As long as companies have sold digital products, the successful ones have actively sought new ways to appeal to customers with more personalized experiences. Product personalization has always been unambiguously positive, with no imaginable upper limit. More has always been better because there was so far to go.
Until now. The pursuit of personalization got an enormous boost with the introduction of infinitely scalable AI. We’re in a new age now. One where can instantly analyze data, run experiments, and generate product experiences tuned to exactly the right person at the right time.
We're on the verge of a new age of ultra-personalized digital product experiences. AI can already talk directly to customers. It can generate and deploy product changes. It’s only a matter of time until every user gets a unique version of a product, custom-tailored to their specific needs. Now that this detailed personalization is possible, teams should be adjusting their roadmap and planning to take advantage of the new possibilities. That would make customers happier than ever. Right?
Slow down.
Personalization is not an endpoint, it’s a spectrum
Before your team kicks off a workstream to accelerate personalization with AI, you need to make sure you’re aiming in the right direction. If you want to personalize as far as you can, slow down. You might be setting unrealistic expectations that end up backfiring on your team.
I like to think of personalization as a spectrum. On one extreme is the monolithic experience, with every user getting exactly the same generic experience. At this end of the spectrum, there is one product version that everyone uses.
On the other extreme is a hyper-tailored experience that uses AI to analyze every user and generate a custom experience for them. This product is a self-aware chameleon, intelligently evolving to consistently meet each user exactly where they are. At this end of the spectrum, there are as many unique experiences as there are unique users. It’s true N-of-1 personalization.
Every team starts with the monolith. Eventually, you run a few experiments and create a few cohorts. You write a few pop-up that appeal to each of those cohorts and implement some basic branching logic to make sure each user sees the right message for them. You also make small changes in the UX to appeal to your cohorts as you learn more about them.
With every new grouping of users and new experience path, you’re making progress along the personalization spectrum. As your product evolves, you will increase the number of experiments, guides, and cohorts you have in play. This effectively creates more versions of your product, each tailored a little bit more to a specific audience.
How Agents advance a team on the personalization spectrum
Amplitude are built to create more personalized experiences. Agents analyze all your company data and take action to run experiments, separate the broader audience into cohorts, suggest copy for guides, etc. They work the way your team already works, only faster and at infinite scale.
Agents increase the speed at which your team can run the standard product improvement cycle. As they learn more about your users, they’ll help segment your large audience into smaller, more accurate groupings and tailor a user experience for each one.
In effect, Agents move your team from one end of the personalization spectrum to the other faster than you ever thought possible. They’ll need some guidance, but Agents can help you take steps (or jumps) to customize your product for your users. Before you start, you need to determine exactly how far you want to push your personalization efforts now that Agents have removed the barriers.
The rest of this post will discuss two tips to help you make those decisions.
Take a small step before you take a big step
The AI wave is exciting. It’s hard not to look a few years down the road and imagine what could be possible with agentic personalization. But it’s best to understand those Agents as an accelerator, not a shortcut.
AI is the ultimate technological deep end. If you try to go too far right away, you’ll drown. Instead, take small steps. Work iteratively. Create one new version of your product for one specific group of users. Start by working with an Agent to identify a new cohort, run a new experiment, and create a new guide. Test the way your audience responds and track how their engagement changes. Measure if your updates impact the time they spend with your product, increase feature usage, improve conversion rates, etc.
Your goal shouldn’t be to embrace the end state of N-of-1 personalization. It should be to accelerate the iterative loop of creating new cohorts and building product experiences for more targeted groups of users. The secret to advancing on the personalization spectrum is not to take bigger steps, it’s to take small steps faster.
Your team has a difficult balancing act to do here. It’s obviously not optimal to only provide generic experiences. But there’s also danger in taking personalization too far. Think of the problems that could come up if unique users at the same organization had wildly different experiences. Or if a customer service rep was helping users who had similar roles but vastly dissimilar products.
By fragmenting your audience into groups that are too small, you lose the ability to catch bigger trends. At a certain level of personalization and cohort splintering, you lose the ability to perform significant analysis. There’s a lot of value in identifying the right time to stop personalizing your product.
Find a place where N-of-1 personalization makes sense
While it may not make sense to customize your entire product for each specific user, there are some parts of their experience where that level of customization is a worthy goal.
For example, a feature that allows customers to chat with a rep from your company should feel personalized. A blanket chat message is easy to identify and always leaves a bad impression on users. You want customers to feel like they have a direct connection to the company during chat interactions. There’s a reason why so many organizations dedicate whole teams of employees to handle this type of work. It’s important to do it well, which requires a high degree of personalization.
That exception helps us prove the rule: N-of-1 personalization should happen at the feature level, not the product level. Find the right features and use generative AI to help customers. Don’t force AI into your product just to try something new.
This AI Agents trend is the start of an exciting new era of product improvement. Honestly, I’m not entirely sure where it’s headed in the long run. The whole landscape is changing so fast and so dramatically that it’s impossible to plan along traditional timelines.
Here’s what I do know: the right step right now is to use AI Agents to make your product more personalized for more users. The technology to do that is already available. All you need to do is start small and keep experimenting to advance your level of customization. Just don’t take it too far.