The “Privacy Sandbox:” breaking down upcoming changes to analytics tracking

In 2025, digital advertising will change on a fundamental level.

What started a decade ago with Europe’s GDPR laws will go global as the Google-backed “Privacy Sandbox” initiative kills third-party cookies. The result will be an entirely new model for ads tracking and attribution.

If that sounded like gibberish to your ears, you aren’t alone. When discussing changes to marketing analytics, technical terms abound. Take the official description from the Privacy Sandbox website for example:

Privacy Sandbox for the Web will phase out third-party cookies by using the latest privacy techniques, like differential privacy, k-anonymity, and on-device processing.

Privacy Sandbox also helps to limit other forms of tracking, like fingerprinting, by restricting the amount of information sites can access so that your information stays private, safe, and secure.

Differential privacy? K-anonymity? What does it all mean?

More importantly, what does it mean for your business? How will your ability to advertise change once third-party cookies go away for good?

Let’s break it down.

Digital advertising 101: how ads tracking works today

Digital ads require user tracking for a few reasons. First and foremost, tracking users allows for serving customized ads.

Targeted and untargeted advertising

Consider a billboard along a highway. Besides geography—its physical location—it is utterly untargeted. Billboard advertising relies on sheer volume. Sure, 99.9% of the people viewing, for instance, a billboard ad for an equipment rental company, will not be relevant. But if the billboard receives thousands of views a day from a diverse audience of commuters, it’s probably reaching your audience anyway.

Online advertising usually can’t work that way. There are hundreds of major highways and millions of streets across North America. There are trillions of web pages on the internet, many of them containing several ads each. The sheer scale of internet traffic—and of advertising competition—means that an untargeted approach cannot work. So we use tracking tools—namely, third party cookies—to ensure ads are relevant to the user who views them.

The role of cookies in advertising

There’s that term again—”third-party cookies.” Let’s talk about cookies. A cookie is a little piece of identifying information that apps and websites attach to users. Cookies get saved to your browser or device, enabling websites and apps to remember things about you. When you return to a website you logged into yesterday and find yourself still logged in, that’s a cookie.

Cookies are made of several parts. Every cookie will have a name describing its purpose; a value, often representing a unique identifier used by the server to track you; a domain, the address of a website or app where the cookie applies; a path, describing the location on the website or app where the cookie applies; an expiration date, describing when the cookie should expire and be deleted from the device; and security settings, defining the way the cookie operates to prevent malicious attacks.

Cookies get added to what’s called a “request header.” Each time a user loads a web page in their browser, they send a request to a server asking for the page’s contents. The request header is kind of like the return address section on an envelope. It includes information about who’s sending the request as well as what the request entails. When a cookie gets added to the request header—represented by its name and unique value—the server knows to adjust the information it sends accordingly. If the cookie determines whether a user’s logged in to their email inbox, then the server knows to skip the login process and send you straight to your inbox.

So we know that a cookie is like a tag attached to a user, describing some attribute of the user, like their login status, for the benefit of a web server. Great. Where do third-party cookies come in, and why are they getting killed?

Third-party cookies and their impact on ads

Third-party cookies are cookies managed by platforms other than the website each user visits. They are currently common practice. Many websites, particularly those driven by ad revenue, may attach dozens of third party cookies to each user.

For example, LinkedIn, Meta (Facebook/Instagram), Google, and X (Twitter) all have their own tracking tools that rely on third party cookies. When these tracking tools get added to a web page, they report back to their associated ad platform, using cookies to identify user behaviours, assign audiences, and otherwise allow for advertising personalization.

Let’s say you visit a blog called bestdogfood.com, reading an article reviewing the top 10 best organic dog foods. The next day, you visit a major news site like CNN and see an ad banner promoting one of the dog food brands you read about yesterday. Using a third-party cookie, the ad platform determined your interest in dog food brands. When you pop up again on another website using the same ad platform, that cookie data tells the platform to serve a dog food ad.

This is the heart of digital advertising as of 2024 and has been since the mid-2000s.

Privacy problems: why third party cookies are going away

Third party cookies enable modern advertising infrastructure, but they create all kinds of concerns about privacy and data ownership.

To explain the issue, let’s reframe how we think about today’s internet.

The internet as a frontier

The internet as we know it now has existed for about twenty years. That’s when Facebook first launched, when “web 2.0” began. It’s around this time that Google became the dominant search engine. YouTube would launch just a few years later. The smart phone—and by extension, the mobile internet—would also emerge within the next few years.

Imagine the first twenty years of the automobile. Then imagine the hundred years that would follow. Because of its complete integration into our lives, it’s easy to forget that today’s internet is still a new frontier.

That doesn’t mean we have digital cowboys duelling in dust-strewn online streets. There are no marshals, no Wanted: Dead or Alive posters, and relatively few horses. The internet is like a frontier because the long arm of the law hasn’t caught up yet. Much like driving laws—it took nearly 80 years from the introduction of the Ford Model T for seat belt use to become mandatory in the United States—internet privacy legislation lags behind technological innovation.

A self-regulating internet?

In the early days of the internet search engine, SEO was cut-throat. The earliest versions of Google didn’t account for bad faith tactics, so bad faith tactics became common. Website owners would stuff keywords into white space on their website, writing white-on-white to make it invisible to users but readable to machines—like Google. So Google adjusted, deprioritizing simple signals like keyword volume, and introducing more thoughtful scoring factors like text readability.

Bad actors got more sophisticated. Instead of simple keyword stuffing, they would create link farms with other scammers, generating fake popularity and interlinking between websites that fooled search engines. So Google adjusted and killed that practice.

Next came tactics like negative SEO—undermining competitors by sabotaging their own SEO through purposefully spammy tactics like low quality link buying. Google adjusted and killed that practice.

In the same way as Google had to effectively regulate website best practices in order to keep internet search useful and effective, it’s joined with many other tech companies to devise new privacy standards, getting ahead of potential legislation that may arrive years too late.

Google as a monopoly: should they be in charge of setting web standards?

As of writing, Google is dealing with a US anti-trust case arguing that they monopolize search advertising. The arguments put forward by the Justice Department are compelling: Google acknowledges that they face no market pressure, that they can set prices at a whim, and that they can “make a product worse” without losing market share. This is pretty damning stuff.

Regardless of the outcome of the case, Google’s dominant position in search advertising makes their control of the Privacy Sandbox initiative interesting. On one hand, who knows the digital advertising space better than Google? As the world’s largest ad platform, they stand to lose the most with the death of third-party cookies. Perhaps they should be the ones to guide publishers toward the best replacement solution.

On the other hand, setting policy and standards for a whole industry affirms Google as monopolists. The Privacy Sandbox may involve many stakeholders, but it’s branded as a Google project, hosted on Google servers, and supported by documentation written and specified by Google through Google’s own developer resources website. One might find oneself asking: is this plan the best way forward for the industry, or for Google?

It’s not the first time Google’s been accused of agenda-setting in the tech world. Their AMP (Accelerated Mobile Pages) initiative. AMP is a strict development standard for mobile web pages that offers high speeds and consistent design. The problem is, Google enforces this standard by rewarding pages that follow it—giving them preferential treatment in search engines and through advertising. If Google can set a standard and compel the internet at large to comply with it, that threatens the openness of the web.

Regardless of whether Google should lead this initiative, they are. It will be interesting to follow this anti-trust story as it unfolds. It may have implications for the Privacy Sandbox and the digital advertising world at large.

The case against third-party cookies

The case put forward by privacy advocates revolves around consent. While users can issue explicit and informed consent to first-party cookies, cookies issued directly by the website they’re using, the way third-party cookies track users across many platforms makes giving that consent tricky.

For instance, I may be happy to have my local news website serve me relevant ads as a means of earning revenue. I use the product, and ads provide a path to monetization that I find acceptable.

If my user data from that local news site then gets used to customize ads on platforms where I do not consent to have my data used, that could be considered a breach of privacy. If the platform itself sells my data to another party, that’s even worse. Without being able to control how third parties handle my data, I cannot trust that my personal information is secure.

Speaking of security, there are also risks of data breaches and improper access. While ad platforms themselves are secure, access to those platforms from potentially millions of advertisers is less secure. If a hacker is able to take over an advertiser’s administrator account, they may be able to download sensitive user data for their own purposes.

This raises a final point made by privacy advocates: the extent to which cookie data is truly anonymous. It’s true that marketing cookies obscure identifying user data, but they may not do a good enough job. There are cases where individual users can be identified by “anonymous” marketing data, which is a major breach of privacy.

The case FOR third-party cookies

If third-party cookies are so bad, why have they persisted for so long?

The answer is simple: they’re effective. Third-party cookies allow for accurate tracking and responsiveness in advertising. They enable website analytics. They help businesses make informed decisions. They are easy to implement.

Keep in mind that the death of third-party cookies is NOT the death of advertising, or of user tracking. If third-party cookies died tomorrow with no replacement, ads would continue to serve. They’d be far less relevant in many cases, but they would persist in roughly equal volume to what you see today. Some advertising advocates would suggest that a targeted ad containing something you may be interested in is less irritating than an untargeted ad that may feel completely irrelevant to you.

What would happen in the event that advertising and tracking became unviable is the death of free internet services. Many users lament the increasingly paywalled state of the internet; this will only become worse. Businesses need to make money, and for many publishers and service providers, advertising is a viable means to do so. A trackerless internet would kill this revenue model. Its likely replacement would be subscriptions and other pay-to-play models.

The digital ads industry is projected for a market size of nearly $800 billion by 2026. Tech companies like Google receive about 80% of their revenue from ads. In a very real sense, ads pay for digital innovation. Advertising revenues go toward R&D efforts that may result in, among other things, powerful new forms of AI.

Without suitable replacements, publishers, tech platforms, and advertisers alike have valid concerns about the economic impact of removing cookies.

The Privacy Sandbox: a vision for privacy-first advertising

The so-called Privacy Sandbox is an attempt at a suitable replacement for third-party cookies. Developed by an interdisciplinary team of developers, marketers, and policy experts from dozens of organizations—but very much led by Google—the Privacy Sandbox initiative will try to introduce safety and anonymity to ad tech, without sacrificing tracking.

Their website claims three goals for the project: to keep information private through new technology, to enable publishers and developers to maintain and ads-driven revenue model, and to build new standards for internet privacy.

The initiative includes separate “sandboxes” for web and for Android apps. This article will focus on the web component.

What is a “privacy sandbox?”

The term ‘sandbox’ is pretty well known among developers and tech enthusiasts but may be confusing if you’re not in that world. In real life, the purpose of a sandbox is to contain sand, usually for kids to play in. Without the box, sand gets everywhere.

This is the same principle for the privacy sandbox. In order to ensure user data is protected, they’re engineering systems that will keep sensitive user data contained within strict boundaries. The sandbox ensures tracking data gets where it needs to go without spilling over into any unintended spaces. It introduces safeguards as well as new and less invasive ways to track user interests and behaviours, which we’ll discuss later.

The privacy sandbox is a conceptual thing, comprised of a variety of technical and policy changes that reduce the amount of personally identifying information absorbed by trackers while still providing strong ad targeting capabilities.

k-anonymity

One of the confusing-sounding concepts raised by the Privacy Sandbox website is the concept of “k-anonymity.” This is a statistical measure that speaks to the overall goal of Privacy Sandbox. A k-anonymity score measures how specific marketing tracking can be. For example, a k-score of 1000 means that the smallest identifiable cohort that can be gleaned from tracking is a group of 1,000+ users. Initiatives within Privacy Sandbox may be evaluated on this k-score to determine how well they support user privacy.

Differential privacy

Differential privacy is somewhat controversial, at least among marketers who rely on accurate user data. Differential privacy purposefully introduces noise to collected data—watering down insights to prevent them from allowing individual identification. Where an unaltered report might allow a marketer to identify which specific recorded session resulted in a conversion, a noisy report will be made inaccurate enough to prevent such granular tracking.

Enhanced privacy technology and standards included in Privacy Sandbox

Let’s cover some of the more significant technologies and standards that will upgrade user privacy across the coming months.

IP Protection

IP addresses are unique identifiers assigned to each device or connection. An IP address tells a server where to send data so that you can interact with the internet. While IP addresses may change over time, they tend to be fairly static, meaning that an IP address can become an identifier used to track down an individual user.

The Privacy Sandbox initiative will introduce intermediaries that obscure user IP addresses. These proxies will prevent the destination origin from collecting IP addresses, while a second layer of security prevents the proxy itself from accessing information shared between the client and the origin.

Partitioned Network States, Storage, and Other Resources

Whenever you browse the internet, your browser accumulates resources. For example, your browser can remember websites it’s visited previously and re-use the same connection configuration to make loading the website faster on subsequent visits. It can also store assets like images and documents in a cache, or share information between website sections—or even different websites—in certain cases to enhance speed and usability.

This is nice for user experience but offers some privacy downsides. These cached network resources, files, and settings are “globally scoped,” meaning that other websites can access information from this cache in some cases. This can help identify user behaviour. For instance, a website might take notice of exceptionally fast loading speeds for one of your frequently-viewed websites. Based on the exceptional speed of the connection, it can infer that you’re a frequent visitor to the website. Using this data, it begins to build a profile of your likely interests and behaviours.

Partitioning cached data puts up firewalls around this information, ensuring other websites cannot access your cached network information. While this may have a modest negative impact on page load speeds, it prevents unscrupulous third parties from sniffing around your browsing behaviour.

User-Agent Reduction

Every time you connect to a website, your browser sends a request to the web server. The request contains something called a “User-Agent string.” This string—a programming term for a line of text—contains a variety of details about your device and connection. It tells the server things like which browser and operating system you use, including version numbers and whether you’re on desktop or mobile.

While sending requests is mandatory to access the internet, including detailed identifying info like specific device models is not. Many websites require User-Agent data for things like responsive layouts. Knowing you’re viewing a website on a phone tells the server which page layout to send to the device so it displays properly. Google’s new User-Agent policy shares limited information in the initial User-Agent string. If the website needs more specific information, it provides a method for the server to securely ask for follow-up data. By broadcasting less user device data initially, this policy reduces how visible and identifiable you are online.

Bounce tracking mitigation

In the face of the extinction of third party cookies, you might imagine that some analytics companies would be willing to fight dirty to maintain their marketing visibility. You’d be right. Bounce tracking is one way some companies get around the decline of third-party cookies.

The way it works is actually pretty simple. When users navigate to a website, they get routed through a third party before landing at their destination. A brief redirect, potentially so brief it would be hard to notice as a user, attaches cookies to your browser that help companies sidestep third-party cookie restrictions. The cookies follow you to the new website and track your behaviour.

New detection patterns built into Google Chrome sniff out these bounce tracking systems. If Chrome notices a redirect where the user didn’t engage with any content—no clicks, scrolls, or time on page—it scraps the cookies associated with that redirect so that no tracking follows through to the destination website.

Privacy budgets

Another means of avoiding undue user tracking is the privacy budget. The right combination of otherwise anonymous data can allow for granular user tracking, which is what Privacy Sandbox works to avoid. Kowing which city or state a user lives in usually isn’t enough to identify them, but what if you also know their top personal interests and most recently visited websites? This combination might be enough to narrow down a cohort of tens of thousands to hundreds, which is specific enough for Privacy Sandbox to care.

Introducing the privacy budget! Imagine each piece of analytics data is assigned a monetary value. Maybe each user’s top 3 interests cost $5 each. A user’s geolocation (city-level) might cost $10. A privacy budget will add up the total value of each piece of data collected by a website. If the collected data exceeds the threshold amount, the data will get blocked or rounded off to be less accurate, thus limiting the extent of personally identifying information that can be collected by one entity.

For example, if the privacy budget is $20, the website could collect a user’s top three interests and come in under budget. If they try to also collect their geolocation at the city level, they will exceed the budget. They may either receive no geolocation data, or they will receive a broader location—Utah instead of Salt Lake City, perhaps.

New methods to track users

The purpose of the new privacy standards and techniques outlined above is to kill invasive alternatives to cookie-based tracking. By restricting many of the ways websites can harvest user data, the Privacy Sandbox encourages advertisers to work within Google’s emerging next-gen tracking ecosystem. What does that system entail? Let’s get into it.

Topics API

After experimenting with other tracking methods, the Topics API emerged as one means of targeting ads. The Topics API is a feature built into the latest versions of Google Chrome. As users visit websites, they collect topics—tags covering around 200 different interests. Each user may be assigned up to five topics at once. Topics rotate out on a weekly basis, updating based on the user’s latest browser activity. Sensitive categories like race, political affiliation, and sexuality are not subject to tracking through this system.

Advertisers can use the topics API to find relevant audiences without compromising the privacy of individual users. The Topics API has some limits. Right now, Chrome is the only browser that uses it. This means Topics can currently only target Chrome users. It’s also an opt-out feature, meaning Chrome users who disable it in their settings will not be tracked.

FLEDGE (First Locally-Executed Decision over Groups Experiment)

Some elements of Privacy Sandbox are already active. Others, like FLEDGE, are still in development.

FLEDGE is a new method of serving ads where targeting happens locally in your browser. That means that instead of sending targeting data over a network connection to a server, the targeting data lives in your browser application.

When a website wants to serve ads, the ad platform talks to your browser. Essentially, it says “here are some possible ads I can show. Which one do you think the user will like best?”

Based on your browsing habits, the browser sends back a message like “the user really likes gardening, science fiction novels, and vegetarian recipes.” If there’s an available ad matching any of those top interests, then the ad platform will display a relevant ad.

This is a subtle but important change from the current way ad targeting works. Instead of sending a big bundle of user data over the internet, the browser simply lists which topics might interest the user. The ad platform, which must be run on a server configured not to collect any additional metadata, only receives a list of interests to determine which ad to show. It does not “know” the user or learn any identifying information about them.

Trust Tokens

Another challenge for publishers and advertisers is verifying the authenticity of their audience. By certain measures, a great deal of the internet’s users are bots. In most cases these bots are innocuous, used to send automated updates or retrieve information from websites and social media platforms. However, a percentage of these bots are malicious. Malicious bots can be used to impersonate humans for the purposes of fraud or create false interactions to trick security systems and ad platforms.

Bot-based ad fraud is particularly common with display advertising, which is highly dependent on the kind of ad tracking discussed in this article. In addition to targeting the right user groups, advertisers want to know those target users are real.

Trust tokens are Google’s solution to this challenge. A trust token is an arbitrary tag attached to a user. It contains no identifying information. Consider them a simpler form of cookie that does not share any identifying information.

For example, when you visit website A for the first time, it starts testing your session to see if you’re likely to be a human. It might test for human mouse movements, clicks, or other page interactions that would not be typical of a bot. If it sees human activity, it assigns you a tag with a random, unique code like “ABC123.” That’s the only information attached to your session.

When you leave website A and navigate to website B, the unique code comes with you. The ad platform on website B checks for the presence of a trust token and verifies the unique code is valid. If it is, the platform will serve ads. If it isn’t, it won’t, thus avoiding bot fraud without collecting personally identifying information. These tokens expire quickly, attaching a refreshed ID on a new session to prevent tracking across sessions. Effectively, the only information shared across websites is a simple “yes/no” value answering the implicit question: is this user a human?

Reporting aggregation

Remember that differential privacy we talked about earlier? Whether through their attribution or aggregated reporting APIs, Privacy Sandbox will result in major restrictions in conversion tracking and audience reporting. Many of today’s conversion tracking techniques will provide specific and identifying information about each conversion. Let’s consider a form submission conversion. Right now, an advertiser will often be able to collect the geolocation, time and date of conversion, device type, and in some cases even the gender and interests of each converting user.

Google’s attempt to solve this is, frankly, quite blunt. Just make the APIs return less specific information. They plan to reduce what is reported about conversions to the bare minimum necessary details, delay reporting to prevent marketers from tracking conversions as they come in, and even add noise to the data to make it harder to sift through.

If you’re noticing some negativity in the tone of this section, you aren’t wrong. As an advertiser, these changes offer no direct upside. Advertisers will know less about their converting audience. It will take longer to learn about their converters, and their data will be corrupted with noise intended to make it hard to track down users. It is the opinion of this humble marketer that this sledgehammer approach to safe conversion tracking will result in worse campaigns for advertisers and users alike, and weaker campaign attribution that will undermine trust in digital advertising.

The increasing importance of the web browser in marketing

Many of the changes proposed by Google hinge on browser-level technologies. Things like on-device processing require browser support to work. This is a potential challenge for Google and other ad platforms, as not every browser will support these features.

Google Chrome is the most commonly used browser by far with about 65% global market share, though it’s less popular in North American and European markets where Apple products are popular and Safari’s market share is higher. For now, Google Chrome is the only browser that includes some of the advanced features used by Privacy Sandbox. Advertisers targeting a North American audience will see their audience size shrink under these new policies. Take the Topics API for instance. If only about 50% of users are on Chrome, that means that you can only target about 50% of users. The Topics API is an opt-out feature. Savvy users can switch it off and avoid being targeted by clicking one button in their browser settings.

Other browsers competing for market share often do so by contrasting their features with Chrome. Mozilla’s Firefox browser includes a call to action on its homepage reading:

No shady privacy policies or back doors for advertisers. Just a lightning fast browser that doesn’t sell you out.

Including any of Google’s proposed Privacy Sandbox features that enable advertising seems highly unlikely. Advertising and user tracking are deeply unpopular practices among the general online community. Competing browsers are sure to highlight their independence from Google—and resulting lack of advertising features baked-in to their software—as a reason to jump ship to their application.

What B2B marketers can expect post-Privacy Sandbox

As so many parts of the Privacy Sandbox initiative are in development or may change prior to their full rollout, it’s hard to determine precisely what marketers can expect from the change. Many of the features discussed in this article have already launched ahead of the January 2025 full rollout deadline, but others are not finalized.

Based on past experience with major changes to marketing tech—namely the deprecation of Google Analytics 3 (UA) for GA4 last year—here’s what we expect to see from this change:

• Short-lived chaos in digital advertising. Just as with the sunsetting of the previous version of Google Analytics, while most marketers will be prepared for the death of third-party cookies, many will not be. We noted a brief dip in competition immediately following the GA shutdown. Advertisers who keep up with the changes may see the same minor boost in performance as their less sophisticated competitors fail to navigate the new advertising paradigm.
• Particularly high impact for low volume lead generation ad campaigns. Adding noise to reports makes them less granular and less accurate. This isn’t so disruptive for large, high-volume campaigns that are common for B2C advertisers—the noise added might only skew results by a single percentage point or less. However, for specialized B2B lead generation campaigns, this reporting noise could be significant. The fewer conversions recorded, the more impactful the noise since there’s less underlying “real” data to obscure. Though this noise will not impact the underlying bidding system of each ad platform, which uses conversion data as a training signal to improve performance, it may impact the ability for advertisers to make data-driven decisions regarding campaign direction. Advertisers will need to be careful about interpreting data lest the reports prove misleading. They may need to consider a more insight-based method for analyzing campaign performance or weigh indirect data sources like their CRM more heavily when making campaign changes.
• Potential for major changes in cost per click/cost per impression. Google will need to demonstrate to advertisers that ads still work despite weaker targeting options. We anticipate that each click on an ad will be worth less as targeting accuracy takes a hit, and that Google and others will need to compensate for that decrease in value. Expect dropping conversion rates counterbalanced by reduced cost per click.
• Gradual realignment of marketing spend. Many B2B companies rely on one or more forms of digital advertising for lead generation. Third-party cookies will be put to rest in Q1 2025. It will be interesting to see how marketing budgets adjust by the next quarter. If lead volume or cost per lead suffer under the new arrangement, budgets may rapidly realign toward other activities such as organic content marketing, trade show marketing, or account-based marketing.
• Less detailed reporting across analytics platforms. Many marketers noted a change in reporting integrity after the rollout of Google Analytics 4, as its greater emphasis on privacy results in less granular reporting. This trend ought to continue, requiring marketers to revise their reporting strategies and accept lower levels of accuracy and certainty.
• Continued reduction of advertising controls. Platforms like Google Ads have been “streamlining” their experience for years, pushing greater levels of automation while decreasing access to specific controls and granular reports. This may accelerate after third-party cookies go away. As reporting accuracy declines due to these changes, advertisers will have less reliable data to inform their decisions, which may require them to put more trust in automated systems anyway. Automated bidding systems will not be affected by noise and reporting delays implemented in reports—they will have access to the unaltered conversion data to drive their decisions.
• Emergence of new applications and services. Privacy Sandbox will not only reduce the accuracy of tracking data. It will also increase the complexity of implementing tracking. Whether through expansion of existing services or the creation of new ones, a market will emerge around simplifying the management of tracking data for advertisers. Expect new applications and services to emerge as part of your advertising workflow through 2025.

Conclusion: more transparency for users means less transparency for advertisers

No matter how you slice it, the Privacy Sandbox will be a boon to privacy-minded consumers and a challenge for data-driven advertisers. By delaying, obfuscating, and otherwise limiting reporting, Google will ensure that individual users aren’t tracked down through their behaviour online. Taken altogether, the changes will result in less data integrity at every level of advertising. While less impacted than analysts themselves, even Google’s internal and automated systems will have access to less data than now. In some ways, the story stays the same: cost per conversion has steadily increased for years just as conversion rates have steadily decreased.