Sample posts

Five posts in three founder voices. The founders, companies and numbers are invented; the point is the range.

Dan Whitlock, CEO, Tracewell

Pipeline monitoring. An engineer who became CEO: dry, specific, short lines, real numbers.

The feature we deleted

We deleted our most-requested feature last month.

Custom alert rules. 41 customers asked for it in 2025, and we built it in six weeks.

90 days later, 7 accounts had made a rule. 3 of those rules had ever fired.

What people wanted was to stop getting paged at 3am for a pipeline that recovers on its own. Custom rules were their guess at a fix, and we took the guess literally.

The replacement is boring. We wait one retry cycle before alerting on a transient failure. There are no settings. Pages to on-call dropped 38% across our customers in the first two weeks.

I still think listening to customers is the job. I just read requests differently now. A feature request describes a problem, written by someone who has already tried to solve it in their head.

Next time someone asks us for a toggle, my first question will be what woke them up.

Our best engineer came from support

The best infrastructure engineer we've hired never applied for the job.

Ana joined as our second support hire. Within a month she was attaching flame graphs to tickets. Within three she'd found a memory leak in our ingest workers that two of us had been chasing since spring.

We had no process for this. Support was support, engineering was engineering, and a Slack channel sat in between.

Two things changed after that. A ticket that needs a code change now goes to the support person who found it, with an engineer reviewing. And anyone can take an on-call shadow week (four people have so far, which surprised me).

Ana's title has changed twice since. Her first ticket is still pinned in #incidents.

I'd love to say we planned this. Mostly she ignored the org chart until we caught up.

Priya Raman, CEO, Ledgerline

Accounts payable automation. Numbers first, a little contrarian, ends on one practical rule.

The third approver

Mid-market finance teams spend about 9 days a month closing the books. We've watched more than 200 of them do it.

The slow part is rarely the reconciliation. It's waiting for a department head to approve an invoice that has sat in their inbox since the 4th.

In our data, an invoice that needs one approval clears in a median of 2.1 days. Two approvals take 6.8 days. Three take 13.

Each extra approver roughly triples the wait, because each one assumes someone else already checked.

Our fastest customers fixed it with a one-line policy: one approver below a dollar threshold, two above it, and nobody ever gets a third.

Those teams closed four days faster on average, and none reported more fraud.

If your approval chain has three names on it, ask who actually reads the invoice. It's usually the first one.

The deal we turned down

We turned down our biggest inbound deal of the quarter.

$40K a year from a logistics company with 14 entities and a CFO who wanted to sign within two weeks.

The catch was a custom integration with an ERP they had heavily modified in 2011. Our estimate was three engineers for two months, plus maintenance nobody would ever invoice for.

I ran the math the way I'd want a customer to run it. Two months of three engineers costs more than the contract's first year, and every other customer would wait those two months for what they'd asked us to build.

So I said no and sent the CFO the numbers. Honestly, I expected that to be the end of it.

He wrote back a week later. They're moving off the old ERP next year and want to talk again then.

Marcus Bell, founder, Kettle

Shift scheduling for hourly teams. Warm and story-led, opens with a person, ends on an honest open question.

My mom's legal pad

My mom managed a diner for 22 years. Every Sunday night she sat at the kitchen table with a pencil, a legal pad and the phone, building the week's schedule.

It took three hours. Then someone's kid got sick on Tuesday and she redid half of it from the back office.

When we started Kettle I thought the problem was the pencil. It wasn't. Most of our customers already had scheduling software. What they didn't have was a way to swap a shift without calling the manager.

So that's the first thing we got right. Two employees agree on a swap in the app, the manager approves with one tap, and the schedule updates.

Last month, 61% of shift changes across our customers happened without a phone call.

I showed my mom the numbers. She asked whether it works when the one who's sick is the manager.

It doesn't yet. That's next quarter.