Skip to content

A business, seen by the hour.

Raw transactions, hundreds of thousands of them, turned into a dashboard that tells an owner in minutes how the business is doing: which hours pay, whether the hours they staff earn their cost, where the profit really comes from, and what needs to change.

The worked example is a two-branch retail agency of a national payments network, with four and a half years of complete transaction history. The tool is general: any business whose activity arrives as transactions can track its performance, its health and the value of its working hours the same way.

The data comes in however it suits the business: from an API on a fixed schedule or in near-real time, or from files at whatever cadence they arrive. Each run rebuilds the dashboard, an independent audit proves it matches the source, an AI agent writes the read-out, and the result is available in real time from anywhere, on desktop or mobile.

Function
Operations and finance, business performance
Solution
A self-contained dashboard rebuilt on every run, with an independent audit
Intake
An API on a fixed schedule or near-real time, or files at any cadence
History
Four and a half years, complete from the business's first day
Audit
Every period reconciled against the source, worst deviation under two euro cents
Runs on
In the cloud or on your own infrastructure, on any device, desktop or mobile

The problem.

The only data the owner had was a monthly export per branch from the network's back office: one row per transaction, in a legacy encoding, with amounts in the old currency for the first years and in euro after the changeover. Merchants and branches had been renamed along the way. Everything worth knowing was in there, and none of it was visible.

Two questions mattered most: where the profit actually comes from, and whether the hours the business staffs match the demand that walks in.

The question was never whether the data existed. It was whether anyone could read it before the next month's export arrived.

What we built.

Any source, one pipeline

Transactions arrive from an API on a fixed schedule or as it happens, or from files at whatever cadence they come, as they did here. Each run regenerates the dashboard: currencies converted at fixed rates, duplicates removed, renamed merchants and branches normalized so every history stays continuous.

The core views, shared filters

Overview, rankings, trends, profit per hour and analyses, all driven by the same filters for merchant, branch, employee and date, each compared against the preceding period of equal length. What is specific to a business comes as an add-on, such as cash-flow monitoring for one that handles cash.

Profit by the hour

Normalized per working day, with a week-by-week tracker for any chosen hour, built to judge a change to opening hours on evidence.

Audited against the source

An independent script re-reads the source with separate logic and reconciles every period against the dashboard, down to the cent.

Available anywhere

Hosted in the cloud or on your own infrastructure and available in real time from any device, desktop or mobile, with a backup of the whole project on every run.

Analysis written for you

After each run an AI agent writes the read-out: what moved, why, and what to watch.

How it flows.

Each run turns the raw transactions into every deliverable, whether they arrive by API or as files, and a separate audit path re-derives each period from the source to confirm the build.

Transaction data

  • An API on a fixed schedule, or continuously
  • Or files at any cadence, as in this case
  • Any encoding, any currency

Build pipeline

  • Converts currencies at fixed rates
  • Removes duplicate rows
  • Normalizes renamed merchants and branches

One run, every deliverable

Aggregated data

  • Hundreds of thousands of transactions
  • Cleaned and reconciled
  • Ready to filter by merchant, branch, employee and date

Live dashboard

  • The core views, plus add-ons for what is specific to the business

Any device

  • Desktop and mobile, in real time

Backups

  • The whole project, on every run

An independent audit re-reads the source with separate logic and reconciles every period against the build.

What the owner sees.

Operations

August loaded, two branches

Operations
10,233
Turnover
€416k
Profit
€6,385
Profit per staffed hour
€13.70
  • Payouts and transfersUp
    Share of transactions
    13%
    Share of profit
    42%
  • Account top-upsDown
    Share of transactions
    45%
    Share of profit
    23%
  • Bill paymentsSteady
    Share of transactions
    27%
    Share of profit
    22%
  • Mobile and internetSteady
    Share of transactions
    12%
    Share of profit
    10%
  • OtherSteady
    Share of transactions
    3%
    Share of profit
    3%
Illustration with sample data.

What the data showed.

The findings that changed how the business is run, redrawn from the dashboard.

Payouts and transfers

Two services against the rest of the mix

  • Share of transactions13%
  • Share of profit42%
Payout and transfer services carry 42% of the profit from 13% of the transactions, about five times the per-transaction value of the rest of the mix.

Activity by weekday and hour

Transactions, full history

fewermore transactions per hour
The working schedule is visible in the data itself: both branches on weekdays, one branch also on Saturdays with a midday break, Sundays closed. The faint Saturday opening cell is the half-hour that turned out not to be worth staffing.

Profit per hour

Average per working day, both branches

A strong morning peak, a lunch-time dip, and a long profitable tail into the evening at the branch that stays open late. This is the profile the staffing questions were answered against.

Seasonality

Each month's profit as a multiple of its year's monthly average, three full years

January is the softest month at three quarters of the year's monthly average, and the year builds steadily to a December peak of almost a third above it. The overlay separates growth from seasonality.

Profit by year

Indexed to the first, partial year

Profit grew almost sevenfold from the partial first year to the third full year. Asterisks mark partial years: ten months of operation in the first, eight months of data in the last.

What changed.

The real profit mix

Payout and transfer services turned out to generate 42% of the profit from 13% of the transactions. Promotion was redirected accordingly, and one branch's underuse of these services became a concrete growth target.

Staffing judged on evidence

Hour-level profit showed the schedules already match demand almost perfectly. The one clear misfit was a nearly dead half-hour in one branch's weekend schedule, and a proposed evening extension at the other branch was re-costed against its own late-day demand, with the tracker to judge it already in place.

Growth told apart from seasonality

Profit grew almost sevenfold from the partial first year to the third full year, and the year-over-year overlay now separates genuine growth from a seasonality that peaks in December and dips every January.

Numbers that can be trusted

Every rebuild is reconciled with the source data by an independent script, whether that data arrived by API or as files. It caught a rounding defect that had silently understated profit, and the deviation across the full history is now under thirty cents.

Available anywhere, in real time

The dashboard is hosted in the cloud or wherever the business prefers, and it is available from any device, desktop or mobile, with nothing for anyone to maintain by hand and a backup on every run.

Where else this applies.

Any business whose transactions can be pulled from an API or exported as files can read itself this way: retail chains, franchise networks, clinics, logistics depots, agencies with several branches. The questions are the same everywhere: how the business is performing, whether the hours and people it pays for earn their cost, and what to change. The data already holds the answers; the work is cleaning, reconciling and presenting it, and writing the analysis so nobody has to.

What it needs is small: access to the transactions, by API or as files, a written version of the rules (names, currency, opening hours), the questions the owner actually asks, and one person to read the summary.

More about our Data & decision intelligence service

More work.

Tell us what you're working on.

A 30-minute conversation about your business, and a clear view of what we would do first. No obligation.

Book an intro callhello@devshiftlabs.com