Reporting & dashboards
A short, honest weekly read.
Most businesses have too much data and not enough reporting. Numbers sit in six systems, the weekly review never quite happens, the monthly board pack takes two days to build, and different departments disagree on the same KPI. We design a small, agreed set of numbers — the ones that actually change decisions — and make them appear on the right desk at the right time, without anyone having to build them.
How it shows up
Signs the work is needed.
- 01 Every department has a different version of the same KPI, because everyone pulls the data themselves.
- 02 A monthly board pack that takes two days to build and is out of date by the second page.
- 03 Dashboards you paid for that nobody looks at, because they show everything and therefore signal nothing.
- 04 Decisions made on gut feel because the data was going to take an afternoon to assemble.
- 05 A sense that the numbers are probably fine — but without enough evidence to be sure.
How we design it
The shape of the setup.
-
a.
We agree the numbers first.
Before any dashboard is built, we agree the six to ten numbers that actually matter to the people running the business. Named, defined, calculated-how, sourced-where. This takes a single afternoon and saves months of future argument. Most engagements discover that three of the proposed metrics are really the same thing with different names.
-
b.
We pipe the sources into one place.
Your CRM, your accounting, your project management, your marketing — wherever the numbers live — into a single warehouse. For small teams this is usually a neatly-structured Google Sheet fed nightly; for larger ones it is a proper BigQuery or Postgres setup. The principle is the same: one place, one refresh schedule, one agreed definition per field.
-
c.
We write the weekly read.
Every Monday at 7am, a short email arrives in the owner’s inbox: the current week's numbers, the change from last week, a one-line comment on anything worth noticing. Written by a light model pass that flags movement, reviewed-but-not-rewritten by the system. Two minutes of reading. No login required.
-
d.
We design two views, not twenty.
An owner's view (the weekly read) and an operator's view (a live dashboard for day-to-day team decisions). That is almost always enough. We refuse to build the fifteen "could be useful" tabs that nobody will look at. If a new view is needed later, we add it then.
What “done” looks like
Outcomes we hold ourselves to.
These are the benchmarks we aim for on a typical engagement. Your specifics will vary — the diagnosis document names the ones that apply to your business, in writing, before work starts.
- An agreed KPI dictionary
- Every important number has a single definition, in writing, that the whole team accepts. Arguments about "which revenue we mean" end.
- Weekly summary in under two minutes
- The owner reads three to five numbers on a Monday morning, already contextualised. No dashboard login required.
- Monthly board pack in under 30 minutes
- The assembly becomes a review, not a compilation.
- One source of truth per number
- If the sales team quotes a number, the finance team quotes the same one. Conversations stop being about the numbers and start being about the business.
A representative engagement
What an average client looks like.
The context
An eight-person SaaS consultancy with clients across the UK and Ireland. Numbers lived in HubSpot, Xero, Harvest, and a half-maintained Google Sheet. The monthly management pack took the operations lead most of two Fridays. Leadership meetings spent the first twenty minutes arguing about definitions.
What we built
- A single source-of-truth Google Sheet, refreshed nightly via n8n pulling from HubSpot, Xero, and Harvest APIs, with each field traceable to its source.
- A written KPI dictionary — six numbers, two paragraphs each — committed to, signed, and pinned in the company wiki.
- A Monday 07:00 email to the managing director: MRR, pipeline coverage, utilisation, DSO, cash on hand, one-liner on anything that moved more than ±10%.
- A single Looker Studio operator dashboard, with five tabs maximum, shown on a wall TV in the office.
Six months on
Management pack time down from two days to forty-five minutes. KPI arguments in leadership meetings largely stopped. The managing director stopped asking the ops lead for numbers mid-week because he already had them on Monday.
This is a composite drawn from the pattern we see most commonly in this area of work. The shape holds; the specifics change.
When this isn’t the right fit
Times we say no.
- You are too early to know what to measure. If product-market fit is still being found, over-instrumenting is a distraction; instrument only enough to learn.
- Your team is under five people and everyone knows the numbers by feel. The overhead of a formalised system is not earned yet.
- Your reporting problem is primarily a data-quality problem. If the CRM is half-full of nonsense, a dashboard will just make the nonsense more visible. Fix the source first.
- You need deep statistical analysis, experimentation, or econometrics. That is a different discipline; we are about operational reporting, not data science.
If this describes the shape of your week, write to us.
Book a discovery call