Your Year in Review Isn’t for You

Why I stopped accepting vanity metrics and built something useful instead

At most tech companies, we’ve largely determined that the thing we generally want people to do is to keep showing up and paying us (either with their data, with their money, or both).

As a result, the metrics that dominate board decks tend to answer a familiar set of questions:

• Do people show up regularly? (Start by measuring your daily or weekly active users,)

• Do they stick around? (Measure your customer retention rate and churn)

• Do people to pay us once? (Measure your conversion rates from free to paid users)

• Do people to pay us more than once? (Measure your recurring revenue streams or repeat customers)

• Are we making more money than before? (Measure your net revenue)

Here’s an example of what this looks like in action for Spotify:

These are sensible questions for running a business. They’re just not very helpful for running a life. The actionable data we need as end users looks quite a bit different.

One thing that strikes me about these “Year in Review” summaries is how much effort goes into producing glossy, sycophantic slideshows that offer almost no actionable insight to the user.

The irony is that the companies behind these reports are the same ones that preach data-driven decision-making. Entire teams (and entire businesses) exist to help organizations extract useful, behavior-changing insights from data.

And yet. When it comes time to reflect that same rigor back to users, the bar drops dramatically.

In tech, we like to say that data only matters if it helps someone make a decision. A number stripped of context, or a metric disconnected from behavior, is just trivia.

But rather than empower users with data that might actually change behavior, we give them applause and a highlight reel. As Saturday Night Live recently joked, we now know more useless facts about ourselves than ever.

Let’s unpack for a minute the shallow interpretation of my personal LinkedIn corpus of data.

This year, I used LinkedIn 344 days out of the year, posted content 259 times, and added 315 new connections. And that’s as much synthesis as LinkedIn decided to offer.

As a user, I’m left asking, “Why should I care?”

Consider instead this alternative framing.

An analysis of my post content and the roles of my new connections could signal intent toward specific career ambitions, especially when layered on top of millions of other users doing the same.

Instead of offering me with banal numbers, LinkedIn could offer a much richer artifact.

• LinkedIn already knows I started a business.

• It knows what I post about.

• It knows who I’m connecting with (and who people like me tend to become).

What it withholds is the only insight that actually matters:

Here’s what we think you should do next.

That’s where they make their money, of course. It would be a shame to give it away for free.

Notably, ChatGPT’s interface and presentation of its annual reported prompted me to do something a little bit different: It invited me to talk back.

The final screen of ChatGPT invited this series of prompts:

• Tell your friends to try ChatGPT

• Make a video with Sora

• Plan your 2026 goals

• Plan a long weekend trip

Now, I didn’t end up doing any of these. But the prompts did give me an idea.

Since ChatGPT already knows so much about me, my year, and my problems (remember: you’re looking at a 1% power user over year), I wondered: What could it help me create to help me solve my problems?

We quickly circled around one primary problem area (managing work/life balance and an overloaded nervous system) and decided to build something to address it.

Less than an hour later, I built Bethany’s Load Balancer, a mobile-optimized daily mood tracker for myself to help me track both daily trends and weekly demands in a light-touch, highly customized way.

ChatGPT helped me think through possible solutions and generate a starter prompt, and I built it with Replit using lightweight authentication and a simple database so it works across devices.

For me, this represented a meaningful shift. Instead of being handed a highlight reel about my behavior, I turned my own data into something I could actually use. And that feels like a much better place to start.

At most tech companies, we’ve largely determined that the thing we generally want people to do is to keep showing up and paying us (either with their data, with their money, or both).

As a result, the metrics that dominate board decks tend to answer a familiar set of questions:

• Do people show up regularly? (Start by measuring your daily or weekly active users,)

• Do they stick around? (Measure your customer retention rate and churn)

• Do people to pay us once? (Measure your conversion rates from free to paid users)

• Do people to pay us more than once? (Measure your recurring revenue streams or repeat customers)

• Are we making more money than before? (Measure your net revenue)

Here’s an example of what this looks like in action for Spotify:

These are sensible questions for running a business. They’re just not very helpful for running a life. The actionable data we need as end users looks quite a bit different.

One thing that strikes me about these “Year in Review” summaries is how much effort goes into producing glossy, sycophantic slideshows that offer almost no actionable insight to the user.

The irony is that the companies behind these reports are the same ones that preach data-driven decision-making. Entire teams (and entire businesses) exist to help organizations extract useful, behavior-changing insights from data.

And yet. When it comes time to reflect that same rigor back to users, the bar drops dramatically.

In tech, we like to say that data only matters if it helps someone make a decision. A number stripped of context, or a metric disconnected from behavior, is just trivia.

But rather than empower users with data that might actually change behavior, we give them applause and a highlight reel. As Saturday Night Live recently joked, we now know more useless facts about ourselves than ever.

Let’s unpack for a minute the shallow interpretation of my personal LinkedIn corpus of data.

This year, I used LinkedIn 344 days out of the year, posted content 259 times, and added 315 new connections. And that’s as much synthesis as LinkedIn decided to offer.

As a user, I’m left asking, “Why should I care?”

Consider instead this alternative framing.

An analysis of my post content and the roles of my new connections could signal intent toward specific career ambitions, especially when layered on top of millions of other users doing the same.

Instead of offering me with banal numbers, LinkedIn could offer a much richer artifact.

• LinkedIn already knows I started a business.

• It knows what I post about.

• It knows who I’m connecting with (and who people like me tend to become).

What it withholds is the only insight that actually matters:

Here’s what we think you should do next.

That’s where they make their money, of course. It would be a shame to give it away for free.

Notably, ChatGPT’s interface and presentation of its annual reported prompted me to do something a little bit different: It invited me to talk back.

The final screen of ChatGPT invited this series of prompts:

• Tell your friends to try ChatGPT

• Make a video with Sora

• Plan your 2026 goals

• Plan a long weekend trip

Now, I didn’t end up doing any of these. But the prompts did give me an idea.

Since ChatGPT already knows so much about me, my year, and my problems (remember: you’re looking at a 1% power user over year), I wondered: What could it help me create to help me solve my problems?

We quickly circled around one primary problem area (managing work/life balance and an overloaded nervous system) and decided to build something to address it.

Less than an hour later, I built Bethany’s Load Balancer, a mobile-optimized daily mood tracker for myself to help me track both daily trends and weekly demands in a light-touch, highly customized way.

ChatGPT helped me think through possible solutions and generate a starter prompt, and I built it with Replit using lightweight authentication and a simple database so it works across devices.

For me, this represented a meaningful shift. Instead of being handed a highlight reel about my behavior, I turned my own data into something I could actually use. And that feels like a much better place to start.