Why the planning tool didn't replace the spreadsheet

I’ve bought two planning tools as a finance leader. The first was Anaplan, at a company big enough to have its own model builders. The second was Pigment, when I was running finance at a company growing fast enough that the annual plan was out of date by March. Both times, most of the planning still happened in Excel.

Neither was a bad decision. Anaplan handled more dimensions than any spreadsheet could, and Pigment was quicker to change and easier for the business to use. But a year and a half after each implementation, the plan was being built in a spreadsheet and loaded into the tool at the end.

I noticed it in a board prep meeting. The CEO asked what the plan looked like if we slowed hiring in the second half, and my head of FP&A opened a spreadsheet. Nobody in the room thought that was odd, including me.

The model was never the hard part

A planning cycle took us about ten weeks. We collected inputs from eight budget owners, and most came in late. Sales sent a bookings forecast that didn’t match the CRM, and reconciling the two took a day. The CEO asked what happens if the enterprise deals slip a quarter, so we rebuilt the model. Someone changed the hiring assumption, and everything downstream of headcount had to be checked by hand. Then the board pack was rebuilt from scratch, and by the time it was presented some of it was two weeks old.

Almost none of that was modeling. It was chasing, reconciling, rebuilding and explaining, and very little of it happened inside Anaplan or Pigment.

Why the spreadsheet kept winning

Nobody had to wait. In Anaplan, a new driver meant a place in the model builders’ queue. Pigment was faster, but a structural change still went through the one or two people who understood the model. In a spreadsheet, we could answer the question before the meeting ended.

And most of the models a finance team builds aren’t the annual plan. They’re a hiring case, a pricing change, a question the board asked on Tuesday. Those never went into the planning tool. They lived in a file with a version number in its name, and they were gone the following quarter.

I tried the obvious fixes. Stricter deadlines got the inputs in earlier and no better. A dedicated admin for the tool made changes faster until they became the bottleneck. More analysts worked for one cycle, then the business grew and they spent their weeks chasing like everyone else. Each fix moved the work to someone else.

The work that needs people and agents

Every step of that cycle mixes two kinds of work. There’s assembly: gathering, matching, recalculating, formatting. And there’s judgment: deciding what a number should be, which scenario to believe, what to tell the board. They’re too tangled to hand a whole workflow to software. What you can do is split each step, so an agent does the assembly and the judgment stays with the person who answers for it.

Collecting inputs. An agent drafts each budget owner’s input from their own systems, like the pipeline in the CRM or open roles in the applicant tracker. The head of sales still owns the number, corrects the draft and says why.

Chasing. An agent knows who owes what and by when. It follows up where each budget owner already works, with their draft attached, and tells FP&A early who is going to miss the date. Deciding when a late input is worth escalating, and talking to the owner who is behind for a good reason, stays with a person.

Working across teams. A change in one team’s numbers lands on everyone else’s. When sales pushes deals out a quarter, the hiring plan, commissions and cash all move, and we used to find out in a meeting two weeks later. An agent can tell each owner what changed and what it does to their numbers, the day it changes. The trade-offs between teams, like whether marketing’s budget moves because sales slipped, are for the leaders to settle, and FP&A approves what goes into the plan.

Reconciling. When the sales forecast and the CRM disagree, an agent finds the gap and shows where each number came from. Deciding which is right is the analyst’s call.

Scenarios and variances. When an assumption moves, an agent reruns the model and checks everything downstream. Each month it ties what moved back to the deals and hires behind it. Which scenario to plan against, and whether a shortfall is a blip or a trend, takes someone who knows the business.

The board pack. An agent assembles it from live numbers, each traceable to its source. The story is the CFO’s: what to lead with, what to flag, what the board needs to hear before it asks.

Two rules make the split work. Every number an agent touches shows its workings, because I never put a number in front of a board that I couldn’t trace. And nothing an agent drafts goes into the plan until a person approves it.

What I’d tell a CFO buying a planning tool

If your planning tool works for the model, keep it. Before you buy a new one, look at where your team’s hours go in a planning cycle. If most of them go on chasing, reconciling and rebuilding, a better place to put the model won’t give them back. Go through the cycle step by step, separate the judgment from the assembly, and take the assembly off your team.

That split is what we’re building at Quro. If your planning tool is sitting beside a spreadsheet that does the real work, I’d like to hear how you ended up there.