WEBVTT - Pinterest CEO Bill Ready Talks Earnings

0:00:02.520 --> 0:00:11.920
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news chefs. Pinterest best day

0:00:12.080 --> 0:00:14.720
<v Speaker 1>in a year. We're currently up eleven percent. The company

0:00:14.760 --> 0:00:18.120
<v Speaker 1>reporting first quarter sales the top down assessments and forecast

0:00:18.160 --> 0:00:22.599
<v Speaker 1>revenue exceeding all street expectations too. Already, pinterest CEO joins

0:00:22.680 --> 0:00:24.520
<v Speaker 1>us now exclusively. It's great to speak with you, Bill.

0:00:24.560 --> 0:00:25.840
<v Speaker 1>What's driving the growth?

0:00:27.160 --> 0:00:30.080
<v Speaker 2>Thanks for having me on. So you know, we're really

0:00:30.160 --> 0:00:33.040
<v Speaker 2>encouraged by the growth on the platform. It's eleven straight

0:00:33.120 --> 0:00:36.920
<v Speaker 2>quarters of record high users, ten straight quarters of double

0:00:36.960 --> 0:00:39.280
<v Speaker 2>disser growth in users, and at the core of that

0:00:39.360 --> 0:00:41.840
<v Speaker 2>is that we've we've really turned Pinterest into an AI

0:00:41.920 --> 0:00:46.600
<v Speaker 2>powered shopping assistant that is really really winning with users,

0:00:46.880 --> 0:00:49.199
<v Speaker 2>especially gen Z that's now more than half our platform

0:00:49.240 --> 0:00:53.600
<v Speaker 2>and our fastest growing demographic pictures is where gen Z

0:00:53.720 --> 0:00:56.640
<v Speaker 2>goes to shop, and that's been really great for users,

0:00:57.160 --> 0:00:59.120
<v Speaker 2>and increasingly advertisers're leaning into that and you see that

0:00:59.200 --> 0:01:00.000
<v Speaker 2>reflecting our results.

0:01:00.600 --> 0:01:04.080
<v Speaker 1>So ad targeting improves, we can see more efficiency, but

0:01:04.400 --> 0:01:08.000
<v Speaker 1>look at getting eighty billion monthly searches, as you say,

0:01:08.040 --> 0:01:12.000
<v Speaker 1>record hive users. What is the AI strategy here longer term?

0:01:12.040 --> 0:01:14.200
<v Speaker 1>Because already you're seeing the efficiency in a flywheel.

0:01:15.480 --> 0:01:18.360
<v Speaker 2>That's right. So we've shared that we have more than

0:01:18.400 --> 0:01:22.080
<v Speaker 2>eighty billion searches per month on the platform, which would

0:01:22.120 --> 0:01:24.760
<v Speaker 2>have been unbelievable to say, even a few years ago.

0:01:25.360 --> 0:01:27.839
<v Speaker 2>And we've done that by really making it a visual

0:01:27.880 --> 0:01:30.800
<v Speaker 2>first platform. So much of shopping is a visual journey,

0:01:30.800 --> 0:01:32.320
<v Speaker 2>and so we are really focused on that. And I

0:01:32.360 --> 0:01:34.759
<v Speaker 2>think in the world of AI, you've had a lot

0:01:34.760 --> 0:01:37.360
<v Speaker 2>of discussion around general purpose winners, but you're also seeing

0:01:37.400 --> 0:01:39.280
<v Speaker 2>specialization play out, and you're seeing that play out and

0:01:39.560 --> 0:01:42.399
<v Speaker 2>sort of consumer versus enterprise. But even within consumer, there

0:01:42.400 --> 0:01:43.840
<v Speaker 2>are a lot of different use cases and we've been

0:01:43.880 --> 0:01:47.319
<v Speaker 2>really focused on shopping and visual search and discovery, and

0:01:47.400 --> 0:01:49.920
<v Speaker 2>we're using AI to power that really based off the

0:01:50.000 --> 0:01:52.520
<v Speaker 2>curation signal on our platform, and we're seeing that really

0:01:52.560 --> 0:01:56.960
<v Speaker 2>really resonate with users. And o those eighty billion searches

0:01:57.000 --> 0:01:59.160
<v Speaker 2>per month, more than half of them are commercial, which

0:01:59.200 --> 0:02:02.840
<v Speaker 2>is a much more magnificant skew towards commerciality then most

0:02:02.960 --> 0:02:05.480
<v Speaker 2>chatbots would have, and that you would see in sort

0:02:05.480 --> 0:02:08.800
<v Speaker 2>of historical general purpose search. So it really has become

0:02:08.800 --> 0:02:11.040
<v Speaker 2>a great shopping destination for our users.

0:02:11.480 --> 0:02:14.120
<v Speaker 1>It's interesting those chatbots had been sort of some of

0:02:14.160 --> 0:02:17.280
<v Speaker 1>the investor anxiety that this is where competition is coming from.

0:02:17.320 --> 0:02:19.520
<v Speaker 1>In visual search, it is Google, it is opening on

0:02:19.520 --> 0:02:23.960
<v Speaker 1>the others. But how dependent are you on are the frenemies?

0:02:24.000 --> 0:02:27.600
<v Speaker 1>Shall we say, for the large language models, you're training

0:02:27.600 --> 0:02:31.959
<v Speaker 1>your own, but how much you're using underlying models of others, that's.

0:02:31.880 --> 0:02:34.840
<v Speaker 2>Right, So we primarily run our own compact fit for

0:02:34.919 --> 0:02:36.480
<v Speaker 2>purpose models. I think this is one of the things

0:02:36.480 --> 0:02:38.640
<v Speaker 2>that you will see over time is there's been so

0:02:38.720 --> 0:02:42.520
<v Speaker 2>much discussion around AGI and general purpose models, but they

0:02:42.520 --> 0:02:45.400
<v Speaker 2>can be really expensive to run, and we've demonstrated it

0:02:45.440 --> 0:02:48.000
<v Speaker 2>through our own compact fit for purpose models as well

0:02:48.000 --> 0:02:51.880
<v Speaker 2>as leveraging open source and retraining that off of our

0:02:52.000 --> 0:02:54.920
<v Speaker 2>unique data set that we're able to see comparable or

0:02:54.960 --> 0:02:58.560
<v Speaker 2>better results at oftentimes less than ten percent of the cost.

0:02:58.639 --> 0:03:01.239
<v Speaker 2>And in terms of shopping, we've shared that we're able

0:03:01.240 --> 0:03:04.480
<v Speaker 2>to get to thirty percent better relevancy of recommendations on

0:03:04.520 --> 0:03:07.320
<v Speaker 2>shopping for our users than what we'd see from leading

0:03:07.360 --> 0:03:10.360
<v Speaker 2>off the shelf proprietary models, which just really gets to

0:03:10.880 --> 0:03:13.639
<v Speaker 2>the uniqueness of pinterest as a shopping destination. And it's

0:03:13.639 --> 0:03:17.320
<v Speaker 2>important to note that what makes our models so powerful,

0:03:17.520 --> 0:03:19.960
<v Speaker 2>it's not just the models themselves. The AI doesn't have

0:03:20.080 --> 0:03:23.160
<v Speaker 2>taste or style. Humans have tastes in style, and so

0:03:23.200 --> 0:03:25.720
<v Speaker 2>people come to Pinterest to figure out their taste, to

0:03:25.760 --> 0:03:27.639
<v Speaker 2>figure out how they want to put outfits together, and

0:03:27.680 --> 0:03:29.320
<v Speaker 2>we're able to learn from that and make better and

0:03:29.320 --> 0:03:31.560
<v Speaker 2>better recommendations that are aligned to a user's taste. And

0:03:31.560 --> 0:03:33.760
<v Speaker 2>the result of that, as usually say things like well,

0:03:33.800 --> 0:03:36.280
<v Speaker 2>Pinterest just gets me. And that's something that we think

0:03:36.280 --> 0:03:39.200
<v Speaker 2>we're doing really uniquely here at Pinterest and where we

0:03:39.280 --> 0:03:42.000
<v Speaker 2>just have totally different signal based on what our users

0:03:42.040 --> 0:03:44.600
<v Speaker 2>do on the platform, and that's what lets us take

0:03:45.000 --> 0:03:48.240
<v Speaker 2>much smaller but really fit for purpose models and deliver

0:03:48.320 --> 0:03:49.320
<v Speaker 2>better results for users.

0:03:49.640 --> 0:03:52.280
<v Speaker 1>It's not just a signal on your platform. Though you've been

0:03:52.440 --> 0:03:55.520
<v Speaker 1>doing M and A talk to us about TV scientific

0:03:55.920 --> 0:03:58.800
<v Speaker 1>and while you're going into performance driven sort of TV ads.

0:04:00.240 --> 0:04:04.560
<v Speaker 2>So we now objectively six hundred and thirty million plus users,

0:04:04.600 --> 0:04:06.360
<v Speaker 2>you know, in eighty billion starches a month, more than

0:04:06.360 --> 0:04:08.200
<v Speaker 2>half of them being commercial, have one of the highest

0:04:08.240 --> 0:04:11.640
<v Speaker 2>commercial intent platforms anywhere in the world. And so far

0:04:11.840 --> 0:04:15.600
<v Speaker 2>we've been making that a really great platform for advertisers

0:04:15.640 --> 0:04:18.359
<v Speaker 2>to connect with users here on Pinterest, but we're also

0:04:18.440 --> 0:04:20.880
<v Speaker 2>looking at how we can make it so that we

0:04:20.920 --> 0:04:23.200
<v Speaker 2>can help those advertisers meet those users in other places.

0:04:23.240 --> 0:04:26.080
<v Speaker 2>Connected TV being you know, one of the fastest growing

0:04:26.120 --> 0:04:30.360
<v Speaker 2>areas of a of ad demand. It's projected to surpass

0:04:30.440 --> 0:04:34.039
<v Speaker 2>linear TV in twenty twenty eight, and we're able to

0:04:34.040 --> 0:04:37.200
<v Speaker 2>help advertisers show more relevant ads that deliver better returns

0:04:37.240 --> 0:04:40.480
<v Speaker 2>for advertisers on TV and users get to see things

0:04:40.520 --> 0:04:42.800
<v Speaker 2>that are actually helpful and useful to them. And we

0:04:42.839 --> 0:04:45.520
<v Speaker 2>showed on this most recent call that when we add

0:04:45.560 --> 0:04:49.120
<v Speaker 2>Pinterest audience data to TV scientifics capabilities, the company that

0:04:49.360 --> 0:04:52.720
<v Speaker 2>we acquired in Q one, we see a sixty five

0:04:52.800 --> 0:04:56.359
<v Speaker 2>percent improvement in the purchases that result from the ads

0:04:56.400 --> 0:04:58.960
<v Speaker 2>that are shown, which is great for advertisers, but also

0:04:59.040 --> 0:05:01.120
<v Speaker 2>just means as a user, when you're watching connective TV,

0:05:01.160 --> 0:05:03.119
<v Speaker 2>you actually see more of thesa say, oh, that's actually

0:05:03.120 --> 0:05:04.440
<v Speaker 2>a product that would be useful for me, and so

0:05:04.440 --> 0:05:06.599
<v Speaker 2>it's a great thing for users too. Thanks.

0:05:06.640 --> 0:05:10.000
<v Speaker 1>It's interesting, it's almost feels like diversification. Briefly, Bill, but

0:05:10.000 --> 0:05:13.720
<v Speaker 1>there's also this moment of gen z and teen bands.

0:05:14.240 --> 0:05:16.480
<v Speaker 1>How do you see that as your role in social

0:05:16.520 --> 0:05:17.160
<v Speaker 1>media right now?

0:05:17.200 --> 0:05:21.480
<v Speaker 2>Briefly, yes, So we've talked about this before, and I've

0:05:21.480 --> 0:05:23.920
<v Speaker 2>been very out spoken on this publicly. When I came

0:05:23.960 --> 0:05:26.359
<v Speaker 2>into Pinterest's CEO nearly four years ago, I wanted to

0:05:26.400 --> 0:05:28.719
<v Speaker 2>prove that there was a more positive business model possible

0:05:28.720 --> 0:05:31.400
<v Speaker 2>in social media, and so youth online safety has been

0:05:31.480 --> 0:05:33.960
<v Speaker 2>very out in front for us. We made Pinterest private

0:05:33.960 --> 0:05:37.719
<v Speaker 2>only for users under sixteen approximately three years ago. We're

0:05:37.720 --> 0:05:39.800
<v Speaker 2>the first and still the only platform to do that,

0:05:40.360 --> 0:05:44.680
<v Speaker 2>and so I've publicly advocated for the fact that social

0:05:44.720 --> 0:05:46.880
<v Speaker 2>media as currently configured, it is not safe for users

0:05:46.920 --> 0:05:49.120
<v Speaker 2>under sixteen. So we turned off social features three years ago,

0:05:49.360 --> 0:05:51.000
<v Speaker 2>and we actually think it's quite an encouraging thing to

0:05:51.040 --> 0:05:53.640
<v Speaker 2>see regulators around the world starting to really pay attention

0:05:53.680 --> 0:05:53.839
<v Speaker 2>to that.

0:05:54.440 --> 0:05:55.960
<v Speaker 1>Do you have to leave it there? Thank you so

0:05:56.080 --> 0:05:58.400
<v Speaker 1>much for talking to us through your numbers and your focus.

0:05:58.400 --> 0:05:59.599
<v Speaker 1>We already Pinterest CEO