Today in Digital Marketing - Hey Copywriters... What is the Most Effective Letter?

Episode Date: September 13, 2024

Threads seems to REEEEEALLY like engagement bait questions. And the reason it does is far more devious than you thought. Plus: Why is Meta changing your campaign destination without your consent? Google Ads locks down your audiences. And the ad format consumers are really starting to get tired of.Links to today's stories📰 Get our free daily newsletter📈 Advertising: Reach Thousands of Marketing Decision-Makers🌍 Follow us on social media or contact usGO PREMIUM!Get these exclusive benefits when you upgrade:✅ Listen ad-free✅ Back catalog of 20+ marketing science interviews✅ Get the show earlier than the free version✅ “Skip to story” audio chapters✅ Member-only monthly livestreams with TodAnd a lot more! Check it out: todayindigital.com/premium✨ Premium tools: Update Credit Card • CancelMORE🆘 Need help with your social media? Check us out: engageQ digital🌟 Rate and Review Us🤝 Our SlackUPGRADE YOUR SKILLSGoogle Ads for Beginners with Jyll Saskin GalesInside Google Ads: Advanced with Jyll Saskin GalesFoxwell Slack Group and CoursesToday in Digital Marketing is hosted by Tod Maffin and produced by engageQ digital on the traditional territories of the Snuneymuxw First Nation on Vancouver Island, Canada. Associate producer: Steph Gunn.Some links in these show notes may provide affiliate revenue to us.Our Sponsors:* Check out Kinsta: https://kinsta.comPrivacy & Opt-Out: https://redcircle.com/privacy

It is Friday, September 13th. Today, Thread seems to really like engagement-based questions, and the reason it does is far more devious than you thought. Plus, why is meta changing your campaign destination without your consent? Google Ads locks down your audiences, and the ad format consumers are really starting to get tired of. I'm Todd Maffin. That's ahead today in digital marketing. A couple of nights ago on my personal threads account, I posted the question,

what's your favorite word starting with the letter B? Nobody replied. So I followed up a minute later with, hey humans, what's the best hand to use? Still nothing. So I added about a dozen more questions. Reply with your favorite color. What one TV show should I never watch? Lavender or orange, please weigh in, people with noses. Nobody ever replied to the posts, except for my friend Rob who liked all of them. But I was proud of them. It felt like an art piece.

None of this will make sense to you unless you've spent a decent amount of time on the platform in the last month or two. It's become clear that simple, low-effort engagement bait, questions like, what one SEO tactic should I avoid, really gain traction. A tech columnist at Business Insider tested this this week by posting a series of provocative questions on threads to see how users would react. The outcome? Her posts attracted a surge of comments, indicating that posting basic, simple questions can significantly boost your visibility on threads. Or to be more precise, simple questions often attract a lot of comment replies. And that's what Thread seems to prioritize. Posts with lots of comments, compared to those with many likes or shares. But there's another layer. Meta is working on its own AI chatbot and needs real human conversations to train it.

By encouraging posts that spark discussions, Meta gathers valuable data for its AI models. This might be why question-driven posts are getting an extra push in the algorithm. So if you're trying to expand your brand's reach on threads, posing thoughtful questions could be your best bet. A good friend of this podcast, Rock Ladnick of Flat Circle Ads, reports today on social media that there seems to be a nasty bug in the Meta Ads Manager. He says the bug is changing

ad set conversion locations from whatever you've set, like website, and forcing it to messaging apps instead. He says this is happening to newly created ad sets and old ones. At our deadline, Meta's status update page was reporting that everything was fine, though take that page with a grain of salt since it's frequently either behind the time or goes completely unaware that

bugs are happening. Either way, might be a good idea to check your campaigns to make sure this hasn't happened to you. For what it's worth, people do complain about bugs a lot on the big platforms. But in the last day, we have noticed an uptick in people reporting on social about new bugs they're finding in Meta's ad manager. We reached out to Meta for comment. They tend to not reply to specific questions about their ad platform, but we will update if there's any big news. While Google's legal battles with the U.S. government grabs headlines, the company continues to launch new tools. This week, Google

Ads introduced confidential matching for your uploaded first-party data. Google says this offers better data protection for your uploaded customer lists. It uses a new system they call trusted execution environments, which the company says should make the data less likely to be vulnerable to hacks, which my first thought was, wait, this stuff wasn't protected before? No, it was, but this is stronger or something. Google says it's sort of like end-to-end encryption in that it isolates business information during processing so that no one, including Google, can access the data. And this new system is now turned on by default. Google also is offering attestation.

This means that third-party audits can verify that data is processed according to those regulations. So why does it matter? Advertisers do worry about data leaks when uploading valuable customer data. And with stricter regulations on consumer information, data security is a growing concern. Some media agencies told Digiday, which reported on this today,

that confidential matching might encourage more ad spending from cautious marketers, especially within tightly regulated industries, like the pharmaceutical or finance sectors. Quoting from Digiday's piece, quote, additionally, those present thought the newly launched data encryption features could assure advertisers that the platform

wouldn't absorb a brand's first-party data in a manner that could later benefit others, including direct rivals, unquote. A new study by Freewheel reveals that issues like latency and poor ad placement are annoying viewers and impacting advertisers on streaming platforms. Freewheel surveyed 420 participants about different ad experiences, no ads, latency issues, and unnatural ad breaks. Nearly 80% found delayed ad load times annoying, affecting how they perceived the program, the ads, and even the advertising brands. Shows without latency were rated 8% higher, and those ads and brands saw quality boosts. Unnatural ad breaks bothered 71% of viewers, making ads seem more intrusive and reducing brand recall by 14%.

Advertisers are paying the same, but getting less impact. Interestingly, the ads didn't reduce program enjoyment. Viewers rated shows with and without ads equally at 6.3 out of 7. Viewers were frustrated by ad slates. Those are the blank screens saying we'll be right back when ads fail to load. Up to 25% of ad slots on streaming channels go unfilled due to these issues. Ads near slates were rated lower. Freewheel said that was hurting brand perception. Programs without these slates saw a 31% increase in joy that was hurting brand perception. Programs without these slates saw a 31% increase in joy. And if you're wondering how they measured joy, it was by training cameras on people's faces using a system called facial coding. Freewheel suggests optimizing

supply paths and connecting directly with publishers to reduce slates and latency. And buying ad space from suppliers with better ad technology can speed up that delivery. Gemini Data, an enterprise AI platform provider, has sued Google for using the name Gemini for its AI services. Earlier this year, Google rebranded its generative AI from Bard to Gemini after introducing the Gemini model family. Gemini Data claims that Google did this despite knowing that Gemini was already their registered trademark. The U.S. Patent and Trademark Office doesn't allow companies in the same industry to use the same name if it might confuse consumers. Google tried to trademark Gemini, but

it was provisionally denied. The USPTO said the name was too similar to existing trademarks like GeminiData's. After the refusal, GeminiData alleges that Google secretly tried to buy the rights to the name. They were contacted by an anonymous party inquiring about the Gemini brand, which they suspect was acting on Google's behalf. Interestingly, if you ask Google's Gemini chatbot about the trademark issue, it reportedly acknowledges the conflict, calling it a developing situation. This isn't the first time Google has faced such issues.

Back in 2009, they named their programming language Go, even though there was already a language called Go. And Google isn't alone in this. Meta recently settled a trademark lawsuit after changing its name from Facebook, showing that big tech companies often tread on existing trademarks. Speaking of Meta, the company is updating how it marks content on Facebook, Instagram and threads that has been edited or created with generative AI. And by updating, I mean hiding.

Instead of placing the AI info tag directly under a user's name, as in the past, the label will now appear under a menu at the top right corner of images and videos. Users will have to tap this menu to see if AI was used. This change follows complaints from creators and photographers who said their real photos were incorrectly tagged as AI-generated. Previously, Meta applied the AI info tag to all AI-related content, whether it was slightly edited using tools like Photoshop or entirely generated from a prompt.

The company says this update aims to, quote, better reflect the extent of AI used, unquote, across images and videos on its platform. As AI editing tools become more advanced, distinguishing between real and altered images could become more challenging. Meta hasn't disclosed which detection system it uses, but mentioned industry signals like Adobe's content credentials metadata. Wow, a lot of feedback on the AI thing

that we played at the end of yesterday's episode. Lots of emails. Let me assure you, that was completely AI generated. We were not pulling a joke there. That's out of Google's new Notebook LM. And we're going to do another one at the end of today's show. This one about attribution models. And again, just in case you didn't hear yesterday's show,

this is Google's new system that kind of creates a podcast episode, I guess. It creates a conversation between two people. It is past Uncanny Valley. It is past Uncanny Valley. It is so good. And all I did was I gave it the Wikipedia page URL for attribution Island. Our associate producer is the intrepid Stefgan. Our production coordinator is Sarah Guild. Ad coordination by Red Circle. Mark Blevis is at feeling when you swipe away all the notifications in a single swoop. I'm Todd Maffin.

Have a restful weekend. I will see you on Monday. So today, we're diving deep into marketing attribution. Okay. We're going to figure out how to make sure that credit is going where credit is due. And to help us navigate all this, because it can get a little complex. It can. We have our expert here.

So by the end of this, you're going to be fluent in attribution models. I love it. And you can think about this like in a play, right? Okay, yeah. The lead actor, they're the one getting all the applause, but who's behind the scenes? The director, the lighting crew, who even wrote the music. Yeah, they're all important.

They all contribute to the experience. So attribution is kind of like acknowledging those, you know, maybe a little less obvious, but still important players in your marketing campaigns. Oh, I like that. It's about the whole picture, not just the final bow. Now, I was reading something the other day. Oh, I like that. It's about the whole picture, not just the final bow. Now, I was reading something the other day.

You know how it is. Everyone's got their favorite article. Right. And it really made me think about how much has changed, even in the past few years, in marketing. Oh, absolutely. I mean, it used to be pretty simple. Right. You had your TV ad. You had your newspaper ad.

Maybe a radio jingle if you were feeling crazy. If you were feeling fancy. Right. You had your TV ad, you had your newspaper ad, maybe a radio jingle if you were feeling crazy. If you were feeling fancy. Yeah. But now it's a whole other world. Totally. It's like everyone's on their phones, on their computers. They're going back and forth between websites, social media, email, like they're having five different conversations at once. Yeah. Yeah. And you're just trying to figure out which one of those conversations convinced them to buy those shoes, you know? Exactly. And that's exactly where these marketing attribution models come in. They help you untangle all of that, that chaotic customer journey. It is chaos. So you can be smart about how you're spending your marketing dollars. Okay. So let's say we're looking at like our analytics. Yeah. You know, trying to see what led to a sales bump.

It'd be easy to just be like, oh, what was the last thing they clicked before they bought it? Maybe it was, like I said, a social media ad or a Google search result. Yeah. That's basically the last click model, right? It is. And it can be helpful. Yeah. But it's a little bit like giving all the credit for a movie to the special effects team.

You know, the explosions are cool. Right. But what about the writing, like giving all the credit for a movie to the special effects team. Okay. You know, the explosions are cool. Right. But what about the writing, the acting, the directing? It takes a village or a film crew, I guess. Exactly, exactly. There's a whole team of marketing efforts at play, just like in movie making. For sure, for sure.

And that's where this fractional attribution idea comes in. Okay, yeah. Tell me more about that. So instead of dumping all the credit onto that last touch point, it tries to spread it out a bit more fairly. Okay. So how does that actually work? Like, is it super complicated? It can get a bit, you know, in the weeds, but there's different ways to do it. One that's pretty common is called a time decay. Okay. Time decay. Yeah. So think about it like this. Let's say you see a billboard for, I don't know, a new brand of coffee on your way to work. Okay. Yeah. You don't like slam on the

brakes and run into the coffee shop. Right. I mean, I hope not. Hopefully not. But maybe later that day you're at the store and you see the coffee again and you decide to buy it. Yeah. It's fresh in your mind. Exactly. So that billboard, even though it was earlier in the day, still played a part in that purchase. So with time decay, are we basically saying that the marketing that happens closer to the actual sale gets a bigger chunk of the credit? You got it. Then there's a linear attribution. And that's basically like every touchpoint gets an equal slice of the pie. It's like no matter what, everyone on the team gets a trophy. I like trophies. Right. It's helpful

if you think that each interaction is kind of equally important in the grand scheme of things. That makes sense. What was that other one? Customer. Oh, yeah. Customer credit. Yeah. That was interesting because you actually get to customize it a bit more based on what you already know about your customers. Ah, so you can be like, OK, for this type of customer, this touch point tends to matter more or whatever. Exactly. If you've got a good sense of your customer data and how they act, that level of customization

can be really powerful. That does sound powerful, but maybe a little advanced. I'm not going to lie. Definitely. Now, you mentioned algorithmic something earlier, and that sounds like really high tech to me. Algorithmic attribution. Yeah. Is that where the robots come in and do all the thinking? Pretty much. Think of it like trying to solve one of those giant jigsaw puzzles. Yeah. Except instead of you doing it yourself, you have this machine and it can analyze all the shapes, the colors colors and just put the whole thing together for you.

OK, that'd be nice. Right. So it's like that with marketing data platforms like Google's double click. They use these really sophisticated algorithms to make sense of massive amounts of data way more than we could ever do on our own. So instead of me sitting here with a calculator being like, okay, this touch point gets this much credit. The algorithm just figures it all out. What kind of stuff would it even be looking for? So let's say someone clicks on one of your display ads, but like weeks ago, the algorithm might be able to see that people who clicked on that ad weeks ago are twice as likely to buy your product eventually, even if they didn't do it right away. Oh, wow. It's like those connections that we might miss, but the algorithm can just pick up on. It's like it sees the matrix or something. Kind of. It's showing you the impact of those campaigns, even if they don't lead to an

immediate sale. Big picture stuff. That's pretty amazing. Yeah. Before we move on too much, can we at least touch on customer-driven something? Customer-driven attribution, sure. Customer-driven attribution. It's like we've been looking at maps this whole time. Right. And then we're like, wait, why don't we just ask someone who actually took the trip? Yeah, exactly. What route did you take? It's going straight to the source. Right. Sometimes the most useful insights you're going to get are going to come from just asking your customers, what made you decide to buy this? Or how did you even hear about us in the first place? It seems so obvious when you say it like that. Yeah.

But I feel like sometimes you can get really caught up in all the data and forget about the humans behind it. Totally. Totally. It's easy to get lost in the numbers. But remember, at the end of the day, the data should be helping you make decisions, not the other way around. And customer feedback can really add a whole other level of understanding to your attribution models. For sure.

OK, so we've talked about last click attribution, all these different fractional models, these really smart algorithms. Yeah. But like if someone's listening to this, what's the main thing you want them to remember? Honestly, I think the biggest thing is just remembering to think about the customer experience. Okay. Not just clicks. Like what is their journey actually like?

So going back to that scavenger hunt idea we talked about. Yeah. It's not just about getting to the prize at the end, right? Yeah. It's about the fun of to the prize at the end. Yeah. It's about the fun of figuring out the clues and getting closer and closer. Exactly. Each step, each little interaction, it's all part of the story. Yeah.

And the better you understand how those pieces fit together, the more engaging you can make the whole thing. Now, you mentioned counterfactual procedures earlier. Yeah. And I have to admit, that sounds kind of intimidating. I know, right? Can you explain that a little bit? What does that even mean when it comes to marketing? So it sounds complicated, but it's a really simple idea at its core.

It's basically just asking, what if? Like, what if someone hadn't seen that one ad? Would they still have bought the thing? Or what if they didn't click that one ad? Would they still have bought the thing? Or what if they didn't click that email? What would have happened? So it's like you're going back in time and changing one little thing. Exactly. To see if it changes the ending.

You got it. And then by looking at those different possibilities, you can start to understand how much each marketing effort is actually doing. That's really cool. Yeah. It helps answer those questions like, would this have happened anyway? Or was it because of something we specifically did? That's like next level stuff right there. Okay. This has been awesome. It has. I feel like we've gone from reading the last page of a mystery to understanding the whole plot. Yeah. And I think that's what we were going for. I think so too. Yeah.

Because once you get the story, you can write an even better one. I like that. So to everyone listening, next time you're looking at all that marketing data, just remember, it's not just numbers on a screen. Yeah.

It's a story. It is. And if you learn how to read it, and attribution is a big part of that, then you can unlock some really incredible things. Absolutely. That's all we have for this deep dive. Until next time,

happy analyzing everyone.