Be ready for identifier-free mobile advertising this year: InMobi’s Sara Camden
How do demand-side platforms change in the privacy era? And, when will you need to be ready for identifier-free mobile advertising?
Traditionally, demand-side platforms matched up ad placements across multiple publishers with specific audience segments, with marketers supplying targeting criteria and DSPs matching those requirements to groups of people via device-based ad identifiers.
Take away the IDFA and (soon) the GAID, and that’s a lot harder. So is campaign optimization, targeting, attribution, and more.
So how does it work now?
DSPs and contextual data for identifier-free mobile advertising
Context now matters more than ever. But not just the kind of context that comes to mind immediately.
“Those individual identifiers that we all have become reliant on are a thing of the past now,” says InMobi head of product marketing Sara Camden. “So it’s really looking at privacy-safe types of signals that have to do with the user’s device, some temporal things, like the time of day or the day of week, the type of supply that you’re targeting, be it the ad exchange, a specific publisher, or the type of publisher from a contextual standpoint … all of those signals have to come together and we have to do a lot more with a lot less.”
Less device data.
Less data about people.
Greater reliance on contextual data, including rough location.
Predictable postbacks in SKAN 4
One of the biggest challenges, Camden says, is not so much the loss of targeting efficiency — although that stings — but the lack of immediate feedback or at least predictable windows for postbacks. That’s one of the things she likes about SKAN 4.
“I think anyone who experienced SKAN 3 has probably cursed the pesky conversion window that kept resetting every 24 hours when a new event occurred,” she told me on a recent Growth Masterminds podcast. “So now we have these 3 fixed windows of the first 48 hours post-install, then days 3 to 7, and then days 8 to 35. And each of those comes with a postback, versus just a single postback before.”
(Of course, you can lock conversions and get postbacks earlier, as you can see in our SKAN 4 transition guide … but that comes at a significant cost in terms of estimated cohort accuracy.)
One of the impacts of SKAN 4 is a resumption of interest in advertisers, says Camden: many who had been completely anti-SKAdNetwork and perhaps even diverted ad dollars away from iOS, as a result, have been attracted back.
While SKAN 4 is still pretty far away from prime time in terms of widespread adoption and usability, InMobi is seeing 20% of bid requests that are SKAN 4 compatible as of April. SKAN 4 postbacks are still few and far between, but clearly there’s a significant uptick in ecosystem adoption that will make SKAN 4 the default attribution protocol for most at some point this year.
Crowd anonymity is better than privacy thresholds
Crowd anonymity seems much more advertiser-friendly than SKAN 3’s privacy thresholds, too.
“Early results that we’re seeing is that now you get some level of conversion insights at around five installs per day per campaign, whereas it was about 15 in the earlier version,” says Camden. “So, you know, that certainly is a marked improvement over the earlier iterations.”
That’s good news for marketers who kept seeing null values in SKAN 3 postbacks.
Of course, more installs per campaign will, however, get you better results: fine conversion values and potentially full source identifiers. So it’s still important to put more wood behind your campaign arrows.
Ready or not, GAID is going away
Camden thinks Google’s record of kicking the can down the calendar on the third-party cookie deprecation is not relevant to the GAID, which she thinks will be on track to be gone by the end of the year.
Part of that is due, perhaps, to the more consolidated mobile space, where Google holds more power over the Android ecosystem than it does over the web environment.
“By the end of this year, I think marketers need to be ready to predominantly be targeting their campaigns and doing attribution without relying on IDs,” she says. “You mentioned how the can keeps getting kicked down the road when it comes to deprecating third-party cookies. I’ve not been seeing that same friction when it comes to the mobile identifiers. I think because they’ve gotten that practice with iOS and App Tracking Transparency that they’re more prepared … so I’m not seeing the type of delays that we’ve seen with cookie deprecation.”
That might be a scary thought to some — and identifier-free mobile advertising — and it’s a good reason to get up to speed with Privacy Sandbox on Android.
Much more in the full podcast …
Watch the full video above, and subscribe to the Growth Masterminds podcast on your favorite podcasting platform. There’s much more there, including Camdens’ take on mashing up Privacy Sandbox on Android and Apple’s SKAdNetwork to make the ultimate solution.
Here’s a full transcript of our conversation …
Note, it’s AI-generated and lightly edited, so it may not be 100% correct.
DSPs in the privacy era and identifier-free mobile advertising: what will change?
How do demand side platforms change in the privacy era?
Hello and welcome to Grow Masterminds. My name is John Koetsier.
Back in the dim mists of time, like three years ago, DSPs were pretty straightforward businesses, right? By which I mean, they were horrifically complicated, incredibly challenging, but at a conceptual level, it was relatively simple. We had a massive demand for ad placements. We matched it up with devices whose IDFAs or GAIDs that indicate compatibility or interest or opportunity.
Since SKAdNetwork, however, and soon Privacy Sandbox, it’s gotten more difficult. There’s less data, there’s less history, there are more challenges finding great supply for a DSP’s demand. So how do DSPs evolve in the era of privacy? To chat about this, we’re joined with a longtime adtech and martech expert. She’s been with Equifax, with Epsilon, she’s now head of product marketing at InMobi: Sarah Camden.
Thank you. It’s great to be here, John.
It’s great to have you. Like I said in our pre-chat, I feel like I’ve seen your name everywhere. I’ve heard your name in lots of places. This is the first time face-to-face. Thanks for taking the time.
Yeah, I appreciate the opportunity. It’s a fun subject to talk about, right?
Absolutely. Everything in ad tech is fun. Super fun. Barrel of laughs. Let’s start here: level set. What the heck does a DSP do?
So DSPs are really there to make lives easier for marketers and ad buyers, you know, and really kind of automate that process. You know, I actually got my start in adtech right around the birth of DSPs when contextual type of targeting was still sort of the primary mechanism. But, you know, marketers realized with all of these devices that people had, they would love to get more granular and individually target people.
So hence the birth of DSPs that can take that data and really personalize as well as measure campaigns at that individual level, which we all have loved. Obviously, things have changed quite a bit, and it’s a little bit of: everything old is new again in the current state of DSPs today.
Back to the future or back to the past, one of those two. Let’s talk about a DSP under SKAdNetwork.
How’s that work? What are you doing? What signals are you using? How are you finding the right placements?
That’s a great question. You know, those individual identifiers that we all have become reliant on are a thing of the past now. So it’s really looking at privacy-safe type of signals that have to do with the user’s device, some temporal things, like the time of day or the day of week, the type of supply that you’re targeting, be it the ad exchange, a specific publisher, or the type of publisher from a contextual standpoint.
So is it sports? Is it news? Is it gaming? All of those signals kind of have to come together and we have to do a lot more with a lot less.
Interesting. Are there some geo signals as well, roughly?
Yeah, yeah, there are some rough GPS signals, like DMA, or just regional type things, you know, those are things that we tend to use at least at InMobi.
So that’s a massive change. What kind of level of impairment, let’s just talk SKAN 3 right now, maybe because that’s what’s dominant right there. What kind of level of impairment have you seen since losing the IDFA?
Can you estimate a percentage of efficiency loss in targeting?
I don’t know that I have an efficiency loss statistic off the top of my head, but you know, one of the big things is just that loss of real time feedback loop. SKAN has a couple of different timing mechanisms that delay when marketers get that insight into their campaign performance, both from the conversion window standpoint, but then also this randomized delay that Apple applies on top of that just to make sure that you can’t triangulate who an individual user is, which kind of goes against the whole spirit of why SKAN exists today.
So that certainly has been a gap that most folks are addressing through various predictive models to try to pinpoint when a specific conversion event occurred and what that event was in some cases, because you have to overcome these privacy restrictions before you even get insight into what exactly that user did from a conversion event standpoint.
Mm-hmm. So now we’re moving into SKAN 4. There’s more data available. You’ve got fine conversion values, not just course conversion values. You’ve got three postbacks, so you can have a little bit of longevity over time. There’s a bunch of other things going on.
How’s that transition going? Where are we in that transition?
Yeah, yeah, that transition is a welcomed one from our perspective. You know, as you mentioned, we get more data more often. It’s also much more predictable.
I think anyone who’s experienced SKAN has probably cursed the pesky delay or the conversion window that kept resetting every 24 hours when a new event occurred. So now we have these 3 fixed windows of the first 48 hours post install. then days 3 to 7, and then days 8 to 35. And each of those comes with a postback versus just a single postback before.
So, you know, you do get that view into more of the consumer’s lifetime value post install, which is incredibly valuable.
And as far as an uptake perspective, I’ve been noticing a lot of attention coming from advertisers that had been completely averse to SKAN in the past. Either they were using mechanisms like probabilistic attribution, or maybe even diverting their budgets away from iOS entirely. They seem to have turned a corner and are now actually wanting to embrace SKAN, both the earlier versions, as well as SKAN 4.
The publishers are coming alongside as well. SKAN 4 officially rolled out into the wild this past October … we were seeing sub 5% of bid requests that were iOS as SKAN 4 compatible. But now coming into April, we’re up to 20%. So, you know, there is supply out there to bid on. And really the biggest impediment at this point is just getting the advertisers to come along and update their conversion value models to align with the scan for standards.
Wow, that is news to me. I had not heard that it’s at 20% right now. That’s obviously nowhere near where it needs to be for full adoption or anything like that, but that’s not insignificant. That’s not insignificant at all. That means that there’s widespread testing. There’s the beginning of widespread adoption.
That is really interesting, and it’ll be super interesting to see how that pace accelerates or continues to grow over time. One thing that’s interesting …
Yeah, one thing that’s interesting to me is, obviously in SKAN 3, if you didn’t have a sufficient number of installs, you hit privacy constraints and you just didn’t get data, right?
Do you feel in SKAN 4, that bar is lower, at least for getting course postbacks for your first postback?
Yeah. Yeah, yeah, absolutely. Early results that we’re seeing is that now you get some level of conversion insights at around five installs per day per campaign, whereas it was about 15 in the earlier version.
So, you know, that certainly is a marked improvement over the earlier iterations.
That is really interesting. And of course marketers want everything. They want the fine conversion value. They want the source identifier. They want all the data you can possibly get when you let’s say beat or exceed crowd anonymity numbers in SKAN 4, but getting at least some data even at very low levels is a huge improvement because getting those blanks in SKAN 3 I think contributed to what you talked about … those marketers that said, I’m opting out. I’m out of here. I’m not dealing with this SKAN business at all because they got null values or else they got values that were insane.
$100 for CPI or $500 for CPI, because so may of their campaigns didn’t meet the privacy requirements. That’s really good news.
Yeah, yeah, I would completely agree. And it really opens up iOS again to some of the smaller budget advertisers because those privacy limitations in earlier versions would just preclude them from being able to spend enough to get over the privacy threshold and actually get some insights into their conversions.
Of course, we’re in multiple transitions right now, right? We’re going SKAN 3 to SKAN 4. And we’ve talked a little bit about how that’s going. We know Privacy Sandbox is happening. And we see every month or two, Google releases a little piece of information, a little thing here, a little thing there.
I think the promise was GAID would be end of lifed end of this year.
Who knows? We’ve seen that the third party cookie has received new leases on life quite frequently from Google. What’s your sense of privacy sandbox and when marketers need to really start paying attention to that?
Yeah, yeah, it’s interesting. You mentioned how the can keeps getting kicked down the road when it comes to deprecating third-party cookies. I’ve not been seeing that same friction when it comes to the mobile identifiers. I think because they’ve gotten that practice with iOS and App Tracking Transparency that they’re more prepared for that.
So I’m not seeing the type of delays that we’ve seen with cookie deprecation. So honestly, by the end of this year, I think marketers need to be ready to predominantly be targeting their campaigns and doing attribution without relying on IDs.
That is super interesting and kind of freaky because I had sort of back my mind, and I wonder maybe a good chunk of marketers, we tend to be optimistic people, right? You know, it’ll be more time, we’ll have more opportunity.
I kind of figured, you know, Google will kick the can down the road, you know, 3 months, 6 months, something like that, who knows, we’ll see where it goes. And I think many marketers have been thinking in similar ways and therefore not really getting ready. Also because guess what: They’re busy with the SKAN 4 transition, which is not easy in and of itself.
But you bring up a good point: that hasn’t moved since Google announced it maybe a year ago, something like that. And also there’s the iOS precedent, right?
So Apple gave some warning, but when it happened, IDFA was out. I mean, it’s possible to get, but it’s really, really hard to get. And you get a small fraction. Maybe with the extensive warning that Google gave … 5hey’re like, hey, that is it. It’s gone. It’s done. I’m not sure if the industry is ready for that yet.
Yeah, yeah, I would agree. I mean, marketers are pretty instant gratification minded. And you know, if there’s workarounds, you can’t blame them for taking advantage. But you know, the way that we try to advise our customers is, by all means, make hay while the sun is shining.
But do not ignore the looming clouds on the horizon because you could be caught flat-footed and you’re going to be in a lot of trouble.
So right now is the time to really be testing and learning and gathering a repository of data on how to make your campaigns perform without IDs because you’re going to be at a big competitive disadvantage when those bandaids get pulled off, whether it be Apple enforcing their policies that ban fingerprinting. or the eventual rollout of Android Privacy Sandbox.
Mm-hmm. As a DSP, when you look forward and see Privacy Sandbox for Android, how do you feel about it? Are you like, well, there’s a lot of data there. There’s more data than even at SKAN 4 level.
Are you cautiously optimistic? Or are you thinking, wow, no GAID, we’re screwed?
Yeah, I’m definitely optimistic. You know, part of why I came to InMobi, like I was very happy at my last job, but I loved sort of their forward looking vision and preparing for that world without identifiers, you know, going all in on that from a strategy standpoint.
So yeah, I’m actually very optimistic based on the results that we’ve been seeing with earlier versions of SKAN, where we’ve been able to get KPIs costs at parity with ID-based traffic and fingerprinting campaigns. So I’m very optimistic about it.
And Google Privacy Sandbox, it addresses some of the things that ScSKANan doesn’t today. It’s almost like if you could mash the two of those solutions up together, it’d be a pretty viable long-term solution because Google’s kind of solved for the front end audience targeting piece of the pie, and then Apple solved for the measurement piece of the pie. So, putting those two things together, I think, is a pretty reasonable solution.
Well, that’d be interesting. Easy solution. Put Apple and Google in a room, force them to figure it out.
Right? They love to work together, right?
Best friends, obviously.
Well, yes, it is interesting. There’s some speculation that, you know, obviously the industry has been pushing for stuff from SKAN 3 that we’re sort of seeing in SKAN 4, but there’s some speculation that privacy sandbox has impacted what Apple has put in SKAdNetwork version 4 and that evolution will continue over time.
Very likely as each massive mobile player sees what the other is doing. It says, okay, we can add this or we can add that. It’s a little different than Google … in Google’s perspective because Google is an ad network. That’s how they make 97.9% of their money. That’s not an official stat. Don’t quote me on that, but it’s somewhere around there.
It’s close enough, it’s close enough.
Exactly. Whereas Apple, it has an ad network, but I mean, it’s a rounding error on their profit. So interesting times ahead. And you mentioned already, marketers need to prepare, be ready for a future without ad IDs.
Talk about how DSPs will help marketers win in this era.
Yeah, yeah. I mean, DSPs as well as MMPs, I think are great consultative partners on this, particularly those that have been investing in this idealistic future for a while, because there’s a lot of them that haven’t.
But, you know, in the case of InMobi, we’ve started very aggressively investing in and staffing up to support SKAN and really building it into our roadmap. You know, back when App Tracking Transparency was first announced, we started running idealistic campaigns. So we’ve got sort of this repository of learnings that goes into all of our bidding models that are ID free.
So, finding those DSP media partners that have the experience that can kind of help them along that journey, as well as leaning on their MMPs for how they should adapt their reporting strategies and attribution strategies to support this idealist future, I think is the key to helping make that jump.
Excellent. Sara, thank you for taking this time.
My pleasure, it was a great chat.