Reforecasting the day — seeing 3pm at 11am

Real-time management · ~6 minute read

The single most valuable thing a real-time analyst does is see the rest of the day before it happens. The run-rate reforecast is simple arithmetic — and powerful precisely because it is available early, while the fix is still a nudge rather than a scramble.

The run-rate reforecast

Once a meaningful chunk of the day’s actuals are in, you can project where the day lands: if actuals are running X% off forecast and the intraday shape holds, the day will land roughly X% off. Concretely: forecast 8,000 contacts, ~45% of the day normally in by 11am, actuals to 11am running +12% — the day projects to land around 8,960, and you can re-derive the afternoon’s interval requirements now, hours before the afternoon arrives.

The formula is just projected day = actual-so-far ÷ share-of-day normally in by now. Nothing clever — and that is the point. The value is not sophistication but earliness: the difference between calmly arranging afternoon cover at 11am and scrambling at 3pm is not analytical brilliance, it is having done the arithmetic four hours sooner.

Adjust for the cause — the assumption has limits

The run-rate’s key assumption is that the day’s shape holds — the divergence is roughly proportional across the remaining intervals. One-off events break it: a spike that will not repeat should not be projected across the afternoon. Shape-changing causes break it too: a campaign that front-loads the day means the morning’s +18% will not run through to close.

This is why the reforecast leans on the variance work that precedes it. A quantified divergence with a known cause can be projected intelligently — a bill run persists all day, a momentary outage does not. The analyst who applies the run-rate blindly misprojects; the analyst who adjusts for what they know about the cause projects the day the operation actually gets. And the projection is not a one-off: refresh it as more actuals arrive, because it sharpens through the day.

Leading indicators — the signals that move first

Beyond the volume projection, the anticipatory desk watches the signals that move before service does: volume climbing against forecast, AHT creeping, availability dropping, a queue starting to build, login numbers falling against plan, a known event — a campaign send, a predictable peak — about to hit an already-stretched operation. Each is a warning delivered while there is still time to use it.

The lagging alternative is the service tile, and by the time service has visibly broken you have already spent the time the leading indicators would have given you. The whole anticipatory advantage of real-time management lives in this gap — acting on what predicts the breach rather than on the breach. A desk that watches only the outcome is, by construction, always late.

Foresight is worthless without action

A reforecast that says the afternoon runs 12% heavy is only valuable if it triggers a response now: arranging cover, holding offline activity for the peak, flexing breaks, pre-warning team leaders. The earlier the projection, the smaller the intervention needed — a nudge at 11am does what a scramble at 3pm cannot, at a fraction of the cost and stress.

The characteristic failure is the desk that projects and then watches: the spreadsheet said the afternoon would be heavy, the afternoon was heavy, nobody had moved anything. Projection without action is wasted foresight — and it is surprisingly common, because projecting feels like work. The loop that matters is read the partial day, reforecast the rest, watch the leading indicators, and convert the projection into early, proportionate action.

The closing principle

See the rest of the day before it happens. The run-rate reforecast — adjusted for the cause, refreshed as the day sharpens — plus the leading indicators that move before service does, convert real-time from reactive to anticipatory. The point of foresight is early, proportionate action; the earlier the projection, the gentler the fix.