fbpx

Colette Nataf is the founder and CEO at Lightning AI. In today’s conversation, Colette describes 3 potential scenarios: best case, worst case and somewhere in the middle – and talks through how we might prepare for the worst. She illustrates the possibilities with the experience of a company that is using web-based flows for their UA today in a pre-iOS-14 world, and outlines exactly how the different aspects of user acquisition might change come January. 






ABOUT COLETTE: LinkedIn  | Twitter | Medium | Lightning AI




ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube


KEY HIGHLIGHTS

🔱  The best, the worst and the middle case scenarios for iOS 14

🔒 How a privacy app manages growth without a single SDK

🌱 It is possible to grow without sophisticated attribution

🧮 Old school tracking with UTM parameters

🍂 The longer the funnel, the greater the drop-offs

💨 Targeting without conversion data doesn’t exist

🌎 Macro trends impact CPMs

🎬 Why it is a good idea to prep the user for a new flow

🔚 It is end times for broad targeting

👬🏻 No user level data means lookalikes will not work

🧶 Interest-based targeting will become more prominent

🗓️ How to set up a web-based flow

🗄️ The critical importance of stitching together data from different sources

🚫 Things that marketers should not attempt

💍 Marriage of convenience: marketing courts engineering

🤖 Bringing data science into the fold

🌪️ Prepare for the worst, even though it may not happen

🌟 Facebook will not remain a superstar

🌈 Which are the new channels on the block?

🖥️ The potential return of display advertising

🦋 How marketing teams will change and evolve.

KEY QUOTES

Prepare for the worst

“The worst case scenario is we’ve lost all tracking; we have no data; MMPs don’t work at all; we can’t use any dashboards anymore; everything looks like it’s organic. It’s a mess. In our case, we’re preparing for all three of those—obviously, we’re hoping for the best, but we do want to have solutions in our back pocket in case it’s something in between, or it really is the worst case scenario, and all the tracking is gone.”

How it is possible to grow without attribution

“Because they don’t have the Facebook SDK, they don’t have an MMP—they need to do things flying blind and figure out ways to understand attribution without that power. 

So this is really my example of the worst case scenario, because they’re already operating in that worst case scenario world, where we don’t have any attribution at all. What they do, and what they’ve found as a solution, is that they run Facebook ads on mobile devices, but to a landing page on mobile web. On the landing page, the user enters their phone number and they’re sent a text message with the link to install and download the app.”

A good time to test the waters

“I think that if we assume iOS 14 is going to roll out in January—January is historically the lowest CPM month, so it’s a good time to be experimenting; which is great because we have some flexibility as marketers.”

How retracing our steps may be the way forward

“When I started advertising, broad was a crazy idea. We were like: “Why would we launch a target that has no targeting? That’s crazy.” It wasn’t even that long ago that these ideas were brand new. So it’s a little bit of a step backwards, but sometimes you have to go backwards to go forward.”

“What you need to be able to do is take the parameters in the URL—say it came from this specific campaign, this targeting, this ad—and feed that into your database and connect it with that link that’s coming through the text message. So you need really: a personalised link in that text message, you need to have their cell phone number, and you need the URL parameters and all that needs to feed in somewhere into a database.”

There will be a diversification of channels

“So let’s just assume that prices increase by 50% on Facebook to get an install. So what that means is that now Facebook is no longer the best option, and a lot of these other networks that people have said are interesting, but aren’t a big chunk of the budget become a lot more competitive. So things like Snapchat, TikTok, definitely; maybe even UAC increases budget.”

Marketing will need to include engineering and data science

“Marketers are great at a lot of things, and we’ve become really good at reporting analytics, but that is very different than becoming a data scientist or data engineer. And the truth is that if we’re going to have to pair data together from three, four or five different sources, that’s where you really need a data person on your team. You’d not want to be putting that together and deal with lookups and pivot tables. We’re all on Macs, and I think they would all break.”

FULL TRANSCRIPT BELOW

Shamanth: I’m very excited to welcome Colette Nataf to The Mobile User Acquisition Show. Colette, welcome to the show.

Colette: Hi, thanks so much for having me.

Shamanth: Yeah, Colette, I’m thrilled to have you, certainly because I loved the last episode that we did record. You talked about automation. And I like that because you are one of those people that has actually skin in the game; you’ve actually built out fairly elaborate automation systems for your company. And you know what you’re talking about. 

And I think it’s a very similar topic to what we’re talking about today, which is what’s happening with iOS 14. And what are some of the worst case scenarios that you’re preparing people for—because you’re clearly advising a lot of people, a lot of companies just now. And I would say, a lot of them can be freaked out—at least certainly, very many people we know are concerned about what might unfold. And certainly, the last time we spoke you told me about how you are preparing people for what could be the best case scenario. 

I think a good place to start would be to look at the lay of the land; what are some of the scenarios that you are preparing people for?

Colette: In terms of scenarios, it starts with the best case. So the best case is: all the MMPs built out the perfect prediction model; they all have probabilistic algorithms that match click data to install data; and they figured everything out. And everything just magically works. So that’s the best case; the best case is nothing changes. We as marketers just go about our day, we switch over to iOS 14, and everything’s perfect. And we have perfect data. And it all just works. That’s option one. 

Option two is somewhere in the middle: some stuff works; it works on some platforms; and we have perfect data for the Apple network, but maybe not so perfect for Facebook or for other networks; maybe bigger networks are better and other ad networks aren’t; or some kind of combination of getting imperfect data. So maybe like click works, but view-through doesn’t, something like that. So that’s the middle of the road. 

The worst case scenario is we’ve lost all tracking; we have no data; MMPs don’t work at all; we can’t use any dashboards anymore; everything looks like it’s organic. It’s a mess. In our case, we’re preparing for all three of those—obviously, we’re hoping for the best, but we do want to have solutions in our back pocket in case it’s something in between, or it really is the worst case scenario, and all the tracking is gone.

And we’ve got to really backtrack.

Shamanth: All three are possible. And I think it’s good to think about what happens in each of those scenarios. And certainly, you’ve seen how very different types of apps, very different types of companies might prepare and adapt for each of these things. 

Specifically, you mentioned the last time we spoke, you’re currently working with a company who does acquisitions without any SDKs. And you mentioned to me the last time that that’s an experience that could be somewhat close to the worst case scenario. So can you speak to this company, how they operate, and what some of the learnings might be from what they’re already doing?

Colette: Yeah, absolutely. I think it’s a great example. So we work with a company that is a privacy app. What that means is that they’re all about data protection, and they want to be able to keep all of your information private, and not allow large corporations to mine your data and use that information for whatever nefarious purposes they might have. 

I think that’s all well and good. But as every mobile app knows, you usually need advertising to get yourself off the ground. And also as mobile app marketers know, Facebook is a really great way to acquire users, especially for people who are concerned about privacy. A lot of people are on Facebook, and people are concerned about the data usage that is going on both on Facebook and, for people who are a little savvy, also on Instagram. 

So this company: it would go very much against their fundamentals to install the Facebook SDK, and they’ve actually taken it a step further. They don’t have any MMPs, they have no SDKs installed in their app, which is pretty incredible. They build everything in-house. And what that means is that because they don’t have any SDKs—

Because they don’t have the Facebook SDK, they don’t have an MMP—they need to do things flying blind and figure out ways to understand attribution without that power. 

So this is really my example of the worst case scenario, because they’re already operating in that worst case scenario world, where we don’t have any attribution at all. What they do, and what they’ve found as a solution, is that they run Facebook ads on mobile devices, but to a landing page on mobile web. On the landing page, the user enters their phone number and they’re sent a text message with the link to install and download the app. 

So what’s actually happening is that they’re kind of going way back in time, before any kind of mobile tracking was a thing. And they’re putting UTM URL parameters into that URL: they have your campaign name, the campaign ID, etc. And they’re linking that to the cell phone number when somebody installs the app. That’s how they marry those two data points together. 

So it’s going to be a little imperfect; maybe somebody gets their cell phone number wrong; maybe they leave the page and go directly to the App Store. But, at least you have some data, so you’re able to optimise your campaigns, you can figure out which of these ads is the best performer, and then they can actually use conversion optimizer on Facebook with the pixel, so when somebody enters in their cell phone number, and confirms, that is the event that they target. 

So, no SDKs and they’re still making it work. So I’m here to say it’s possible: this worst case scenario happening in the world where mobile marketers who made it work.

Shamanth: Yeah, I think that’s incredible. Also, because they didn’t have to do this, but it was an ideological stance, which I find to be incredibly rare in this day and world. They’ve built out this solution which is a very creative way of still growing—they’re not sacrificing growth at all. 

So out of curiosity, did prices change for them? What was the CPM like when they just ran a web-based flow?

Colette: Yeah, that’s a great question. So in terms of CPMs, I haven’t noticed a huge difference there: mobile app installs versus mobile web. The difference really comes down the funnel, and it’s really in that install event: there’s more steps to complete, you’re not going directly to the App Store, and so the install rates are low. And that’s just going to be true; there’s no way to avoid it. Any time you add steps in your funnel, you’re going to see drop offs. 

I think their install rate is probably half of what it would be if it were running directly to the App Store. It’s hard to tell also because, obviously, with click and view through attribution, there are some tangibles that we don’t know about. But I think that costs are definitely higher overall, because of that. 

And similarly, Facebook uses the conversion data to feed into their algorithms, so that they’re finding the right users. So if they don’t have that install data, it could be anybody. It could be people who like putting in their cell phone numbers, but don’t want to install any apps, and especially the privacy app. You don’t necessarily know who those people are going to be, or what things they’re going to be happy doing. 

But I think, overall, you’re adding to the funnel, and you have to expect that drop off. So there are going to be changes, not necessarily on the CPM level—maybe not even at the click level—but at that conversion rate to install, you’re gonna see a drop off and that’s going to make a difference. 

Shamanth: Yeah, so for a mobile marketer that is looking to explore the signup flow, what sort of delta should they expect in the CPI, if they shifted?

Colette: Yeah, I would expect up to double the CPI. So it’s a pretty big increase, and that’s just what happens when you don’t have any data.

Shamanth: Yeah. And clearly, this app has made it work. And do you anticipate that, because that sort of personalization goes away, maybe the CPMs will drop; maybe that’ll make up for the 2X evolution in the downstream conversion rates? Is that within the realm of possibility? How do you think about that?

Colette: It’s hard to know; there’s always supply and demand questions. Right now, we’re in the middle of October, and I’m seeing huge CPM increases that I assume from the election in the US. So there’s a lot of macro trends that are going on at the same time, so it’s hard to know. 

I think that if we assume iOS 14 is going to roll out in January—January is historically the lowest CPM month, so it’s a good time to be experimenting; which is great because we have some flexibility as marketers.

But that being said, if you’re making your budget right now for 2021, I’d pad January. I would just be aware of the reality that things are changing and we don’t know what that means. There’s no way to prepare 100% for something that is unknown. So hopefully we’ll see low CPMs in January; all the holiday stuff going to be done; elections are going to be done, maybe? 

Shamanth: One hopes!

Colette: That would be ideal. I think the reality is that we’re going to change our apps. And while you’re preparing for the worst case scenario, web-based flow, it’s a good idea to maybe change your end card. So you say: “Give us your cell phone number, and we’ll text you to install the app” or even have an example of what that looks like. 

This is a while ago now—six years ago—but when I worked for MileIQ, we actually, in our TV spot, we had a picture of someone sitting on the couch, watching the TV ad, and we would show the guy going to the App Store and installing the app. So, not necessarily a bad idea to explain what that flow is going to look like to prep the user first. So that when they actually get there, they’re not like: “This isn’t the App Store! Are you lying to me? What is this weird place where I am?”

Shamanth: Yeah, the App Store user flow isn’t intuitively clear for a lot of people. So certainly, I think that makes sense, especially if you’re watching it in a different context; be it a webpage or a TV ad. 

To switch gears a bit, how do you anticipate targeting to change? Should web-based campaigns become more prevalent?

Colette: Yeah, absolutely. So, right away, I would say, broad targeting is just not going to work. Broad works right now, because we have the data going in about who’s making purchases, who’s doing those downstream events. If the best thing that you have is something before install, that’s not gonna work for broad any more. So I think that’s going to be really different. A lot of advertisers rely on broad targeting to be their bread and butter, and, if that whole bucket is gone, you’re going to have to reallocate your spend to other places. 

The other thing that might not work as well—or even at all—are lookalikes. If we don’t have a way of using dynamic data that Facebook can pull in directly for purchases, for engaged users, it’s possible they might not work. And if they do work, it’s probably going to be either at the very low or very high end of the spectrum. I can imagine that either like a 1% is going to work very well, because it’s going to be a very targeted audience or like we need to have a splattering to cover everybody. And so you need that larger 10% in order to make it work. 

What I’m telling people—and what I’m also banking on—is that interest-based targeting is going to be a bigger and bigger chunk of that budget, and a bigger set of groups that people are going to be using. 

The good news is Facebook has so much data about everybody, which is why privacy apps make sense to advertise on Facebook. But Facebook knows everything about everyone pretty much, so I assume the targeting works. And when you have all interest groups, you can make sure that you’re finding exactly the right audience. The downside is that those are smaller groups, so you have to find more of them in order to make that successful. 

But I think definitely broad is going to be out. I don’t know how retargeting is going to work, but hopefully we can figure that one out. We’ll see. It’s going to be interesting.

Shamanth: Yeah. And for what it’s worth, I tell people that this is how we did things before lookalikes came along, and there is a world of that sort. It is certainly something we can go back to.

Colette: Absolutely.

When I started advertising, broad was a crazy idea. We were like: “Why would we launch a target that has no targeting? That’s crazy.” It wasn’t even that long ago that these ideas were brand new. So it’s a little bit of a step backwards, but sometimes you have to go backwards to go forward.

Shamanth: Yeah, totally. And let’s say, somebody had to set up a landing page flow of the sort you described. So the user goes there, enters their phone number and then gets a link to download the app. How easy would this be to set up? Are there off the shelf solutions? Let’s just say somebody was like: “Let’s start preparing.” What would be their first steps?

Colette: The good news is that there’s easy ways to make a web page. There’s a whole bunch of different places where you can make landing pages. And probably whatever push solution or email automation solution you are using also has some kind of SMS option. Twilio is pretty easy and cheap. So that part is easy. 

The hard part is connecting those two pieces together, because what you need to do is not just have a landing page where somebody can put in their phone number. Because if you just have that, the user is still lost; we don’t know where they’re coming from.

What you need to be able to do is take the parameters in the URL—say it came from this specific campaign, this targeting, this ad—and feed that into your database and connect it with that link that’s coming through the text message. So you need really: a personalised link in that text message, you need to have their cell phone number, and you need the URL parameters and all that needs to feed in somewhere into a database.

And that’s the hard part; what marketers can’t do on our own. So really to do that, you need a backend engineer, or some kind of data person who is a skilled programmer. I feel like it’s the kind of thing where I know enough to be dangerous, and could end up screwing up my database. You don’t want to do that! You have to call for reinforcements here. This is serious; this is writing information into the big bucket of truth—you want to make sure it’s actually right. So I would definitely try and get on your engineering team’s radar right now. 

And again, it might not be something that you even end up using. So that’s the reality. You might be building this; you might kind of do all this work; and then it doesn’t even matter. But I personally hedge on the side of caution, and so for me, being prepared and having this as a backup solution, I feel is a better option than just throwing up your hands and saying: “Well, January, we’re just lost.” So definitely, it’s time to start talking to your product manager, or your engineering team or someone who can help you out with this. Because you’re going to need some real backend dev; it’s not something that you can fully do on your own.

Shamanth: Yeah. Are there off the shelf solutions that might be sort of plug and play?

Colette: I don’t know any that are going to take the UTM parameters and then put it into your database. The only thing that I was thinking of is maybe Segment you could use, because Segment will store the data from the website or have the URL. And then maybe a smart data person could join that table with data downstream from the mobile app. I haven’t done it with Segment directly, so I’m not sure. Maybe mParticle has the same idea that they’re bringing data from these different places. And so you might be able to do it. But nothing like this exists because we never needed it.

Shamanth: Yeah. And like you said, we might never need it. 

Colette: Yeah. I don’t think anybody’s sitting at home like: “Oh, I’ve got a new business idea.” because it might be a terrible idea. I don’t think it is preparing to like put an SDK into your app to track something that may or may not be relevant. 

So the good news is, as marketers, we’re pretty good at building landing pages. That part we’ve got down. I think it’s just the rest of it; that’s the connectivity that makes this a little bit of a challenge.

Shamanth: Certainly. And all of this, obviously, we’ve talked about in the context of Facebook. What happens to non-Facebook channels, and how do you see budgets shifting? Should this be the prevalent dynamic that prevails?

Colette: Yeah, I think it’s going to be really interesting.

So let’s just assume that prices increase by 50% on Facebook to get an install. So what that means is that now Facebook is no longer the best option, and a lot of these other networks that people have said are interesting, but aren’t a big chunk of the budget become a lot more competitive. So things like Snapchat, TikTok, definitely; maybe even UAC increases budget.

I don’t know how UAC is going to work also, because there’s no targeting. So I don’t know about that one but let’s assume Google figures it out. 

I think that those kinds of other places are going to become a lot more competitive, because Facebook is no longer the star. That’s where it gets interesting in my mind. What is the next rising option? Is it Pinterest? Pinterest has a lot of targeting options that people can be able to use and leverage, but the Pinterest audience—while large in certain demographics—is small overall. Same with Snapchat. TikTok, I don’t know. I use Tick Tock every day, but maybe I’m the only one in my age group. 

So I think that’s the question: What’s that next option? And this is a question we’ve been asking since I got into marketing: what’s the option after Google and Facebook? And we’ve never really had an answer. That’s why at MileIQ, we ended up running TV ads to begin with. We were like: “What do we do with all this money that we got from Microsoft?” Like, thanks for the acquisition; now we have millions and millions of dollars. We can’t spend it only on Facebook. So let’s throw some ads on TV. 

Shamanth: Yeah.

Colette: I don’t know. Maybe TV is that channel?

Shamanth: Yeah. What I’m hearing you say is there will be budgets migrating outside of Facebook; there will be more fragmentation. That certainly makes sense. And are there any interesting or surprising channels that you think might open up that may not be somewhat on the current radar?

Colette: Yeah, so what’s interesting is if we’re doing a mobile web flow, then UAC doesn’t matter. So what I’m really excited about is can we use YouTube again? I would love to be able to use YouTube targeting again. 

The best set of ads I ever ran—ever in my 8 years of marketing—were YouTube ads directed at people who listened to rap music, who were watching videos on how to fix their dishwasher. You can’t make that up. I don’t know why it performed so well; it crushed it. Best ads I’ve ever run. So I would love to see that, again. I’d love to be able to use regular display advertising again, and see how it works on the web; see if it does anything. 

I think there’s a lot of web-focused channels that don’t necessarily have mobile app install products, or they don’t have those really built out well, so that could open up. Similarly, if costs are increasing, we might even see LinkedIn for business apps. In fact, it would be interesting too. Although I will say my LinkedIn feed is 99% junk right now, so I don’t know how well that’s performing. But there’s a lot of channels that we ditched, because they don’t have the mobile app install unit or it’s not very good. It becomes more interesting when you’re thinking about mobile web.

Shamanth: Definitely. And speaking of YouTube, search becomes an option and certainly, native ads can be huge. Native ads do have app install products, but certainly having a web thing can make them much, much more powerful, much more impactful. 

How does all of this change how marketing teams are structured? How do you see that playing out?

Colette: Yeah, if this is going to really become a thing, you’re going to need to have at least a part time engineering person around to make sure everything’s working. 

And you’re definitely going to need to have a data person.

Marketers are great at a lot of things, and we’ve become really good at reporting analytics, but that is very different than becoming a data scientist or data engineer. And the truth is that if we’re going to have to pair data together from three, four or five different sources, that’s where you really need a data person on your team. You’d not want to be putting that together and deal with lookups and pivot tables. We’re all on Macs, and I think they would all break. 

So we’re definitely gonna want somebody with some coding experience, some SQL, some database, and they don’t necessarily have to come from mobile. And it can be anybody. The good news is that data science is a very hot field right now. So there’s a lot of people coming into the industry out of college who could be really good fits for this. I don’t think you need to have the most advanced data scientist in the world, but you are going to need somebody who has that skill set. 

And also if you have a full analytics team or some kind of BI or engineering, that person can communicate between the engineering teams—especially in case something is wrong—and with marketing. Since you’re doing it in-house, you don’t have to necessarily wait for an update from AppsFlyer. 

The hard thing is now you’re going to be facing engineering, and so you’re going to be waiting in the queue that everybody else is in. And you’re really gonna need to be able to make your case. And so I think having somebody who’s a little bit more technical on the team is going to make a really big difference. Not only with the data analysis, but also just the politics of the world of engineering, and making sure that those very limited resources are allocated to marketing when that team needs help.

Shamanth: Certainly, and I think there’s been an ongoing shift towards having more engineering resources in marketing over the last couple of years. And what I’m hearing you say is this will only accelerate that process of marketing being not just media buying, so to speak.

Colette: Yeah, absolutely. And when you’re controlling millions and millions of dollars, you have to be able to figure out whether it is doing its job.

Shamanth: Yeah, certainly. Colette, I think this has been incredible with a new way of looking at everything that we’re confronting, and we’re going through. And, as has been the case every time I’ve spoken to you, there’s just so much I’ve learned; just a new way of looking at the circumstances that we are confronting. 

This is perhaps a good place for us to wrap up, but before we do that can you tell folks how they can find out more about you and everything you do?

Colette: Yeah, absolutely. So I’m Colette Nataf, I’m CEO and founder of a company called LightningAI. You can find us on lightningai.com, or you can find me on LinkedIn. Send me a message and just say: “I’ve heard about you on a podcast.” I’ve actually had people do that before, so that was fun. I do reply! If you want to talk about the future of the world, or anything else about marketing and ads, then I’m here. I love meeting new people and getting connected. So find me and let’s chat.

Shamanth: Absolutely. We’ll link to all of that in the show notes. And something that, Colette, maybe you’re perhaps too humble to mention is your Medium; certainly something I’ve enjoyed reading. Certainly something I found very inspiring.

Colette: Well, thank you very much, I appreciate that. I had somebody write me a comment on one of my articles: “Trump doctrine” which was the craziest comment I’ve ever gotten on the internet. I don’t know what it was, how I appeared the wrong way, but I’m such a progressive Bernie supporter. I still have my first sign in my window. That was crazy.

Shamanth: Excellent, Colette. So this has been great. Thank you so much for being on The Mobile User Acquisition Show.

Colette: Yeah, thanks so much. Been fun, as always.

A REQUEST BEFORE YOU GO

I have a very important favor to ask, which as those of you who know me know I don’t do often. If you get any pleasure or inspiration from this episode, could you PLEASE leave a review on your favorite podcasting platform – be it iTunes, Overcast, Spotify or wherever you get your podcast fix. This podcast is very much a labor of love – and each episode takes many many hours to put together. When you write a review, it will not only be a great deal of encouragement to us, but it will also support getting the word out about the Mobile User Acquisition Show.

Constructive criticism and suggestions for improvement are welcome, whether on podcasting platforms – or by email to shamanth at rocketshiphq.com. We read all reviews & I want to make this podcast better.

Thank you – and I look forward to seeing you with the next episode!

WANT TO SCALE PROFITABLY IN A POST IDENTIFIER WORLD?

Get our free newsletter. The Mobile User Acquisition Show is a show by practitioners, for practitioners, featuring insights from the bleeding-edge of growth. Our guests are some of the smartest folks we know that are on the hardest problems in growth.