Mini-Episode: Real Talk on AI-Driven Ad Results

A mini episode while Peter’s on the road this week. Real user feedback on Meta’s AI Ad settings, and a creative advertising AI from Pinterest.  

Share Your AI Stories/Experiences

Email: [e-mail] 

LinkedIn: Peter Malick, Grace Mathews, InboundAV 

Instagram: @InboundAV

Facebook: @InboundAV

Subscribe to our newsletter 

Contact us

Chapters

00:00 – Intro

00:25 – Real User Experience With Meta AI Ads

01:55 – Pinterest’s More Targeted AI Strategy

03:52 – Why Specificity Matters in AI Adoption

05:16 – A Call for Listener Stories

Sources

“Pinterest tests an AI feature that lets advertisers turn their catalogs into shoppable collages” by Aisha Malik for Tech Crunch

 

Episode Transcript

(00:00):
Grace (00:00)Hello, Accelerated Velocity listeners. Welcome to what I'm calling episode 9.5. Technically it's episode 10, but it's going to be a bit more of a quick check-in because Peter is currently traveling, but we still wanted to keep things going and make sure we showed up with an episode for you this week. So Peter has a follow-up to share on our conversation last week about AI-driven ad campaigns. Peter, I'll let you take it away.

(00:01):
Peter (00:25)And we talked specifically about Facebook and the announcement that by the end of the year, somebody is going to be able to have Facebook create with AI the perfect ad campaign. could include video, could include images, and AI will design the whole thing. we talked about, well, you know, if every ad on Facebook is perfect, what does that mean? Because if everything's perfect, nothing's really going to stand out from anything else. You know, it's always a... ⁓
⁓ a backside to any of these huge claims I feel. And I always want to look at that backside. Like what could possibly go wrong? I love everything to go right, but I think it's also really interesting to look at what could possibly go wrong. So anyway, we've included a little blurb about that in our newsletter. And about 20 minutes after the newsletter was sent out, somebody on the newsletter emailed back and said like, yeah, well, I had...
I had Facebook use AI to find my custom audiences this past month and my business went down by 40%. So it's, I love to be a cheerleader of something that's great, but if something's not ready for prime time, I think it's really prudent to be realistic about it. Recognize the potential of whatever they're building. But at the same time,
If it's not really ready for use, recognize that too.
Grace (01:55)And I'll piggyback off of that with an example of a completely different approach to AI ad features and what we can glean from that. And that is Pinterest, who is testing out a new feature that automatically turns product catalog images into dynamic AI generated collages. So here's a quick example. Imagine a fashion brand who has a product catalog up there with
clothes and bags and whatnot, AI takes those product photos and arranges them into stylized collages, maybe by color or theme or to create an outfit to match what Pinterest users are already typically engaging with. It's a very Pinterest-y aesthetic. Early tests are showing that these collages are getting three times more clicks than standard product listings, which is an interesting signal about how people are responding.
to this type of creative AI content. What stands out to me is that this is a very focused and intentional use of AI. One narrow feature, which is clearly tied to user behavior and creative design, it's a big contrast to Meta's ambition of start to finish AI-built ad campaigns that they claim will be rolled out very soon.
And I'm not saying one approach is better than the other because it's very much an apples to oranges comparison. But I do think that it highlights the difference between all in one start to finish AI driven automation and specialized task specific AI tools. And that distinction doesn't just apply to ads. It also matters for marketing campaigns and service workflows and sales automation.
So there's a lot to explore in how we use AI to solve real problems without over-engineering the solution.
Peter (03:52)Yeah, you know, I think that's really poignant. And the reason is, I think that if you're asking AI to do a very specific thing, in other words, let's say right now you can sell puppies and you can sell software with a Facebook ad and they're going to write the ad and flesh it out for you, do the video, whatever it is, that might be a reach. But
If you're just going to sell puppies or have somebody adopt puppies, that's a very attainable goal because it's very specific. And to have the AI just go after that specific goal, I think is the way we should all start with AI. we're actually a HubSpot app that we're building for AI within HubSpot is exactly that. We're looking at specific use cases that a company might need to use AI for.
And it could have to do with lead generation. could have to do with one of the things we're looking at is creating a target account within HubSpot. In other words, a large company that you want to go after. And it's a very finite thing. are certain steps that would be very, very helpful to automate. And that's kind of the way we're looking at AI. I think your Pinterest example is perfect, because it sounds to me, and I haven't seen it.
but it sounds like something that could be really effectively done.
Grace (05:16)Yeah, and as we wrap up, just want to mention it's really great hearing about different AI approaches. So if you're listening now and you have a story about your experience with a specific AI tool, whether it's for a broad range of functionalities or a very isolated feature, we want to know how that's playing out for your business. What's working, what's not ready yet, let us know. And you can do that by contacting us through any of the details that I'm going to be linking in the show notes. And I believe that's it.
for our quick drop-in episode. As always, remember to subscribe, leave us a positive review, hopefully it's positive, and check out the show notes for our website, our socials, and anything else you might be interested in from us. Thanks again for tuning in, and we'll be back next week with a full episode.