You Didn't Go Over Budget Last Week. You Found Out Last Week.
The project hits 94% and nobody on the delivery team saw it coming. Not because the signal was missing — because it lived in a tool none of them open, and arrived a month after it mattered.
The export lands on the first of the month. Someone opens it, and the project everyone thought was fine is at 94%.
There are two weeks of work left in the plan.
What follows is familiar. A quick huddle. Someone pulls the timesheets apart line by line. Someone else drafts a careful email to the client. And underneath all of it is the same question nobody quite says out loud: how did none of us see this coming?
You weren’t missing the signal. You were missing it on time.
Here’s the thing about that 94%. It didn’t happen on the first of the month. It happened gradually, over six weeks, and it was visible the entire time — in Harvest.
Which nobody on the delivery team opens. Because why would they? They’re in Linear and Figma and Slack all day. Harvest is where you log time, not where you go to find things out.
So the number existed, accurately, the whole time. It just lived one tool away from every decision that could have changed it.
Budget overruns aren’t sudden. They’re discovered suddenly. That distinction matters, because the two have completely different fixes. If the problem were the spending, you’d need more discipline. If the problem is the discovery, you need the number somewhere people already are.
The signal that shows up first
The total is a lagging indicator. By the time it’s alarming, the money is gone.
The leading indicator is much less dramatic and shows up much earlier: someone was booked for ten hours and logged twenty-one.
That’s week two. That’s not a crisis, and it’s not anyone doing anything wrong — a task turned out to be harder than the estimate, which is the most ordinary thing in software. But it’s a fact that, seen at the time, changes what you do next. Seen at the end of the month, it’s just part of the explanation for a number you can no longer change.
Almost nobody looks at that comparison, because it lives in two systems. Scheduled allocation is in one tool. Actual logged hours are in another. Putting them side by side is a manual job somebody has to decide to do — which means it happens when someone is already worried, which is after the useful moment has passed.
Where Spectr helps
Connect Harvest to a project and the budget picture stays current: budget versus spend, hours logged per member, and the resulting health, updated on sync rather than reconstructed at month-end.
Connect Forecast on top and you get the comparison that actually predicts the overrun — scheduled allocation next to actual logged hours, per person, per week. That’s where “booked 10, logged 21” stops being an archaeological finding and becomes something you notice while it’s still week two.
None of it is editable, deliberately. There’s no endpoint that accepts a budget or a cost. The numbers only ever come from the provider, so what you’re looking at is what your time-tracking says, not what someone typed in optimistically.
The part that makes it usable
Here’s the detail that decides whether any of this survives contact with a real team: hours are visible to everyone, dollar figures are admin-only.
That’s enforced server-side, not hidden in the UI. A developer seeing the project is at 80% of its hours is useful — it’s context for the estimate they’re about to give. A developer seeing the client’s rate card is a completely different conversation, and not one most agencies want to have by accident.
Split that way, the number can live where the work is decided instead of behind a permission wall. Which was the whole problem.
The honest limit
This doesn’t stop you going over budget. Nothing does — you’ll still misjudge scope, and things will still turn out harder than they looked.
What it changes is when you find out. Week two instead of week six is the difference between a conversation about trade-offs and a conversation about apologies. You still might land at 94%. But you’ll have chosen it, and you’ll have told the client while they still had options too.
The worst version of this isn’t going over. It’s going over quietly, and everyone finding out at the same time — including the client.
Spectr syncs project budget and per-member hours from Harvest and Forecast, alongside the specs and stories it generates from your meetings — with dollar figures kept to admins and hours visible to the team. Free tier available. Try it at spectr.pm.
