How Midmark cut month-end close by two business days

Month-end close at most mid-market companies looks the same: the controller pulls the trial balance, the FP&A lead chases variance commentary, a senior accountant reconciles inter-company, and somewhere along the way someone has to walk into the CFO's office and explain why the deferred-revenue number moved.

At Midmark — a medical-device manufacturer where the close has to land before the executive review every month — that work used to be a four-tool relay race. The ERP held the journals. Salesforce held the bookings. Power BI held the variance dashboards. Spreadsheets held the actual checklist. Each lap moved the work between humans, and each handoff was where time leaked.

Eight weeks after rolling out MetaSpark, the close runs two full business days faster. Same team. Same systems. The agents took over the relay.

The relay race, before

Jon Dill, Midmark's Chief Financial Officer, walked us through the old shape of the close.

• Day −5 to Day 0: the controller's team works the trial balance in the ERP. Anyone who needs context Slacks the FP&A lead and waits.
• Day 0 to Day +2: FP&A pulls cross-system reports and builds variance commentary in spreadsheets. The version everyone is looking at is whichever one was emailed most recently.
• Day +2 to Day +4: senior accountants chase reconciliations. The list of open items lives in someone's head plus a shared sheet that goes stale by lunch.
• Day +4: the close meeting. The first 30 minutes are the team finding out from each other what's still open, what's blocked, and who owns what.

Nothing was broken. Nobody was bad at their job. The cost was simply the handoffs — every transition between tools or humans was where the close lost a day.

What changed

MetaSpark wired three connectors into the existing stack on the second call: the ERP (a Microsoft-stack accounting system), Salesforce, and the Power BI workspace where finance keeps its variance views. Two of those had ready-made connectors. The third was a small in-house tool the team used for inter-company reconciliations — the connector compiler authored it during the demo while Jon's team watched.

From there, the change was concentrated in one surface: a single ranked queue showing the entire finance department what was actually open across all four systems, with the gap-closing work assigned to the right person and the most urgent items at the top.

What the agents took on

• Pulled open journal entries, unreconciled accounts, and variance flags out of the ERP and BI every 15 minutes.
• Cross-referenced Salesforce bookings against revenue postings so reconciliation breaks showed up before someone had to ask.
• Drafted the variance commentary as a starting point — the FP&A lead edits rather than writes from scratch.
• Generated the close-meeting prep deck the morning of, with the ranked open-items list as the first slide.

Day-in-the-life, after

Now the close has a different rhythm. The finance team opens MetaSpark first thing during close week, sees the same prioritized list everyone else sees, and gets to work on what's at the top. Slack volume during close week is down — the ambient questions ("do we have the AR aging yet?" "is the inventory adjustment posted?") got answered before anyone had to ask, because the agents post the status into the queue.

The close meeting changed shape too. Instead of starting with status, it starts with decisions. The 30 minutes of mutual catch-up disappeared because everyone walked in with the same view of what was open and what was settled.

The numbers

• Close cycle: down two business days (~22% reduction from the baseline).
• Time spent on cross-system status-pulling: 8.5 hours per close cycle for the FP&A lead, down to under 60 minutes.
• Connector authoring time for the in-house reconciliation tool: under one minute on the demo call.
• Slack volume on the #finance-close channel during close week: down 60% week-over-week vs the previous quarter.

The number Jon cares most about isn't on that list. It's the share of close meetings that get to the strategic conversation — the bit about the business, not the bit about the close.

What this replicates

The Midmark pattern is finance-flavored but the shape is general. Any function that runs a recurring cycle across multiple systems — close, QBR, on-call rotation, monthly health review, board prep — pays the same handoff tax. The fix is not another dashboard. It is an agent that reads the systems you already run on, ranks the work that's actually open today, and assigns each piece to the human who can move it. Then the humans show up to the meeting that matters, not the meeting where they catch each other up.

What's portable from Midmark's setup

• Pick a high-stakes recurring cycle. The close is ideal because it has a fixed deadline and clear ownership.
• Connect every system that holds part of the truth. If we don't have a connector, the compiler writes one — bring the niche internal tool with you.
• Resist the urge to redesign the workflow. The agents work on top of the relay you already have; the relay just stops costing days.
• Measure handoff time, not raw output. The win is the day you stop running the catch-up meeting before the meeting.

If your team runs a cycle like this and you want to walk through what the first two weeks would look like on your stack, book the demo. We'll write any connector you're missing live on the call — same way we did for Midmark.