Services / Data / Live dashboards

The numbers you run on, live, on every screen you own

Sales, cash, stock, delivery, utilisation, pulled straight from the systems that already hold them, laid out so the answer takes one glance, and refreshed without anyone exporting anything.

Start the conversation →See the technology

Why it matters

The cost is the wait

Almost every business already owns the numbers it needs. They are simply spread across a till system, an accounting package, a CRM, three spreadsheets and somebody's inbox, so answering "how did last week actually go?" costs a person half a morning of exporting, pasting and reconciling. By the time the answer exists it describes a week that is already over, and half the meeting is spent debating whether the figures are right rather than deciding what to do about them.

A dashboard that works is not a wall of charts. It is a short list of decisions somebody makes every week, each one answered by a number that is current, defined once, and trusted enough to act on without a follow-up email. That means the boring parts matter more than the visuals: where each figure comes from, what it counts and excludes, who is allowed to see it, and what happens on the morning the source system fails to deliver.

We build the whole path, connections into your existing systems, a tidy place to keep the data, one agreed definition per metric, the access rules, the screens themselves, and the training that decides whether anyone still opens them in six months. You own the lot: the queries, the definitions, the data and the dashboards.

What you actually get

Built to be trusted

A dashboard project succeeds or fails on the details that never make it into the demo. These are the ones.

Built around decisions, not chart types

We start from the calls people actually make (which site is slipping, which project is eating its margin, whether we can cover payroll) and build back from there. A chart that nobody would act on gets cut, however good it looks.

One definition per metric

"Revenue" is defined once, written down in plain English with a named owner, and reused by every report that mentions it. It ends the meeting where two departments present different numbers for the same thing and spend an hour arguing about whose is right.

Fast enough that people keep using it

Figures are pre-calculated so a page opens in a second or two rather than grinding through millions of rows on every click. A dashboard that takes twenty seconds to load is a dashboard people stop opening, and no amount of design fixes that.

Everyone sees their own slice

Permissions applied to the data itself, so a regional manager opening the report sees only their region: automatically, from one report, not a separate copy per team that quietly drifts out of step with the others.

Right on the wall, the laptop and the phone

The same numbers laid out for a boardroom TV, a desk and a pocket, plus scheduled email delivery for the people who will never open a dashboard at all. Distribution is usually what decides whether a reporting project changes any decisions.

It tells you when it is stale

Every panel carries a last-refreshed stamp, and automatic checks look at whether the data arrived, arrived complete and sits in a sensible range. When a source fails the dashboard says so, rather than confidently showing you yesterday.

Where it earns its keep

Same pattern, different desks

The shape is the same wherever we find it, a leadership team making this week's decisions on last week's numbers, one export away from being wrong.

01 · Multi-site retail & hospitality

Twelve sites, twelve versions of yesterday

The problem: Each site manager keeps their own spreadsheet, head office stitches them into a weekly summary on Tuesday for a week that ended on Sunday, and a site that is quietly losing money gets noticed a fortnight late, if the person assembling the summary happens to spot it.
What we build: Takings, labour as a share of revenue, stock and wastage pulled nightly from the tills and the rota system into one comparison view. Managers see their own site and how it ranks; the operations director sees all of them side by side, sorted worst to best.
What changes: A bad week is visible on the day it happens instead of a fortnight later, the Tuesday assembly job disappears entirely, and site conversations start from the same figures on both sides of the table.

02 · Manufacturing & logistics

A whiteboard that only works if you are standing next to it

The problem: Output against plan, downtime and on-time delivery live on a whiteboard by the line and half-buried in the ERP. Anyone off-site (a director, an account manager fielding a "where is my order" call) has to ring someone on the floor and interrupt them to find out.
What we build: Wallboards on the shop floor showing output against plan, current downtime and the reason for it, and which orders are at risk of missing their promised date, with the same underlying numbers available as a phone view for the office and a weekly trend for the board.
What changes: The floor and the office work from one picture, chasing calls stop interrupting production, and the recurring causes of downtime become obvious because they are now counted rather than remembered.

03 · Professional services & agencies

Utilisation you only discover at month end

The problem: Timesheets are in one tool, invoices in the accounting package and the pipeline in the CRM. The three are reconciled at month end, by which point a fixed-fee project can already be well past its budget and there is nothing left to do but absorb it.
What we build: Time, invoices and pipeline joined into a live view: utilisation per person against target, spend against budget on every live project, and forecast fee income for the coming quarter based on work already won.
What changes: A project heading the wrong way is caught in week two rather than at invoicing, and resourcing arguments are settled with a number instead of a feeling about who is busy.

The technology

The tools behind it, named

Which tool is right depends on what you already pay for, how many people need access, and whether the dashboard is for your own team or something you sell to your customers. We pick per project and tell you why.

6 layers · 30 technologies

The dashboard layer

The part your team actually opens. All of these do charts well; they differ enormously on licensing, on who can build reports, and on how they handle permissions.

Power BI
Metabase
Looker
Apache Superset
Grafana
Tableau

Where the numbers are kept

Dashboards need somewhere to read from that is fast and separate from the systems running your business, so a heavy report never slows down the till or the CRM.

Google BigQuery
Snowflake
PostgreSQL
ClickHouse
DuckDB

Shaping the numbers

Between the raw export and the chart sits the modelling layer: where a metric gets its one definition, its tests, and its documented lineage back to the source.

dbt
SQL
Python
pandas
Apache Airflow

Analytics inside your own product

When the dashboard is a feature you sell, it gets built into your application with your design, your permissions and your pricing tiers: no third-party branding, no per-viewer licence as you grow.

React
Next.js
TypeScript
Apache ECharts
Plotly

Who is allowed to see what

Sign-in through the accounts your staff already have, and permissions applied to the data itself rather than to the chart, so a filter can never be removed to reveal someone else's region.

Microsoft Entra ID
Okta
Auth0
Keycloak
Row-level security

Where it runs

Hosted close to your people so pages open quickly, or inside your own infrastructure where the data is not allowed to leave the building.

Cloudflare
Google Cloud
Docker
Vercel

Product names and logos are the property of their respective owners and are shown to describe the technologies we work with. Their use does not imply any partnership, sponsorship or endorsement.

How we deliver it

Live behind a human first

Usually four to eight weeks to a dashboard people are running the business on, depending on how many systems have to be connected first.

We sit in your weekly meeting

Before designing anything we watch how decisions actually get made, which figures get quoted, which get argued about, and which spreadsheet somebody rebuilds every week by hand. That list is the specification.

We agree what each number means

Is revenue booked at order or at invoice? Does it include VAT, shipping, refunds, intercompany? We write the answers down with you and get them signed off, because this is where reporting projects quietly fail.

We connect the sources

Reliable feeds from the systems that hold the data, pulling only what has changed and coping gracefully when one of them renames a field. Anything too messy to trust yet gets flagged to you rather than silently averaged.

We build the first screen and show you early

One dashboard, the most valuable one, in front of the people who will use it within the first couple of weeks. Real reactions to a real screen change the design far more usefully than another requirements workshop.

We check it against something you already trust

Every headline figure is reconciled against your existing month-end numbers or a report you already believe. Where they disagree we explain exactly why, often because the old number was wrong, and you should know that before the board does.

We train your team and hand over the keys

Hands-on sessions with the people who will use it, plain-language documentation of every metric, and enough training for your own staff to build their own views. Then a review after the first month, when the real requests appear.

Before you commit

The questions worth asking

Can a dashboard fix our messy data?

No, and anyone who says otherwise is selling you something. A dashboard is a window; if what is behind it is inconsistent, the dashboard makes that inconsistency visible in high resolution. That is genuinely useful, most clients discover real problems in the first fortnight but the fix is upstream, in how the data is captured and maintained. Where a source is too unreliable to report on, we will tell you that and propose cleaning it up first rather than quietly averaging over the gaps.

How current is "live"?

Usually hourly or nightly, and for most decisions that is plenty. Genuinely real-time data (seconds, not hours) costs meaningfully more to build and run, so we recommend it only where the decision truly cannot wait: live stock during a sale, fraud, a production line stopping. We will happily talk you out of it where the value is not there.

We already have Power BI. Do we need something else?

Probably not. If you are on Microsoft 365 the licensing and sign-in are already in place, and the sensible move is to make what you have work properly rather than introduce another tool. Most of what we do in that situation is the layer underneath (the connections, the definitions, the modelling) which is where the actual problem usually is.

What happens when a source system goes down?

The load retries, catches up when the system comes back, and raises an alert if it cannot. Meanwhile the dashboard marks the affected panels as stale rather than showing you yesterday's figure as though it were today's. We would rather a director sees "this data is 26 hours old" than makes a decision on a number that silently stopped moving.

Who maintains it once you have gone?

You can, and we build it so that is realistic, everything documented, everything in your accounts, your team trained to add views and change metrics. Plenty of clients keep us on a small retainer for the changes that come with new systems or new reporting needs, but that is a choice, not a dependency we engineer in.

How much of our own time will this take?

More than most suppliers admit, and it is concentrated at the start. Expect a few hours a week from someone who genuinely knows how the business counts things, usually a finance or operations lead, for the first month, mostly spent settling metric definitions and access to systems. Projects where nobody was available for that are the ones that overrun.

Tell us the numbers you wish you could see every morning

Send us the report someone rebuilds by hand every week. We will tell you honestly what it would take to make it build itself, and what has to be tidied up first.

Start the conversation →

Explore more

The rest of Data & Dashboards

Connecting systems Automated reporting Forecasting & alerts ← All of Data