Reporting

There are two failure modes and most organisations have both at once.

The first is reporting that answers questions nobody asked. A dashboard arrives with eleven charts, somebody looks at it for a fortnight, and then it becomes wallpaper. The measures were chosen because they were easy to compute rather than because anybody makes a decision with them, and a measure nobody acts on is a measure no one needs.

The second is worse and more common: numbers no one can check. A figure appears, it disagrees with someone's spreadsheet, and there is no way to establish which is right because the calculation is somewhere nobody can see. What happens next is predictable. People stop using the number, go back to their own spreadsheet, and the organisation runs on several private versions of the truth that quietly disagree.

Both are avoidable and neither is avoided by better charts.

Only the measures a decision depends on

The dashboard covers five things, and they were chosen by asking what a small organisation actually looks at on a Monday morning. What is open and how long it has been open. What closed, and why. What is coming, weighted. What is late. Where enquiries came from, followed through to whether they converted.

That is a deliberately short list. Every additional measure competes for attention with the ones that matter, and a dashboard with eleven charts is one where nobody can tell you which two are important.

Every number opens

The second principle is that any figure can be clicked to see the records behind it. Not a summary of them, the actual rows, which you can read and count by hand if you want to.

This sounds like a convenience and it is a matter of trust. A number you can verify is a number you will argue with, and arguing with numbers is how an organisation discovers that its definition of an active customer is not what everyone assumed. A number you cannot get behind is one people quietly stop believing, and then quietly stop reading.

Not a business intelligence tool

No pivot builder, no custom visualisations, no modelling layer, no joining against external data sets. If you want to do serious analysis, export and use something built for it. A weak version of a business intelligence tool inside a CRM helps nobody and we would rather not build one.

No scheduled reports yet

You cannot have a report arrive by email on the first of the month. For an organisation that wants a number without anybody remembering to look, that is a real gap. It is named on the roadmap rather than described as coming soon.

No individual scoring

You can see work by owner, which is the honest version of the question. What the product deliberately does not do is compute a performance score or a ranking per person. A number that summarises a person is a number that gets managed toward, and the behaviour it produces is worse than the behaviour it was meant to measure.

No forecast beyond arithmetic

The weighted forecast is value multiplied by the likelihood on the stage. There is no model, no prediction from historical patterns, and no confidence interval. That is deliberate: a forecast produced by something nobody can explain is one no one argues with, and no one has therefore checked.

Pick the decision before the measure

For each number you intend to watch, write down what you would do differently if it moved. If the honest answer is nothing, stop watching it. That single exercise removes about half of what most organisations put on a dashboard.

The numbers that survive are usually unglamorous: how much is due to close, what has not moved, what is overdue, what came in and from where. All four have obvious actions attached, which is precisely why they get looked at.

Wait a quarter before believing anything

The first month of any measure in a new system is describing your import and your adoption instead of your business. Conversion looks wrong because half the pipeline arrived on one afternoon. Cycle time looks short because nothing has had time to be long.

Read it, by all means, and treat it as a check that things are being recorded rather than as information about the business.

Compare against the same period last year, not against a target

Once you have a year of history. Most businesses of this size are seasonal in ways nobody has written down, and a month on month comparison in a seasonal business generates alarm and relief in roughly equal measure, both of them meaningless.

Let people see their own numbers

Reports are scoped by the same permissions as everything else, which is what makes this possible. In most small organisations ordinary staff see no numbers at all, and the reason is not secrecy, it is that the only way to show someone some was to show them everything.

Somebody who can see their own overdue list, their own pipeline and their own enquiries will act on them without being asked. That is worth considerably more than a monthly report to management, and it costs nothing to switch on.

The period is offered as the last seven days, the last thirty, the last ninety or the last twelve months, and the page opens on thirty because that is the span someone running a small organisation is usually asking about. Not every tile obeys it, and the ones that do not are the ones people notice first, so they are marked below.

Open sales, and the weighted figure beside it

Open sales is the total value of every opportunity whose outcome is still open, with the count of them underneath. Next to it sits the weighted figure: each open opportunity multiplied by the likelihood recorded on the stage it is currently sitting in, then added up.

Neither obeys the period, and that is intentional. Work in progress is a statement about now. A pipeline total for the last thirty days would be an answer to a question nobody has ever asked out loud.

The arithmetic behind the weighted figure is worth saying plainly, because it is the entire model: value multiplied by stage likelihood, summed, and rounded once at the end rather than once per opportunity. There is nothing else in it, and nothing you cannot reproduce on paper.

Won and lost in the period

Two tiles, each carrying a value and a count, covering opportunities that closed inside the period you chose. Closed rather than opened: a deal that began in February and was won in May counts in May, where the work finished and where somebody will want the credit.

Lost is shown at the same size as won, in the same row, with the same prominence. A page that displays wins loudly and losses quietly teaches people not to close the things they lost, and a pipeline in which nothing is ever lost is the single most common way these numbers become useless.

Leads being worked

The headline figure is how many leads are open right now, meaning those marked new, working or qualified. Underneath it are two numbers that do obey the period: how many enquiries arrived in it, and how many converted in it.

Converted is counted at the moment of conversion rather than at the moment of capture. A January enquiry that becomes a customer in March lands in March, because that is where the work happened, and because counting it in January would mean last month's figures changing every time somebody finally closes something old.

Relationships

The total number of people and organisations you are allowed to see, with the number added during the period underneath. It is the least interesting tile on the page and it earns its place by being the denominator of half the questions anyone asks about the others.

Work overdue, and meetings ahead

Overdue counts tasks that have a due time in the past and that nobody has completed. The figure beneath it, labelled due today, means due within the next twenty four hours. Meetings ahead counts appointments starting in the next seven days that have not been cancelled.

All three read from the present moment rather than from the period you selected, which surprises people the first time they change the period to twelve months and watch these three tiles refuse to move. Lateness is a fact about now. There is no useful sense in which something was overdue during a period that has ended.

Where enquiries come from

One table below the tiles, four columns wide: the source, how many enquiries it produced, how many of those became customers, and the rate between the two. It is counted over everything you can see rather than over the chosen period, because a source is judged over time, and a referrer who sent you four customers last year should not disappear because you happened to be looking at a fortnight.

The sources are whatever your leads are carrying. A lead someone types in by hand carries the source manual, which means that table frequently has one large row in it describing your own staff at a keyboard. That row is not a channel and no marketing decision should be made about it.

What is absent, and why absent is not nought

Each block asks the permission that governs the records behind it. Sales asks the opportunity view permission, leads asks lead view, work asks task view, meetings asks appointment view. Somebody who holds none of them does not see a row of zeroes. They see nothing where that block would have been.

The distinction is the whole point. Nought is a claim about the business. Absence is a claim about access. Showing the second as the first tells a new starter that their organisation has no pipeline, and they will believe it, because there is no reason for a number on a screen to be lying.

What they had

Nine people distributing industrial valves. Six sell, two are in the office, one owns the place. They sell in sterling at home and in euros to two customers in the Republic of Ireland. Enquiries arrive from a trade directory, from a search advertisement, from two manufacturers who pass work on, and from people ringing up because somebody recommended them.

Before the move, the owner assembled a figure in a spreadsheet on the first Monday of each month and nobody else could reproduce it. Sources were recorded in the sense that somebody usually remembered to ask.

The first Monday

The open sales tile reads considerably higher than the owner expected, and the weighted figure beside it reads considerably lower. Both are correct. The first is the sum of everything six salespeople have not yet given up on, and the second is that same set after each deal has been multiplied by the likelihood written on its stage.

The gap between them is the useful part, and the argument it starts is the right argument: not whether the software is wrong, but whether the likelihood on the proposal stage is really what everyone has been assuming.

The euro win that was not in the headline

In the second month they win a substantial order from Cork. It does not appear in the won tile, and one of the office staff spends twenty minutes convinced something is broken.

Nothing is. The open pipeline is overwhelmingly in sterling, so sterling is the currency the headline follows, and the euro figures are held separately rather than converted at a rate no one in the building has agreed. Once that is understood it becomes a fact about the business rather than a complaint about the software: they are two operations of very different sizes and one page cannot honestly average them.

What the stage history said at the end of the quarter

Everyone in the building would have told you the slow part was the customer deciding. The history says the slow part is between the site visit and the quotation going out, and that it is slow because quotations are written by one person who also does the buying.

No one could have discovered that from the current state of the board, because the current state of the board shows where things are and not how long they have been there. It is only visible because an entry was written every time something moved, from the first day, before anyone knew they would want it.

What the sources table said, which was worse

The trade directory produced the most enquiries by a wide margin and the fewest customers. The two manufacturers who pass work on produced a handful of enquiries and converted most of them. The directory was the only source anybody was paying for.

The manual row was also the largest row in the table, which told them something else entirely: most of what they were counting as an enquiry was a salesperson typing in someone they already knew. That is not a channel and it had been quietly inflating every conversion calculation they had ever done.

What they changed

They corrected the stage likelihoods from what the quarter actually did, which moved the weighted figure down and made it believable for the first time. They stopped paying the directory. They gave the quotation writing to someone else on Tuesdays and Thursdays.

They did not build a report. Nothing in that quarter required one, which is the part of this example most likely to be true of your organisation as well.