Forecast Room UK

A number on its own cannot be healthy or unhealthy. Health needs three things in the same glance: where it stands, what it should be, and which way it is moving. Every row here carries all three, or says plainly which one is missing. There is no score, no traffic light and no single index, because the moment a page compresses this much judgement into one colour the first person who disagrees with the threshold discredits the whole thing. Every figure is read from the same model as the forecast tab, so the two cannot disagree, and the targets at the bottom are what turn a readout into a judgement.

Type a number and the vital above it starts being measured against it. Targets are held per region, they save with the forecast calls, and they are carried through a rebuild, so this table is where the team's definition of healthy actually lives rather than in somebody's slide. Three carry a default because they are conventions rather than opinions; a default says so on screen so nobody mistakes it for something agreed.

UK and Ireland only. Two columns, and both are arithmetic you can do by hand. Forecast is the run rate plus every large deal somebody has put their hand up for. Upside is that plus the deals we are less sure of, and it includes the forecast rather than sitting on top of it. No deal is counted at a percentage, and nothing is called best case, because a number nobody believes does more harm than good.

Closed actuals to the last complete month, then forecast and upside for every month to December. The dotted line is run rate on its own, so you can see how much of each month is being carried by named deals. Click the build-up on any month to see exactly which deals make the number.

The same forecast, split partner by partner and month by month, so each line can be checked against that partner's own pipeline. Switch between forecast and upside. Every partner carries their own run rate, measured from their own history rather than a slice of the total; the jumps are named deals.

Every open deal of 50 subscriptions or more, in close-date order, filterable by month. One decision per deal: is it in the forecast or is it upside. Nothing in between, because a percentage is a way of avoiding the question. A deal nobody has called counts in neither column. Write why in the note, because that note is what gets read out when someone asks. Where a region's book is small enough to read in one sitting the table opens on every open opportunity; the ones under the threshold are listed for completeness and carry no call, because business that size is already forecast by the run rate.

Every rate here is measured over a rolling twelve months of closed won and closed lost together, ending at the data date. Rolling rather than year to date, so the window always holds the same amount of trading and the number moves when the business moves rather than when the calendar turns over. Two very different answers depending on whether you count opportunities or subscriptions, and only one of them is about revenue. The size gradient further down is the anchor the reality check sits on, so it is worth reading properly before arguing with any number above.

The whole method, shown rather than described. Every closed-won deal under 50 subscriptions, averaged over a window of complete months. Then the same arithmetic again, partner by partner, using Account Name on the closed export. Measured, not apportioned. Nothing weighted, nothing modelled, nothing seasonal.

Counting a deal in full is a claim that it will land. This section tests the claim against what deals of that size have actually converted at over the current measurement window. It is not a second forecast and it does not change the number above. It tells you how hard the notes on those deals are working.

Every week of the year, new against existing customer, with the win rate on what was decided that week and a total closing off each month. This is the view Oli asked for, and it is also what turns every probability on this page from an assumption into a measurement as the weeks accumulate.

Every lost opportunity of 50 subscriptions or more that Salesforce holds, not just this year, with the partner, the recorded reason and the last note. Read alongside the win rate table: this is the evidence behind the number, and it is the page a board will ask for once they see the conversion rate on large deals.

Straight from the Loss Reason field, not inferred from a rep's last note, and now across the whole loss book rather than a single created-date cohort. Where a reason is missing the table says so, because a blank is a finding too.

Measured from created date to close date, wins and losses side by side, by deal size, over every deal since Created Date started being captured. These are typical figures rather than averages, because one nine-month deal drags an average and says nothing about the normal case. Typical means half take longer and half take less.

Nothing here is hard-coded. Change a value, watch the forecast move, then save so everyone is arguing about the same numbers. Where a value is measured from our own data it says so, and where it is still judgement it says that too.

The model is only as good as the fields behind it. These are the gaps that remain, in the order they hurt, with the fix beside each one.

A forecast with no scorecard is an opinion. Every time the weekly data is loaded, the standing forecast for each open month is stamped here. When the month closes the actual lands beside it and the error is calculated. Four scored quarters of this is the asset, not the model.

Every call and note on this page lives inside the page itself. That is what makes it openable by anyone with the link and no login, and it is also the weakness: when the model is rebuilt from source, the new file arrives with no calls in it. This browser keeps its own permanent copy and puts them back automatically, but that copy is only on this machine. Save the file and send it over before anyone rebuilds the model. The build now refuses to run without it, so this is the step that protects the calls rather than a nice-to-have. It is also how you move your calls to another machine, or hand them to someone else.

Two Salesforce reports, dropped straight in as .xlsx or .csv in the format they already come out in. Column headings are matched automatically and the report's own title, filter and subtotal rows are ignored. Load both together: the closed file is what measures the win rate and what strips deals from the pipeline that have already died.