Reports, and building your own

Last updated September 21, 2026

Reports → Report Library lists everything the club has, grouped by subject. Pick one on the left and it runs; the answer appears beside it.

Your club starts with twenty, written to be run rather than admired: who is currently a member, how many of each type, who has left, members with no email address, how busy each rink is, what has been taken by each payment method, which duties are unfilled, what maintenance has cost. They are yours from the moment they appear — rename them, change them, copy them, delete them.

📤 Getting the figures out

Every report exports as CSV, Excel or PDF, taken from the rows on the screen rather than run again. A report exported an hour after it was read, with figures that have moved in between, is how two people end up arguing about which of them has the right total.

The CSV holds plain values rather than tidied ones, because a CSV gets opened in a spreadsheet and added up — “£120.00” is text and will not total, and “10 Aug 2026” will not sort as a date. The screen and the PDF are the other way round, because those are read rather than calculated with.

🛠 Building one

Reports → Build a Report. Choose the subject, then drag fields across — or press the + beside one, which is the same thing and works on a tablet. Drop a field on Grouped by, Only rows where or Sorted by to use it there instead.

Try it runs the first fifty rows without saving anything. Save keeps your changes; Save as makes a copy, which is how most reports actually get written — open one that is nearly right, change it, keep both.

🔗 Fields from a linked record

Some fields belong to another record and are marked as such. A booking has one rink and one booker, so the rink name and the booker’s email are columns on a booking report.

It only works one way round, on purpose. A member has many bookings, so “this member’s bookings” is a report of its own rather than a column — following that kind of link either multiplies the rows or silently picks one, and a report whose row count changed when somebody added a column is a report nobody can trust.

🧮 What grouping does

Group by membership type and a row stops being one member and becomes a bucket. Every other column then has to be a summary — how many, total, average, earliest, latest — because there is no true answer to “which telephone number” for a bucket of forty people. Columns that are neither the grouping nor a summary are left out rather than guessed at.

📊 Charts

A chart wants one label and one number, which in practice means a grouped report. Thirteen kinds are offered, from a picker that draws a working example of each rather than listing names:

Comparing sizes — Bars, Bars across (for long names), Lollipops (kinder on a long list).
Following a run — Line, Filled line. Only sensible where the labels are in order.
Splitting a whole — Pie, Doughnut (the total in the middle), Share bar (one bar split into its parts, which keeps working past the six slices where a pie gives up) and Blocks (boxes sized by their figure, for many parts of very different sizes).
Getting to a total — Running total. Each bar starts where the last one ended and a closing bar shows what is left, so income up and costs down read as the arithmetic they are.
A shape to compare — Web. Bookings by day of the week drawn the same way every season, so this year laid beside last says “Wednesdays have gone” without anybody reading a number. Needs at least three groups.
Stages falling away — Funnel. Enquiries, taster session, applied, approved, paid up: where the drop is, as a shape rather than as arithmetic. Needs at least two rows. Set Percentages are to say what the figures beside it are shares of.
One number, judged — Dial. The total against a target you set. The only chart here that answers “are we all right?”.

⚙ More chart settings

Under More chart settings in the builder, beneath the two column boxes. Only the settings that mean something for the chart you have chosen are shown — a gridline on a pie is not a preference, so it is not offered.

Order — as the report has it, biggest first, smallest first, or by label A–Z. A run over time wants the report’s own order; anything being compared wants the biggest first.
How many, 3 to 20, and add the rest together — everything past the limit shown as one “others” rather than quietly dropped. The tail is always taken by size, so ordering by label cannot push the club’s biggest figure off the chart because its name begins with W.
Colours — one colour (the club’s teal), a colour for each, or light-to-dark by size. Where every bar is the same kind of thing, one colour is the honest choice: a different colour per bar says a difference that is not there.
The figures on the chart — where they fit, always, or never.
Figures read as — a plain number, money (with your own currency symbol) or a percentage.
Gridlines — faint lines behind the bars so a height can be read off rather than guessed. Turned off, the gutter that held the figures goes with them and the chart uses the full width.
Target — a dashed line across the chart at that figure, labelled. The Dial needs one; everything else is better for having one when there is a budget to measure against.
Start the scale at zero — on for bars always, because a bar cut off at the bottom lies about its size. Offered for lines, where a membership count that only ever moves between 118 and 124 is a flat line from zero and the movement is the entire point.
Percentages are (funnel only) — a share of them all added up, a share of the first row, or what was kept from the row above. A line at the top of the chart always says which is in force.
Funnel shape (funnel only) — a funnel, with the widths eased, or widths exactly in proportion to the figures. A funnel narrows all the way down, so eased draws two equal figures a little differently; the real figures are printed against every row either way.

The chart is drawn on the canvas as you build it, above the table, from the same code that draws the finished one — so a setting can be watched working rather than saved and hunted for afterwards.

📌 On the dashboard

Once a report has a chart, Put on the dashboard pins it to the admin dashboard, and clicking the tile opens the report behind it.

Charts sit in panels of three — three across a row is where a chart is still readable, and past that each is a thumbnail. Press the ＋ on a panel header to add another panel and choose the charts for it with its ⚙; six panels, eighteen charts, is the most a dashboard will take. Each panel moves and hides on its own like any other dashboard section. Put on the dashboard drops a chart into the first panel with room and starts a new one when they are all full.

Dashboard charts are admin only. They are club-wide figures, and they have no place on the members’ app or a public page where somebody would be shown the club’s finances and their neighbours’ habits.

Common questions

Can I delete one of the reports we started with?: Yes, and it will not come back on the next upgrade. A report that reappeared after somebody deliberately removed it would be a bug rather than a feature.
A report says nothing matched.: That may well be the right answer — but check the conditions before reporting it as one. An empty report looks exactly like a fact.
It says the first 5,000 rows.: A report longer than that is really an export. Narrow it with a condition, or take the CSV, which has the lot.
Who can build and save reports?: Anybody who may see reports can run and export them. Saving, changing and pinning to the dashboard needs the settings permission — a report is a thing the whole committee then sees.
Why can I not report on two subjects at once?: You can bring in fields from a linked record, which covers most of what people mean by that. Joining two lists properly changes what a row means, and that is where reports quietly start lying.
Where are the chart settings? I only see the two column boxes.: Under More chart settings, folded away beneath them — most charts want none of it, and eight boxes on the way to “bars, by rink” is a builder that feels like work. Open it and you get order, how many, colours, whether the figures are printed, the number format, gridlines, a target and the zero baseline. Only the ones that mean something for the chart you have chosen appear.
Which chart should I use?: Comparing sizes: Bars, or Bars across when the labels are long names. Shares of a whole: a Pie up to about six parts, a Share bar or Blocks beyond that. Over time: a Line. Income against costs: Running total. Against a budget: a Dial with a target. Stages: a Funnel. The picker draws a working example of each, so the quickest answer is to look.
Every bar is a different colour and it looks busy.: Set Colours to One colour — the club’s teal. Where every bar is the same kind of thing, a colour per bar is decoration claiming to be information. Use Light to dark, by size if you want the colour to say something, since then it says the same thing as the length.
Can I show a budget or target on a chart?: Yes. Type it into Target and it is drawn as a labelled dashed line across bars, lollipops, lines and the running total — and the chart’s scale stretches to reach it, so a target above every bar is still on the picture. The Dial is built around one: it shows the report’s total against the target as a percentage.
My chart cut off the smaller groups.: Raise How many — up to twenty — or leave Add the rest together ticked, which is the default: everything past the limit is added into a single “others” so the tail is accounted for rather than dropped. Untick it only if you genuinely want the top few and nothing else.
My funnel says the top row is 100%.: Change Percentages are under More chart settings to a share of them all added up. Both answers are right for different data and the software cannot tell which you have: on genuine stages — enquiries, applied, paid up — the same people are counted at each step, there is no meaningful total, and everything is measured against the first row, which is therefore 100% by definition. On a breakdown of different things, such as membership types, the total is real and the largest is 55% of it rather than 100% of itself. Since 10.50.3 the share of the total is the default, and the small grey line above the chart always names which you are looking at.
Should membership types be a funnel at all?: Probably not. A funnel draws what falls away between stages of one process. Membership types are not stages — nobody starts as Full and drops to Honorary — they are parts of a whole, which is what a Share bar, a Pie or plain Bars are for. The funnel will draw it, and with the percentages set to a share of the total it will draw it honestly, but a share bar answers the question better.
The funnel edges look jagged.: Fixed in 10.50.3. Each stage used to be drawn as its own straight-sided shape, so the outline arrived at every boundary at one angle and left at another and the whole thing read as a sawtooth — worst where one row is much larger than the next. The outline is now a single curve through all the widths, with each stage a slice of it.
My funnel looks like a pair of wings rather than a funnel.: Fixed in 10.50.4, and it happens when one row dwarfs the next — eleven members falling to three puts the top row at four times the width of everything below it. Two things changed. The widths are now eased rather than held exactly to the figures, so the funnel narrows all the way down instead of collapsing in one step; and the outline leaves the top corners square before it starts curving in, so a wide row reads as the mouth of a hopper rather than as two thin triangles reaching for the corners. Set Funnel shape to widths exactly in proportion if you would rather have the old behaviour.
The web chart says it needs three groups.: A web draws a shape across a set of headings, and two headings are a line rather than a shape. Either group by something with more values in it — days of the week, membership types — or use bars, which compare two figures better than a web ever would.
Are the chart colours readable for a colourblind member?: Yes, and it is checked rather than assumed. The colours are handed out in a fixed order, and every neighbouring pair in that order is tested to be clearly distinguishable under red-green colourblindness — which affects roughly one man in twelve. The order was changed in 10.49.0 because the old one put a green beside a red.