Predicting User Behavior in a Cookieless World

Recent changes to data privacy legislation and commitments by the browser giants to cease support for cookies mean that predicting user behavior is about to become much more difficult. However, these predictions are vital for those operating online, and companies who fail to find alternatives may fail more generally. First-party data can provide an answer, allowing providers to ethically and reliably determine how users are likely to act on their sites and to respond appropriately. Let’s take a look at what these changes mean for prediction, and how to continue to act in a user-informed way in a cookieless world. 

A New Age for Data Privacy

The Wild West of data has finally been conquered. Where advertisers once collected and shared personalized data recklessly and with near impunity, a slew of changes to legislation has drastically changed the internet landscape. Alongside the CCPA and GDPR laws in Europe and Silicon Valley, twelve other US states have passed comprehensive data privacy legislation (Virginia, Colorado, Connecticut, Utah, Iowa, Indiana, Tennessee, Texas, Florida, Montana, Oregon, and Delaware). So far, only California, Virginia (VCDPA), Colorado (CPA), and Connecticut’s (CTDPA) laws are effective, but more will follow in 2024.

These changes mean businesses who have previously been laissez faire when it comes to data handling, may be subject to harsh penalties unless they clean up their acts with haste. Those who continue to proffer overcomplicated opt-out systems and unsafe storage and collection of personalized data may fall foul of the extensive new legislation. To make matters more complicated, the demise of third-party cookies is imminent. Firefox, Google Chrome, and Safari have all committed to phasing out these devices in the near future, with Chrome scheduled to discontinue their use in 2025.

What Does All of This Mean for Predictive AI?

While some have placed their hopes in Google’s Privacy Sandbox, a feature intended to allow advertisers to continue to collect data online, the data this feature will offer is considerably less detailed than what advertisers have grown used to. Not only will it be entirely anonymized, but it won’t shed any light on what users do on your sites. Vice president of the Trade Desk, Bill Simmons, writes that ‘Privacy Sandbox will obscure the identity data about an individual user, making it harder to coordinate advertising across devices, and thus harder to measure and optimize performance. It could limit the ability to run attribution models, randomizing interest groups and contextual information’.

Even more significantly, cookies will no longer deliver the much relied-upon data about what users do on your websites, meaning that prediction can no longer rely on third-party data. Essentially, these changes mean that the user data that will be available through advertising will be far less detailed, harder to come by, and anonymized, obscuring insights regarding how your users are likely to act. 

With all of the above seemingly spelling doom for prediction, should one simply abandon the practice? The answer is absolutely not. The online realm is more competitive than ever and those who fail to predict user behavior will inevitably lose out to those who find alternative means. Failure to predict means treating all customers as though they are identical. Not only will this mean that users aren’t provided with the services and goods they most desire, but it may significantly reduce profitability for online businesses. Recent research shows that 91% of users are more likely to interact with sites that remember their preferences, while 65% of sites reported increased conversions are adopting a personalization strategy. So how can one continue to predict with accuracy? 

Cookies are Gone? Find a Healthier Recipe for Prediction

First-party data can be combined with AI to allow you to continue to predict and personalize based on a huge number of data points registered when a user visits your site. This data includes their location, device, the weather in a user’s location, what extensions they use, and whether they use incognito mode or favorite your page. AI software can be employed to analyze these data points as they relate to every user who visits your site, eventually building an incredibly detailed picture of your customers, who are broken into segments and can be targeted appropriately with promotions from there. 

For instance, those running a travel company can offer promotions based on the weather of the locations they service. If the weather is particularly fine in a given location, people may be less likely to travel, at which point larger discounts can be offered to incentivize a trip. However, if a user is operating from an expensive device and an affluent area, the discount can be decreased, given that the user is likely to have access to more disposable income. Location can similarly be harnessed by those offering subscription services to online publications or social media platforms, to suggest stories or profiles from the user’s area.

Meanwhile, for those operating online training and education services, eCommerce businesses, or otherwise looking for conversions, a prospective customer who favorites a page will be much more likely to pay for training or products than one who uses incognito mode. In the former case, a lower discount can be offered to make the most of a likely sale, while in the latter, a larger discount may convert an unlikely customer. 

The advantage of using AI in conjunction with first-party data is that it continues to learn and become more accurate over time, segmenting customers into increasingly smaller cohorts who can then be targeted with utmost precision. This doesn’t simply allow you to tailor deals to them, but to detect exactly when to deliver the promotion. For instance, those who use coupon browser extensions like Honey or Coupert, are likely to be more price-conscious and eager for savings. These customers can be targeted with larger discounts and at an earlier time than other customers, increasing the likelihood that they will purchase your services. 

Predicting the Future of Online Business

With radical changes coming to the digital landscape, those operating online must learn to predict not just the behavior of their customers, but how navigating internet privacy will alter in the coming months and years. Fortunately, first-party data provides a solution for both issues. With the demise of cookies and ever more stringent data security laws on the horizon, the anonymized data collected on your website is an ethical, accurate, and future-proof alternative. Using this data in conjunction with AI, your predictions regarding user behavior will segment customers into increasingly tailored clusters, to target them with products and promotions both desirable to them and profitable for your business.