. 13:12ad to the side. So, they haven't got a
364. 13:14lot of inventory there to put your ad in
365. 13:17front of. So, that could be a real
366. 13:19problem as well. The audience for actual
367. 13:22purchases and the the available spaces
368. 13:24that you've got on Chat GPT, there's
369. 13:27Todd wandering in there, our new UGC
370. 13:29creator guy. He's off to rob the studio
371. 13:31again. Anyway,
372. 13:33I think that's the That's the problem.
373. 13:35The problem is that there is one when it
374. 13:38initially finds what it needs to really
375. 13:40target, it turns out the CPCs are
376. 13:42incredibly high, and then it finds out
377. 13:45that the audience is incredibly small.
378. 13:47So, that's my type of
379. 13:48idea about why chat GBT ads aren't
380. 13:51currently working. It's running a
381. 13:53different form of advertising. It's not
382. 13:55It's not doing the subconscious
383. 13:58advertising that
384. 13:59Facebook does or Meta does, and it's not
385. 14:02doing the high buyer intent advertising
386. 14:04that Google Ads does. And both of those
387. 14:06platforms have the huge visibility of
388. 14:08what you're actually doing around the
389. 14:09internet. What it's doing is disclosure
390. 14:11advertising. There's a new one. I might
391. 14:13be credited with that idea one day. So,
392. 14:16that's its problem. It own You can only
393. 14:18advertise on what it discloses, and it
394. 14:20doesn't really know what your buying
395. 14:22intent is because
396. 14:23back to that book the the about the the
397. 14:25path through the jungle with the chimp.
398. 14:27The other thing I forgot to mention is
399. 14:28the chimp is the one that often makes a
400. 14:30buying decision. The human rationalizes
401. 14:33a decision. So, if I see the sunglasses,
402. 14:35my chimp goes, "Ooh, sunglasses. Love
403. 14:37those. Oh god, that looks He looks cool
404. 14:39with them. I'll have them. I imagine
405. 14:41them on myself. No problem at all."
406. 14:43Chimp buys it. It's the same when you're
407. 14:45on Google Ads. On Google Ads, you're
408. 14:48You've actually the chimp has got a
409. 14:50problem, and it's in control because
410. 14:53anything that really catches the chimp's
411. 14:55attention, like anything he wants,
412. 14:57anything he desires, it takes over the
413. 14:59control of the buying process. So, when
414. 15:01you're there on Google typing in Ray-Ban
415. 15:05sunglasses, the chimp is trying to solve
416. 15:07that problem, and it's trying to solve
417. 15:08it now. The chimp has got no idea of
418. 15:10time. All it ever wants to do is solve
419. 15:13problem now. As very much like a
420. 15:15chimpanzee or any human would want to
421. 15:17do. And therefore, it sees solution, it
422. 15:19buys solution. But when you're on chat
423. 15:22GBT, remember you're dealing with the
424. 15:24human, the human that does logic,
425. 15:26evaluates, looks at the options. But,
426. 15:28the human never makes buying decisions.
427. 15:30Buying decisions are nearly always made
428. 15:33by the emotional chimp side of your
429. 15:34brain. So, what you've got is two
430. 15:36advertising platforms over here that
431. 15:38actually hit the chimp, and that's why
432. 15:40you get purchases and conversions. And
433. 15:42over here, you got chat GBT dealing with
434. 15:45the rational human one that rarely makes
435. 15:47a buying decision at all, if ever.
436. 15:50So, there you go. That's my thoughts on
437. 15:52it at the moment and where we are with
438. 15:53chat GBT ads. I know I think really my
439. 15:56advice to you at the moment is I go set
440. 15:58an account up, have a little play around
441. 15:59on it, put $50 on it, 50 pounds on it,
442. 16:02have a look, just be a bit familiar with
443. 16:03it. The rest of the guys, like the ones
444. 16:05at Money Penny where they're putting
445. 16:07huge volumes of money against it, great
446. 16:09guys, keep putting the money against it.
447. 16:10You can afford it. You need to pioneer
448. 16:13this so that you know, you can be at the
449. 16:15absolutely bleeding edge of this,
450. 16:17helping Open AI refine the advertising
451. 16:20system to get it to work. I genuinely
452. 16:22don't think it's going to really work
453. 16:24until they have the visibility of where
454. 16:26we are on the internet when the pixels
455. 16:29are in enough sites to see what you're
456. 16:30doing off the platform. But, for
457. 16:32everybody else, I'd open the account,
458. 16:34play around with it a little bit, and
459. 16:35that's it. Just be Just be ready to go
460. 16:38when it really starts to work. But, at
461. 16:40the moment, it's probably not a viable
462. 16:42platform for most advertisers other than
463. 16:44doing a few experiments on it, just
464. 16:46keeping yourself out there. Now, I've
465. 16:47got all the batteries have gone anyway.
466. 16:49If you've got problems with scaling your
467. 16:50ads and they just don't seem to be going
468. 16:51anywhere and you're struggling and it
469. 16:53just seems so hard to do anything with
470. 16:55Google Ads and you're at the point you
471. 16:56think, "God, I've turned this thing off.
472. 16:58It just doesn't work." It genuinely does
473. 17:00work. The problem is that you're not at
474. 17:01the top of the funnel, throwing people
475. 17:03in the top so they can come out the
476. 17:04bottom of the funnel. This whole idea of
477. 17:06just being able to advertise at the
478. 17:07bottom of the funnel is really it's the
479. 17:10old way. It's gone. The new way is we
480. 17:12need to put people in at the top of the
481. 17:13funnel on YouTube, on Meta, everything
482. 17:16like that to make them aware of you,
483. 17:18bring them down the funnel, so when they
484. 17:19get to the bottom on Google Ads, you've
485. 17:21got this larger warm audience of people
486. 17:24that you can actually convert. So, as I
487. 17:25said, just nip on to scan the QR code on
488. 17:28screen here somewhere.
489. 17:30Go visit clickconvertmarketing.com
490. 17:32or give us a call up or drop us an
491. 17:34email.
492. 17:35[email protected].
493. 17:37Just do that and we'll jump on, we'll go
494. 17:39through the account with you, we'll show
495. 17:41you why it's working the way it is, how
496. 17:43it should be working, and if you like
497. 17:44what you see, we're here to help you. If
498. 17:46you don't, no problem. We're all adults,
499. 17:48just say no, and maybe a bit further
500. 17:49down the road we'll do something
501. 17:51together. But until next time, guys,
502. 17:53bye.