The Privacy Imperative, What This Means For Content Creation
May 17, 2022 | Tune in here!
iPhone users - do you let apps track your data? I have a colleague who said that she does because otherwise she would feel like a hypocrite. I completely understand. But I take a slightly different approach. I use it the way I think it should be used.
If a brand I like is asking, then yes you can track my data. I want you to curate suggestions and let me know when something I would like is available. However, if you are an app I needed for one time only, then nah. Try again another way.
Google recently hosted its Privacy Imperative on April 13th, which was a webinar that discussed Google’s take on driving growth in a privacy-first world. They had 3 main discussion points - first-party data collection, measurement and automation.
Google has been very hungry lately for first-party data. This is data owned by advertisers because users have provided this data to them. It includes ad engagement, site behavior, conversion data and all other data provided to companies by individual users. There’s now a distinction between zero-party and first-party data, where the former is data knowingly offered to the brand. Here, think of forms you need to fill out in joining rewards programs. You don’t always want to offer the info but you know you are because it's a condition of joining.
Measurement and automation are both things that Google has long been pushing to brands and agencies.
Let’s tackle automation first. I always try to analyze it from the perspective of why would Google go through the hassle of developing the technology to automate bids. And, it really goes back to Google being Google, wanting to make more money.
I imagine there are thousands of agencies out there who inflate campaign budgets but keep bids at a level where you would not consume the budget. This allows you to never cap out of budget. Or, there are a lot of IOs out there that have not been fully consumed. Either way, Google is seeing that brands have a lot of budget left that they are not spending. Of course, also on top of Google’s egotistical mindset of ‘brands need us’.
But, let’s say there’s merit to this and that there are millions of campaigns out there where bids have not been changed in a few days or so. Also, that this brand actually has the budget to spend and can actually benefit from more ads on Google.
Then yeah, it actually does make sense for Google to activate on this untapped potential.
Brands actually embraced this technology for a while. I should probably say agencies because there is substantial reduction in overhead when done correctly. It was hard for me to get out of the habit of checking bids daily but I’d say it went from maybe an hour to about 15 mins. Afterall, blindly letting AI make decisions for you is probably not the best idea.
It wasn’t long before everyone realized that automation only works if it’s automating what truly matters. That is, data that drives business.
So, if the technology is automating more leads but the true goal is revenue and a decent chunk of leads are duds then the technology is not working. Hence, first-party data. It is what will allow Google to have insight into businesses it would otherwise not have and make its technology more sticky with advertisers. Instead of leads that may or may not result in revenue, closing the loop will drive more revenue.
However, in the age of privacy, what do we do when there’s gaps in data?
That’s what ‘Measurement’ is about. Google is now migrating to GA4, which is the topic of my last podcast, so that it can heighten the accuracy of Enhanced Conversions and Conversion Modeling. Both are similar so can get pretty confusing.
Enhanced Conversions essentially verifies users who convert against Google’s database of its own logged in users. The benefit of this is really on Google’s side. They’re able to create more signals so that they can use in bidding wherever else they need to, you included. Theoretically, it helps with remarketing because you know these users and Google knows these users so it will prioritize them for you. I am pretty sure, though, that they will offer these signals to your competitors. Not to alarm anyone.
Conversion Modeling is about connecting dots where they should have been easily connected. If someone makes a search, clicks an ad and then purchases a product, then the brand registers 1 sale. However, if this person clicks an ad on mobile and then switches browsers for whatever reason but does not click on another ad, then this is where the modeling comes in. Google will still credit the conversion to the ad first clicked.
Modeling is essentially Google’s way of taking credit for conversions that are not easily seen, which is becoming more and more these days because of the emphasis on user privacy. For impressionable advertisers, this would result in higher conversions from Google campaigns that may or may not be accurate. So a good way to verify is by closing the loop with offline conversion data, the type of data that would ultimately lead to an invoice.
HISTORY
Now for a little more history.
Even as someone who has been immersed in digital marketing for quite a while now, I can’t quite off the top of my head remember what catapulted the call for privacy over the internet.
The earliest major riveting change I can remember is Safari ITP 2.0 and all its subsequent updates. For a while, Apple was becoming more stringent on third party cookies but it was ITP 2.1 where there started to be changes to first party cookies. This change limited first party cookie tracking to 7 days from the 2 years it previously had. As you can imagine, this caused a major chill down the spine of the industry. This took effect in March 2019. A month later, ITP 2.2 took effect in April where the 7 day window would be reduced further to 24 hours.
It is so like Apple to pull that move. But hey, I am benefitting from the stock price especially in this market so I won’t complain too much.
GDPR or General Data Protection Regulation is a framework that the European Union enacted to govern the way that PII or Personable Identifiable Information is stored by a website, which basically requires it to be anonymized. Enter UID but more to come on that. This took effect in 2018 so a year before Apple’s ITP but didn’t have as big of an impact on the industry I don’t think.
What really catapulted privacy as a core issue in digital marketing was Mark Zuckerberg having to testify in Congress about practices at Facebook that allowed other parties to gain access to user data and use it to influence major elections.
Now why I don’t immediately associate privacy with this event is that Facebook, now Meta, is nowhere near the issue of privacy and has lost $230 billion to the heightened implementation of privacy guardrails.
They were actually on the other side of things, so to speak, taking out large newspaper ads against Apple’s updates in iOS14.5 to put major emphasis on user privacy. This is the version when Apple allowed users to decide whether or not to let apps track their data.
Oh and there was the whole boycott Facebook movement. #StopHateForProfit.
That was about hate speech and misinformation and not exactly about user privacy but they are very much intertwined. User privacy is about preventing the exploitation of any person’s online data.
It’s pretty much a foregone conclusion that this movement did not have any financial impact on the company. Stock price at the beginning of July 2020 was $233 and it closed at the end of the month at $253.
But it really did raise awareness for more governance across the internet.
REASON
So now let’s get into what this has become.
If anyone remembers Zuckerberg’s testimony in Congress, you’ll remember that Congress was pretty clueless about how Facebook works as a business. And so, this call for more privacy is really led by industry leaders in the private sector.
With the deprecation of the third-party cookie, we’re now seeing a divergence in the industry. Walled gardens of Google, Meta, Microsoft and others are getting higher because they each now have their own version of how to track user activity on their platform.
It isn’t a smooth transition for anyone. Google first announced FLoC or Federated Learning of Cohorts which quickly died because it didn’t actually create more privacy for people. It grouped users together based on recent browser history but that opened the potential for discrimination because Google would make assumptions about you based on the sites you visited. Or rather, allow you to be excluded because of sites that would make it infer that you are part of a protected group.
Google has since announced that Topics would replace the third-party cookie in its ecosystem. I may be oversimplifying this but I see it as the reverse of FLoC. Instead of browser history putting you in a cohort with others who have similar browsing histories, Chrome is basically trying to understand which sites you’ve recently visited and put you in a topic which advertisers can use for targeting. The difference is that there are already topics where users can be placed. So naturally, the groupings are larger and broader because you don’t create clusters as people are browsing.
I just did a Google search for ‘Microsoft Parakeet’ and nothing recent came up. You know what. I’m gonna do a Bing search. Yeah, the same article from April 2021 came up.
But what we need to know about Microsoft is that they are cooking something up on their own.
Meta, Snapchat, LinkedIn and virtually all the other social media platforms are basically encouraging Conversions API or CAPI, which is the same as Google asking for closed-loop first-party data uploads.
On the other side, across programmatic buying, you have different players coming up with different things.
The Trade Desk has UID 2.0 or Unified ID 2.0 which is its response to third-party cookies going away. And this is likely going to be the most prevalent solution moving forward.
This is because they are the largest, generating about $1.2B revenue annually. Actually, Criteo may be larger by total company revenue but they have a lot going on in marketplaces. Mediamath is another one but revenue estimates I saw was around $600M.
Regardless, The Trade Desk is certainly the frontrunner because it is actively investing in a solution.
The way UID 2.0 works is that instead of cookies leaving crumbs even when users are not logged in, part of being logged into a website allows UID 2.0 to leave its version of cookies with the user. They’re calling these tokens.
These tokens are refreshed all the time and it passes back salted hashed data. Hashed is where your PII is encrypted or translated into some gibberish and salted is when there is additional gibberish added so it will be even harder to decrypt something that cannot be decrypted.
LiveRamp is also a contender for an identity solution. Their product is called RampID, which was rebranded from IdentityLink and their edge is being the leading CRM provider. Since they have emails and other PII from a ton of different brands, they are using this to match users. I’m trying to understand where consent comes in but I guess for you to have been included in a CRM list, then you have consented to something at some point in time.
I’ve actually seen RampID work for a client and I remember telling my colleague that it is actually a much creepier way of matching people. Match rates were so much better that user data passed back was so much higher than cookies.
Now of course there’s the response to social clamor for a more respectful internet that tech companies need to heed but all these new solutions that are replacing third-party cookies are really just ways for companies to gain more control over user data.
With more data providers using UID 2.0, RampID or whatever becomes the leader, the more influence these solutions have over the industry. This is really what tech companies are after.
STRATEGY
Ok, strategy. What are digital marketers to do?
I am optimistic enough to think that the behaviors we are tracking now will still be the same behaviors we will track in the future. Therefore, these will still be the ones available for targeting. After all, what we are replacing is the technology, not the usage itself.
But do we just wait for things to catch up? Well, no.
Do we just test UID? Well, also no.
What we need to be thinking of is ‘how do we use new technology to our advantage’. If Google, Meta and others are going to be modeling conversions, how are we going to use this such that it benefits us or the bottom line of our clients?
Think of matryoshka dolls.
The outermost doll is a seed list which is the total possible users who are qualified to become a customer. This is where you should spend the most time.
Anticipating demographics, online behavior, user action and offline behavior for targeting is where we should spend the most time because we need to decide which of these correlate best with the brands we support. As an example, if your brand is one where phone calls are only for support, then you will not want to create an audience of people who are calling in.
More commonly, if you are supporting a brand where the purchase cycle is long, you will want to understand which types of users will complete the entire journey.
So we need to think a lot about who we are letting into our garden or ecosystem because letting the wrong people in would deteriorate our program.
The middle are those who we’ve convinced to engage with us even at a superficial level. This is where we start to collect data that we can own and use again for targeting or first-party data.
What we need to think of here is what is the next step that users would take to move along their journey. It begins with actions they’ve taken on the site but it still needs to be complemented with other data sources. If someone interacts with some tool on the site and then searches for something more specific, visits a location or performs another action, which of these is the right next step to target further?
The innermost doll are those that we have won. They are part of this circle because acquiring a customer is not a one time event. It is a favored chance to continue to do business with this user. Because we know who they are, it would be easy for us to target them again on any platform in any channel.
Might not have been the best visual but that really is how targeting kinda works. What about conversion data?
Due to data loss from iOS, there’s this whole education going on for advertisers to cut their expected ROAS. If you used to get a $10, now you might just get a $4.5. This $4.5, though, represents the entire $10.
In a way, I kinda like this. It’s like ROAS decided to do a stock split. Intrinsic value is still there but now a different number is representing it. But instead of focusing on hitting a specific number, you should now be looking at trend of performance.
What I’m trying to figure out now is whether or not this is short-lived. With enhanced conversions and CAPI, are we able to get back the complete picture?
I don’t know that we’re ready as an industry to answer this but I am a very strong advocate for looking at trends.
If you are a company or marketing for a company that is so focused on hitting a number, then you are probably not in the best shape. Conversely, if you pay attention to how things are moving, you can strategize about what to do next.
For example, if revenue is declining even as impression share is holding steady, then you might need to find other keywords that represent other products to bid on.
There’s a lot of benefit to connecting your data to your engines and platforms. You’re able to target accurately and have a fuller picture of results.
So why not do it?
There is real fear among advertisers to relinquish their data to platforms. I originally dismissed this fear because it is unlikely that these tech companies will venture into other industries. They will likely stick within the realm of advertising, except for Amazon. That’s one company that will actually figure out how to sell your product.
Putting that aside though, I think a valid concern is whether or not these platforms will create an unfair advantage for a brand or select brands. Either knowingly or unknowingly as well. Probably the latter because the former smells like a lawsuit.
Let’s play this out. Imagine 2 brands that are selling similar products. Brand A is a healthy advertiser on any platform and their philosophy is to spend more on advertising to increase their sales. Their goal is to maximize revenue.
Brand B is different in that their goal is to maximize profit. Their spend is lower than Brand A because they are cognizant of margins.
How will automation treat these 2 brands differently?
I’m grouping users into 3 categories. First are those who buy few but frequently and they are the perfect customer for Brand A because they spend a lot. Second is a group who buys a lot but infrequently. Obviously, this goes to Brand B. But what about new customers who don’t have a track record. Where will automation send these customers?
Google, Meta, etc. will all say that there are other signals that they can take advantage of to decide how to treat these customers. But as privacy falls more in the hands of the consumer, so does the size of the unknown increase. And maybe, the signals that they have are not enough.
So, if this were the case. Will these tech giants favor higher spending advertisers? After all, what they’re doing is just prioritizing more valuable customers. This is something we tell our clients everyday. The answer is… maybe. The point is… they can.
Then the question becomes how do you hedge against these tech companies putting you at a disadvantage because you don’t need to have as high of a budget as they would like?
This is where content comes in. The more you have of it out there, the more people will engage with it and the more signals you create so that these platforms will recognize how important you are in your space.
Content can really be anything. Organic posts, listings, paid ads, blogs and really anything else. What’s important is that there is enough of your content and therefore activity to be recognized digitally.
ALTERNATIVES
Alright, alternatives!
This one is hard because everyone is doing the same thing. Google is not alone in enhancing conversions, measuring conversions or modeling conversions. Literally every other provider in the industry has some sort of AI/machine learning mechanism in whatever they offer.
The alternative is also not avoiding any of this. If you resist change, you will be consumed by it.
I think what we need to be afraid of is government intervention. There could very well be a central bank of PII data where agencies need to go in order to target and track users.
While this sounds far fetched, it’s not impossible. Censorship is happening in some countries so regulation is certainly a possibility. But even if we don’t go too nuclear, cookies were created by Netscape in 1994 and it became the standard tracking mechanism. So it is equally probable that we all rally behind one solution because it is the most accessible and reliable. Kinda like barcodes. I don’t want to get into how that’s becoming obsolete but the theory behind it could exist.
I’m not at all advocating for this. I just think it is a concept to ponder on because it will happen in some ways. Many providers will use The Trade Desk’s UID 2.0 because it is reliable and compliant. So they will have a following and maybe LiveRamp too and whoever else. Actually, what we might have is a credit card situation where you choose between Visa, Mastercard, Amex, Discover and JCB.
It’s about a little over a year away from when Chrome phases out cookies. Between now and then we will see contenders become frontrunners and ultimately the standard. What’s important is that we continuously think of ways to be in control even in the face of machine learning and automation.