Vibe coding better content analytics

Eric Ulken on March 10, 2026

It’s been… a while since I posted here, but I wanted to share a little vibe coding project I’ve been working on during my time as a John S. Knight Journalism Fellow at Stanford: It’s a rough prototype of a tool that demonstrates what I call “value analytics” (as opposed to the volume analytics we primarily use in the news industry). In particular, I am introducing two new metrics:

  • Quality completion rate: What percentage of users who started reading this article completed it in a reasonable amount of time? (To compute what’s reasonable, we measure the interval between when the user starts scrolling and when they reach the bottom of an article and compare that against the article’s “minimum skimmable” engagement time based on word count. If the dwell time on the article is at least as big as the minimum reasonable dwell time, we log a completion.) The Guardian has a version of this metric they call “deeply read“.
  • Value votes: How many completing users said the article they just read was valuable (compared to the total number of completions)?

Combined, these metrics go way further than volume metrics (like page views, engaged time and conversions) at capturing the actual value a user receives from a given article. (They can also form the basis of some interesting user-level engagement metrics, but that’s for a different post.)

How does it work? Gradually scroll to the bottom of this post and you’ll see. If you don’t scroll too fast, you should see a prompt to give feedback on whether this article was valuable. (There is frequency capping, so you might not see the prompt on every page view.)

A dashboard I built in Looker Studio then aggregates the responses into value metrics by article:

The dashboard can show completion and value scores by article or by topic.

If you want to see more about what is going behind the scenes, pull up this article in debug mode.

What do you think? Valuable?

An experiment: The Seattle Times in 10 Tweets

Eric Ulken on December 5, 2012

Inspired by Patrick LaForge‘s Times in 10 Tweets experiment at the NYT, I decided to do the same with a day in the life of The Seattle Times, based on the tweets of Seattle Times staffers. Here’s the result:


What do you think? Interesting? Or a waste of time?

Required reading from 20 years ago

Eric Ulken on September 8, 2012

The world is changing with amazing speed, and we need to pay close attention to what is happening. … No one in our business has yet launched a really impressive or successful electronic product, but someone surely will. I’d bet it will happen rather soon. The Post ought to be in the forefront of this — not for the adventure, but for important defensive purposes. We’ll only defeat electronic competitors by playing their game better than they can play it. And we can.

Here is a fascinating glimpse of the future of digital media and newspapers’ place in it as seen from the year 1992 by then-Washington Post Managing Editor Robert Kaiser. Read Mark Potts’ write-up, and then read the full memo.

As an aside, it’s worth noting that Potts debunks the now-popular notion that newspaper execs’ “original sin” was not pursuing paid digital models from the get-go. In fact, newspapers tried many variations on paid content over the years, and most failed. (Newspaper executives were negligent in many other ways where digital is concerned, but on this count I think they deserve a pass.)

What’s most striking to me, though, is how the radical changes in technology that were predicted at this conference 20 years ago have unfolded more-or-less the way the technologists foresaw. (Remember, this is before anybody outside academe and big science had heard of the web, and 2400 bits per second was probably the most common connection speed for home computers to online services.) In fact, one of the few major technological advancements of the last two decades not mentioned in Kaiser’s memo is the explosion of mobile.

The thing nobody seems to have adequately imagined is how these changes in technology would shift the foundations of mass media in two fundamental ways: By disrupting the advertising business model and by giving rise to search and social media (and thus disaggregating the “bundle” that gave newspapers much of their value).

It would seem the near future is fairly easy to predict in purely technological terms. What’s harder to see are the economic and social implications of those changes. Credit to Bob Kaiser for trying.

Flash in the pan?

Eric Ulken on August 20, 2010

I am an ex-Flash user. I uninstalled the Flash plug-in on my primary browser about a month ago, and I haven’t looked back. Here’s how it happened:

Back when Apple announced that its forthcoming iPad would lack Flash support, it sounded to me like a boneheaded move. If a device built for consuming multimedia doesn’t support the web’s leading format for multimedia presentation, what good is it?

But after using my own iPad for a while, I decided I didn’t miss Flash nearly as much as I thought I would. (And I discovered that a lot of web apps I’d assumed were Flash-based were actually built with JavaScript.) Which got me thinking: Could I do without Flash on my main computer as well?

I use a woefully underpowered first-generation MacBook Air that I’d rather not replace just yet. I’d done about all I can think of to squeeze a little more performance out of it, including installing a solid-state hard drive and upgrading the OS to Snow Leopard. Still I found many common activities, particularly web browsing in multiple tabs or windows, painfully slow.

So I decided, as an experiment, to remove the Flash plug-in from my primary web browser, Google Chrome. I still have it in Safari, which I fire up when I need to look at Flash content.

After about a month, here are my impressions:

  • The speed increase on web browsing is much more dramatic than the performance boost I got by adding Snow Leopard and the SSD. And since most of my computing time is spent in a web browser, that gives my old laptop a new lease on life.
  • Fewer obnoxious ads! That alone might make this “upgrade” worthwhile.
  • YouTube and Vimeo both have stable HTML5 video players, though most of the commercial content on YouTube is available only in the Flash player.
  • The Wall Street Journal’s video player works beautifully. (NYT and CNN not so much.)
  • Interactive charts in Google Analytics and Google Finance, sadly, are Flash-based and don’t downgrade gracefully.
  • The Google Maps API is JavaScript based (though there is also a Flash API) so most of the apps produced by my former colleagues at the L.A. Times’ Data Desk still work without a hitch.

In short, for me, the performance jump is worth the occasional inconvenience, but YMMV.

What do you think? Could you live without Flash?

Search trends and geography

Eric Ulken on August 22, 2009

I know it’s been around for a year or more now, but I still can’t stop playing with Google Insights for Search, that small window into the universe of data that Google collects on user behavior. It’s a trend-spotter’s dream, and — particularly with its geographical filters — a potential source of story ideas for journalists.

For example, I can see the fastest rising search terms in Los Angeles and Berlin in the past week.

But what I’m finding fun right now is plugging in brand names and seeing where they’re strong. I offer, by way of example, some vehicle brands and maps showing search volume in the United States:

The four-wheel-drive Subaru line is understandably popular in mountainous, cold-weather states:

Geographic search trends for "subaru"

Saab has some fans in New England, but that’s about it:

Geographic search trends for "saab"

Another ailing GM make, Hummer, still gets some interest in Nevada and Texas (I suspect the bump in interest in Michigan may be mostly from concerned GM stakeholders):

Geographic search trends for "hummer"

Toyota’s Prius, meanwhile, is especially popular in eco-conscious places such as California and Vermont:

Geographic search trends for "prius"

And the Vespa scooter craze seems to have taken hold on the West Coast and, inexplicably, Utah:

Geographic search trends for "vespa"

When you look at popularity over time for all five brands, you can clearly see how interest in the brands associated with fuel-sipping vehicles spikes during periods when fuel prices are high. No huge surprises here, but it’s fun to see how well search data tracks real-world trends.

Have you used Google’s search trend data for story ideas? Share your tips in the comments.