Skip to content
Case 04 · Automated trend tracking

It reads the whole internet. The team reads one page.

The things your customers will care about next month start small, in forums, videos and newsletters, long before anyone calls them a trend. Nobody has time to watch all of it. This began as a sketch of a tireless listener: spot the early signs, sort them into stories, judge how likely each one is to break out, and tell the team what’s worth acting on. Today that report lands every Monday morning.

the idea: listen group score judge report to the writers …every Monday, before anyone’s had coffee
the first sketch↓ what it became
1Monday, 9 a.m.

It listens where trends start.

Forums, video channels, journalists, newsletters, brand blogs. It listens for the topics the team has chosen to follow, notices when a conversation starts to percolate, and rates each one against a few clear signs of where it’s heading.

Designed forListening where the audience actually talks, not just where it’s easy to collect. When a platform closes a door, the system finds a polite way around it or a person checks by hand.
forums videos journalists newsletters a manual pass every source, one place
This week434 new posts
2Monday, 9:04 a.m.

It groups the noise into stories.

Hundreds of posts become a handful of stories people keep coming back to. One post that went viral on its own doesn’t make the cut.

Designed forPatterns, not spikes. A story has to recur across many posts and places before it counts.
the one-night reset group-trip money fights pet-friendly everything one viral post ≠ a trend
3Monday, 9:06 a.m.

It scores how far each story has spread.

It looks for five signs that a story is breaking out of its corner of the internet. “The one-night reset” shows up in lots of communities, on several platforms, and brands are starting to use it. Nobody can tell yet whether journalists have noticed, so it says so.

Designed forEvidence over excitement. When the data can’t show a signal, it’s marked unknown, never guessed.
StoryThe one-night reset
 many communities more than one platform journalists noticing brands joining in picking up speed unknown 4 of 5 breaking out
4Monday, 9:07 a.m.

It knows when to break its own rule.

“Who pays for the group trip?” scores 1 of 5, which usually means noise. It’s recommended anyway: it isn’t a wave that will pass, it’s a headache every group trip has, and nobody is helping with it.

Designed forJudgment, not just counting. The score informs the call; it doesn’t make it.
StoryWho pays for the group trip?
 1 of 5 · usually noise act anyway it’s not a wave. it’s a problem every group trip has.
5Monday, 9:08 a.m.

It says when it isn’t sure.

It had been tracking “staycation” to measure the one-night reset, and the line was flat. It noticed people don’t say that anymore. They say “reset weekend.” Its advice: fix the measurement before deciding.

Designed forHonesty about its own blind spots. A system that admits a bad measurement is one you can trust when it’s confident.
measure the right thing first “staycation”: flat “reset weekend” weeks →
6Monday, 9:15 a.m.

The report lands.

One page the team can read over coffee: what to act on this week, what to keep watching, and a hard-pass list with reasons, so nobody asks twice. Today a person picks which stories go to the writers. Next, the two systems run as one, with room for the team’s own ideas.

Designed forA report people actually read. Every story comes with its evidence one click away, and every illustration is safety-checked. If in doubt, it ships without one.
The Monday report · week 7Act nowThe one-night reset ●●○●●Who pays for the group trip? ●○○○○On the radarPet-friendly everythingHard passAirport-lounge hacks · one viral postCastle rentals · noise
 so nobody asks twice read in 5 minutes → the writers (case 05)

What it took to make it hold up.

Spotting one trend is luck. Reading everything every week, telling a story from a spike, and admitting what it can’t see is the real job.

The real system: 48 sources watched · about 4,900 posts in its archive · 5 to 8 stories a week, each scored on 5 signals

A

Accurate

A story has to recur across many posts and places. When the data can’t show a signal, it’s marked unknown rather than guessed.

R

Robust

It keeps listening when platforms get difficult, and the moment a weekly run fails, its owner gets an email with the reason.

L

Learns as it goes

Each week it reads last week’s stories, notes what sped up or faded, and adjusts the words it tracks.

People stay in charge.

  • The people who know the brand set its topics, audiences and calendar moments.
  • A person decides which stories become content.
  • Every hard pass is recorded with a reason.

Same idea, other places.

It fits anywhere a team needs to hear what people are saying before it becomes obvious.

  • Product and research teams listening to customers
  • Retail and seasonal planning
  • Newsrooms and editorial desks
  • Brand and reputation monitoring