This year, personalization and first-party data will collide, deep learning will advance, and data privacy regulations will help platforms survive.
Like a lot of startups, we’re spending more time lately looking back on 2022 and planning ahead for 2023. Before my co-founder and I started Miso, back when we were graduate students at the small data lab at CornellTech in 2014, the idea of privacy-first personalization felt like an intellectual challenge we couldn’t resist.
At times, even as the National Science Foundation and others took notice, it felt like we were building a solution for a technology problem that didn’t yet fully exist in the world. Data scientists weren’t paying attention to personal data, privacy, and especially personalization the way we are now. That’s changing, and 2023 is going to be the year when personalization and first-party data collide. As such, here are three trends data and analytics professionals need to watch in 2023.
Trend #1: As app tracking and third-party cookies continue to crumble, enterprises will need to rethink their first-party playbook
One simple truth of 2022 is this: when offered a choice by an app or website to track them across the web, a large majority of users just say no. Consumers will instinctually opt for essential cookies if given the choice, or will decline app tracking when iOS prompts them, and remember -- this is just consumers. Safari and Firefox (and soon Google Chrome) are proving to be hostile terrains for third-party cookies as well. Apple’s iOS continues to up the ante on privacy and with GDPR and CCPA fully in place, the times of widespread cookies and tracking are drawing to a close.
Although this is a move in the right direction, it won’t come without challenges, as third-party ads continue to decline in effectiveness and rise in cost. As a result, businesses will need to rethink their third-party strategies. The key is living in the pools and lakes of data we’ve all likely been ignoring for some time now -- first-party clickstream data. This approach enables practitioners to put to work the data they already have about what users are interested in and engaged with.
Thanks to strides in artificial intelligence (AI) and natural language processing (NLP), we can now operationalize individual user insights into hyper-engaging search, recommendations, and discovery experiences that keep customers coming back. These insights can be applied to email, SMS, push notification strategies -- even loyalty and referral programs. It’s a win-win for businesses and consumers.
Trend #2: Deep learning will continue growing, with further advancements in neural search and deep learning recommendation systems
Investing in machine learning to process the aforementioned first-party data is going to be critical, and the timing has never been better. Given what deep learning and transformer models have unlocked across neural search, recommendation systems, and predictive modeling, the possibilities are exciting. Not long ago, first-party clickstream logs were the sort of data that only a project manager or a BI analyst could love. For years the value was derived from collecting and analyzing them in commercial products.
Now, we finally have deep learning-powered tools and systems for understanding these clicks and the story they tell about users and visitors. For the first time, we’re seeing transformer models and encoder systems come online that can analyze and semantically interpret product catalogs, content, and ads beyond what they are and more for what they’re about.
When you combine these product or content insights with a user’s clickstream, you suddenly have an interpretive lens through which to see what interests a user. One that you can spin off into personalized semantic search or recommendation systems, multi-objective rankings for affiliate marketing, or promoted listings. You can even extend this into target marketing and determine what offers to make. This can lead to improved user experiences and revenue growth, and businesses will be able to do it quickly and at scale.
Trend #3: New data privacy regulations and practices will help more platforms survive and thrive
Changing the way we do business to address ethical and privacy concerns will be painful in the short term. It’s forcing a sea change in how businesses, publishers, and platforms view their internal data and what they do with it. From this, we’ll see a huge shift in the proficiency of how websites and apps personalize the user experience and put their insights to work in better direct marketing, loyalty programs, and advertising.
In the long run, this can only be good. The teams that fully commit to this strategy are going to see higher engagement, retention, and organic growth. They’ll unlock new revenue streams without exposing their user’s data or mistreating their trust. This ultimately means more platforms, marketplaces, brands, and publishers will have the agency and the ability to succeed. Essentially, stronger data regulations could actually be the rising tide that lifts all boats.
A Final Word
For a long time, we’ve had a “winner takes all” view of the internet. There could only be one big online grocer or ridesharing service. That isn’t particularly good or healthy for our online economy. The real thing we want is a vibrant ecosystem -- one in which consumers have fair choices for where they can shop, get their entertainment, and be informed. We have a shot at unlocking a completely new phase of the web.
It’s an exciting time in the world of data and analytics. Although some of the changes ahead may seem daunting, we’re at an inflection point. Businesses that find smarter, more ethical, and privacy-first ways to engage with their customers will be the clear winners, while those that don’t adapt will crumble like the third-party cookies that will soon be a thing of the past.
About the Author
Lucky Gunasekara is the founder and CEO of Miso. He has more than a decade of experience assembling interdisciplinary teams of scientists, designers, and engineers to build products quickly, based on the latest advances in NLP, machine learning, data visualization, and RecSys. You can contact him via LinkedIn.