Larry Weidig
Let’s skip the buzzwords. If data isn’t helping your team fix issues faster, answer questions better, or see problems before they happen, it’s not useful. It’s noise.
In our most recent Bandwidth episode, we tackled what it actually means to be data-driven across an ISP. Not just for execs running P&Ls, but for the folks handling installs, support tickets, dispatch, and everything in between.
Here’s the punchline: data is only valuable if it’s visible, timely, and relevant to the person looking at it.
And most people are getting at least one of those wrong.
Here’s the punchline: data is only valuable if it’s visible, timely, and relevant to the person looking at it.
And most people are getting at least one of those wrong.
Reports Don’t Help If You Forget to Open Them
We all like a well-crafted report. But unless it’s automatically delivered, scheduled properly, and filtered to be relevant to someone’s job, it’s probably going unread.
Sending a report is not the same as getting someone to use it. Even sending that report does not ensure it is opened and read. If the user is not finding the report valuable, it will just get lost in the inbox.
Frontline teams don’t need a 12-tab spreadsheet showing quarter-over-quarter churn by service area. They need to know why the person on the other end of the line isn’t getting a signal or why their billing flagged them for suspension.
Make it visual. Make it simple. Mount a screen on the wall and let the data run. You’d be surprised how many problems solve themselves when the right information is just in front of people.
Dashboards for the NOC Do Not Equal Dashboards for Support
There’s a big difference between metrics that help the Network Operations Center keep uptime above 99.99% and metrics that help a CSR de-escalate a billing call.
If you’re only thinking about data from a NOC perspective like signal levels, latency alerts, node saturation, you’re missing half the picture. The same goes for support leaders who only look at CSAT and call handle times. Not saying these don’t matter, they most certainly do, but they are not real-time.
Your field techs, dispatchers, billing reps, and installers each need a view tailored to their job. Same goes for timing. Real-time matters more than monthly summaries when a customer is on the phone asking why their Netflix isn’t working.
The Customer Might’ve Told You. You Just Didn’t Ask
One of the easiest ways to get useful data is to ask your customers. You’d be shocked how many operators skip CSAT or NPS in the first year or two. Why? Usually because they’re focused on buildout and think they already know what customers are experiencing.
That’s not how it works.
You don’t know what you don’t measure. And if you don’t give customers a way to tell you what’s broken or what’s working, you’re going to find out through churn. That’s a far more expensive feedback loop.
Gamifying the Metrics Doesn’t Have to Be Cringey
One of the better ideas floated during our discussion was gamifying data for your team. Not everything needs to be a contest, but if you’re measuring something valuable like upsells, first-call resolution, or successful installs, then show it.
Make wins visible.
We’re not talking about cheesy leaderboards or corporate noise. I mean surfacing meaningful metrics that show people what good looks like and how they’re doing relative to that. That’s it.
The Data You Ignore Will Eventually Cost You
There was a point in the episode where we talked about where data goes to die. It usually happens when someone sees a bad metric like churn ticking up or ARPU dropping or installs failing and no one takes ownership.
Either the responsibility isn’t clearly defined or it’s been defined but ignored.
Fixing that starts with accountability. Who owns this metric? Who’s responsible for actioning it? And what happens if they don’t?
If the answer is unclear, the result will be too. Everything that has been deemed worth capturing the data for needs an owner that accepts responsibility – EVERYTHING!
AI Doesn’t Replace Judgment. It Amplifies It
I’ll be honest. I’ve become a convert on AI. Not because it’s magic, but because it’s useful.
Take hundreds of support calls, run them through AI models, and you’ll find trends like sentiment, intent, or training gaps that you’d never catch by listening to five random recordings.
That said, don’t let it replace your brain. If an AI flags a node for future saturation, validate it. If it shows you something unexpected, dig in.
You don’t blindly follow AI outputs. But you shouldn’t ignore them either. If you are, what was the point of collecting the data in the first place?
Don’t Be a Packrat. But Don’t Be a Minimalist Either
Data hoarding is real, especially now that storage is cheap. And for good reason. Sometimes that weird latency blip from 6 months ago does come back around.
But not all data needs to be stored forever.
Roll up older data. Aggregate it. Keep it trend-level useful. You don’t need 5-minute SNMP logs from five years ago. But you might want usage trends or install success rates by season.
The goal isn’t to collect it all. It’s to keep what matters.
Final Thought: Data Is Everyone’s Job Now
If your field techs can’t see live tower status
If your CSRs don’t know the churn rate
If your dispatch team doesn’t see install failure trends
You don’t have a data problem. You have a visibility problem.
It’s not about building a single dashboard for leadership. It’s about making sure every person in your org has the info they need to do their job better. When they do, everything else improves. Faster installs. Lower churn. Better experience.
Don’t just collect the data. Use it.
Listen to the full episode on Spotify, Apple Podcasts, or YouTube.