When it comes to choosing a product and content recommendations vendor, evaluating the solutions available on the market can be a challenge.

There’s no debate that, when properly implemented, ecommerce recommendations are key to relevant, successful customer experiences. This ensures customers easily find the right products. Unfortunately, it has become increasingly hard for buyers to differentiate between recommendations vendors as they all seem to promise the exact same capabilities. 

To stand out from the crowd, some vendors have started to tout themselves as “cookieless” and that they are ready for the future of the internet without cookies and tracking. While the promise of cookieless technology seems to be intriguing, let’s look closer.

The reality is they are defining themselves by a negative…  but why? Even the notion of cookieless is really an admittance that they don’t have any means of personalization. So the question becomes, is it possible to have both personalization and privacy-conscious technology? 

In this article, we explain whyit is neither wise (nor necessary) to throw the baby (personalized recommendations) out with the bathwater (cross-site tracking and third-party cookies).

But first, a quick review.

What Are Cookies and How Do They Work?

It’s important to understand how cookies were intended to work, and how the internet used to run on consumer data. 

Back then, when a person visited a site, a third party would leave a cookie, a small piece of code, on that person’s browser. These ‘cookies’ would identify that user’s browsing data across multiple websites, enabling the tracking of consumer behavior across the web. This consumer data collected and sold, and helped retailers and other businesses track visitors, personalize experiences, and analyze performance. 

Third-party cookies have played an enormous role over the last 10-15 years in providing shoppers with recommendations in tune with what they are searching for. It was like your personal concierge knowing all about you – and alerting the next site that you visited that you were interested in gold kitchen sink faucets. 

An infographic describes the differences between first and third party cookies.

But all websites depend on cookies. And that’s especially crucial with sites you have to be authenticated to visit.

First-party cookies generate first-party data. This data allows a user to navigate a site without having to log in every time they go to a new page. This data contains very detailed records of what a visitor does when navigating an individual website or online store. It also requires asking customer permission. 

Initially,  consumers liked the advantages of a personalized experience that cookies were able to deliver. But as marketers started to push cookies into every portion of our online life — things got creepy.

Awareness of all this tracking and the data privacy practices needed to constrain it became a priority and eventually the law. Third-party cookies were put on the chopping block. This is why Google is phasing out third-party cookies in 2024. Browsers developers are either blocking third-party cookie creation or setting them to delete in a set number of days.

To demonstrate compliance with this new future, vendors have started to crow about how they are “going cookieless!” and there’s nothing to worry about. And that might sound great, until you remember that cookies are what helps personalize recommendations.

‘Cookieless’ Puts Positive Spin on ‘Non-Personalized’

Since May 2018 when GDPR went into effect (and then 2020 with California’s CCPA) sites have had to explicitly ask visitors permission to collect personal data from them.

And this is why the whole cookieless thing became the new must-have feature for recommendations.

But what does cookieless really mean and how is it affecting recommendations? 

We think it’s an attempt to distract from the real cost of non-personalized recommendations and cast them in a more favorable light.

Because when you go cookieless you lose cookies — and you lose personalization. 

The Cost of Non-Personalized Recommendations 

Consulting firm McKinsey estimated that 35% of what consumers purchase on Amazon and 75% of what they watch on Netflix comes from personalized product recommendations. Indeed, Amazon and Netflix have a unique ability to engage (and indulge) with each of their customers on a personal level. 

But if a shopper’s browsing behavior is collected by the company that owns that site — and not sold to a third-party — and used by the company’s recommendation engine, this is now considered within compliance since it’s a first-party cookie!

That’s why we think “going cookieless” is throwing the personalization baby out along with the cross-site tracking bathwater. Instead, companies need to find the vendors that are ready to comply with privacy regulation and handle technological developments without giving up on personalization.

To this end, make sure the vendor you choose offers appropriate tools for first- and zero-party data collection and a complete suite of recommendations strategies that encompasses models for personalization. Settling for cookieless recommendations means giving up on opportunities to generate more revenue and profit through personalization.

The Rise of First-Party and Zero-Party Data

Beyond first-party data, there is another type of data: Zero-party data. 

This is a specific type of data that a shopper intentionally and proactively shares with a company or site. It can be the result of filling out a survey, answering questions, completing a profile, or any other information the individual freely and consciously gives to you to recognize them. 

Both of these types of data are likely to play a critical role in personalizing experiences for your visitors.

While we’re talking about the rise of first-party data, it’s important to note that there are some who believe that the removal of third-party cookies will just be the first step towards removing all cookies. Yes, this is being discussed and proposals are being considered. However, it is far from clear what the alternative should and would look like — and whether these alternatives would indeed represent improvements from the perspective of privacy protection. 

In fact,the very notion of “privacy” means different things to different people and even different organizations, making these disputes especially hard to settle. Since there is no consensus in the industry with regard to the appeal of privacy-conscious alternatives to first-party cookies, it is safe to assume that at least for the foreseeable future, they are here to stay.

Suffice it to say, Coveo watches this space carefully, and will respond accordingly, if and when the regulations require it.

First-Party Personalization vs Third-Party Cookieless

Using the phrase cookieless has been a means of some vendors saying they are in compliance — while hiding from you that their products and software can no longer personalize. At Coveo we don’t like to say cookieless because we know we can still leverage the first- and zero-party data that customers have given permission to be used. With that data, and powerful machine learning models, we can offer a highly personalized experience and stay in compliance.

There are certainly some challenges when users don’t accept first party cookies, as in those cases it is not possible to track the impact of recommendations for those visitors (e.g. if they clicked or went on to buy).  

However, even when visitors don’t wish to be tracked on a website, it is possible and, in fact, imperative that the recommendations shoppers receive are always relevant.

Session-Based Recommendations

One example of a recommendations method that uses first-party data instead of third-party but still delivers powerful, privacy-conscious personalization, is so-called session-based recommendations. 

These personalized recommendations work by understanding visitor intent based on interactions observed in an ongoing browser session as someone visits a site and then analyzing the search and browsing behavior in real-time. 

Session-based data tailors a user’s experience to their interests — making them more likely to buy.

For example, imagine I’ve searched for some golf pants on an ecommerce website, which I’ve promptly added to my cart. Session-based recommendations may start showing me with golf-related items automatically. 

This is a key feature and a game-changer that helps ecommerce players deliver personalized shopping experiences without the need for high volumes of data or logged-in users.​ 

These session-based recommendations must remain relevant throughout the visitor’s session by also detecting intent changes and adjusting accordingly. For instance, you wouldn’t want to keep seeing golf accessories once you start shopping for your next tennis outfit. Simply put, any form of personalization that looks at visitors is dependent on cookies or analogous technologies because people must be identified. 

Innovative vendors that deliver privacy-conscious recommendations tend to leverage zero-party data and clickstream data for in-session personalization. 

It is to this end that Coveo has developed cutting-edge technology and provided “Personalization as you go,” using shoppers’ signals to tailor and improve the experiences that customers receive in real-time.

What to Do When First-Party Data Is Unavailable? 

So what do you do, when site visitors won’t click accept that you can collect first-party data at all? 

Well first of all, only a low percentage of people actually block first-party cookies. By rejecting them they are telling you they have no expectation of personalization.  Second, when first-party data is not an option, then you have to make do with just product and content recommendations.

Any model that is based on global data (popular products, trending products, new products) or that is based on product-to-product relationships (complementary products, similar products) can still function without cookies and personalization. 

While not as effective as personalized recommendations, cookieless recommendations can be used  to showcase  the most popular products of your store. These recommendations can help new shoppers save time, giving them access to some of the most valuable and potentially relevant products.

But again, even these come in two flavors – machine learning generated or not. Those that have ML powering recommendations will have far greater relevancy than those that are simple rule-based. So even when first-party cookies cannot be leveraged to deliver personalized experiences, it is critical that websites deliver experiences that are easy, effective and relevant. 

Unfortunately, this is not the case on most ecommerce sites. For instance, we can see here that when a user is viewing a protection case for an iPhone 12, the recommendations received aren’t really that relevant, as they’re for Samsung or iPhone 6. 

So to combat that, you need a recommendations engine that offers robust machine learning that associate complementary products – or have in place a myriad of manual rules that need to be continuously updated).

Privacy-Conscious Personalization

Rather than worry about finding a vendor that offers cookieless recommendations, we believe you should be looking for one that offers privacy-conscious personalization.

Don’t be surprised that few vendors are able to offer this. And the reality is that won’t change in the near future either. For example, Coveo has leveraged substantial amounts of funding, 15 years of machine learning experience, and the acquisition of a team of scientists and academics to be able to offer this mature personalization.

And because of this Coveo is a best-in-class recommendation and personalization engine solution that  provides the most relevant, personalized experiences when users accept first-party cookies. But also, when users do not accept first-party cookies, because of its machine learning models, Coveo excels at delivering relevant recommendations. 

Dig Deeper 

Want to find out more about types of personalization elements? Explore the Recommendations Periodic Table. Discover the 5 categories and 17 elements you can be using on your site, the types of data they consume and the KPIs that will be impacted!

Introducing Ecommerce Alchemy
The Periodic Table of Recommendations