A GM I talked to last quarter had a board meeting at ten on Tuesday. He started writing the board report on Saturday at nine in the morning. His kid had a baseball game at one. He missed it. Not the first one he had missed for a board report, either.
He runs a regional fiber operator. Forty thousand subscribers across two states, six people on his board. A couple of outside investors, two operators who have actually run networks, two community members. They are sharp. They read fast. They ask the questions you were hoping they would not.
His board reports are good. They are also expensive. Each one costs him about ten hours of weekend, and they are the kind of hours where your head is buried in EBITDA-to-budget variance and everyone else in the house has quietly given up on you for the day.
Here is the part that always gets me. The data was never the problem. The numbers already exist. Subscribers are sitting in the BSS. ARPU is in the financials. Net adds are in the report his ops manager already wrote for the weekly. EBITDA is in the P&L his CFO already closed. Nobody is inventing anything.
The actual work of a board report is translation. You are taking numbers that already live in five different systems and turning them into one document a financially fluent board can skim in twelve minutes and then poke holes in. That translation is what eats the Saturday. Not the math.
I keep coming back to one question lately. Of all the back-office work in an ISP, which piece is most worth handing to AI first? The answer is almost never the flashy stuff. The most useful AI jobs for a Tier 2 or Tier 3 operator are the quiet ones that hand the executive back twenty percent of their thinking time. Board reporting is right at the top of that list.
So here is what I would do if the board report were eating my weekends. I would open Claude or ChatGPT. I run both, the $20 tiers, nothing fancy. I would paste in a prompt that already knows how a board report is supposed to be built: executive summary, financial performance, operational highlights, strategic priorities, items that need the board's input.
Then I would feed it the numbers I already track. Subscribers this quarter versus last. Net adds. Churn. ARPU. Revenue against budget. EBITDA. Project updates. Anything notable. The same numbers already sitting in the financial close package and the ops dashboard.
What comes back is a draft. Not finished. Not ready to send. A first draft with the right bones and a reasonable story. Then I do the part only a human can do. I read it. I edit it. I argue with it where it got the emphasis wrong. I add the judgment about what this board actually needs to focus on this quarter. And I check every number twice, because AI will hand you a confident wrong figure with a totally straight face.
The full prompt is in the playbook, along with the ones that cluster around it. The board meeting talking points. The investor update letter. The executive summary for the giant operational report nobody wants to read. The annual report narrative.
The reason this matters is not really the ten Saturday hours, though I will happily take those back. It is what those hours were costing him. They were the hours he was not spending on the questions the board was actually going to ask. Whether to push into the next service area. Whether the BEAD timeline is real or wishful. Whether his leadership team is deep enough for what is coming.
The work is the meeting. The document is just the thing you walk in with.
That GM did his board meeting on Tuesday at ten. The board asked the questions he expected and a couple he did not. He handled the expected ones fine, because he had the data. He handled the surprises better than he would have, because he had spent his weekend thinking instead of typing.
Sonar is what runs underneath the day-to-day at a lot of regional fiber operators. Billing, network, subscriber management, inventory, field techs, in one place. We built the AI playbook because most of the back-office work happens around the platform, not inside it, and it is free whether or not you run Sonar. You can see the platform at sonar.software.
Get the full ungated playbook HERE.
If you are running ops or finance at a Tier 2 or Tier 3 operator and the board report is quietly eating your weekends, the full prompt is in there. Free download. No form.
Tell me if it helps. And if it does not, tell me what is still hard.
How long should an ISP board report take with AI?
Once you have a reusable prompt and the numbers you already track, the first draft comes back in well under an hour. The time then goes where it should: editing, checking the figures, and adding your own read on what the board needs to focus on.
What numbers go in a Tier 2 or Tier 3 board report?
The usual set is subscribers this quarter versus last, net adds, churn, ARPU, revenue against budget, EBITDA, key project updates, and anything notable. Most of it already lives in your financial close package and your ops dashboard.
Can AI write the whole board report for you?
No, and you would not want it to. It gives you a structured first draft with reasonable bones. You still verify every number and add the strategic judgment, because that part is the actual job.
____________________________________________________________
Lindsay Kelley is VP of Sales & Marketing at Sonar Software, a BSS/OSS platform built for Tier 2 and Tier 3 broadband operators. She writes about AI, marketing & sales operations, and what it actually looks like to scale a lean team.