Resources
Build a cash-flow dashboard with Australian bank data
Build a weekly cash-flow dashboard from consented Australian bank data. Run a two-account example to see how transfers, pending payments and data freshness affect the totals.
A business moves A$600 from its operating account to its reserve account. The operating account shows money out and the reserve account shows money in. Across the two accounts, the business has exactly as much cash as it did before the transfer.
To build a cash-flow dashboard from Australian bank data, choose the accounts and period first. Add up posted credits and debits that cross that boundary. Show confirmed transfers within the boundary in the transaction list and exclude both legs from the headline totals. In the example below, that gives A$2,410 in, A$1,500 out and A$910 net.
Start with the accounts the customer shared
This tutorial starts after the customer has connected their bank. If you still need the connection flow, follow How to connect banking data to your app with Fiskil. The customer selects which accounts to share and approves access for the requested data. On your backend, resolve the signed-in customer to their Fiskil End User and check that the selected accounts belong to that connection. When sharing ends, stop fetching under that consent. Fiskil consent guide.
For this view, select two AUD deposit accounts for one business and a reporting week in Australia/Sydney. Keep other currencies in separate views. Use the beginning of the local week and the beginning of the next week as your boundaries. Convert timestamps to Sydney time before assigning a transaction to a day.
Fetch each selected account’s transactions on the backend with GET /v1/banking/transactions, using end_user_id and account_id. Follow links.next through the last page. If pages overlap or an import retries, update the existing record by its fiskil_id within the same customer and account. The date filter uses posting time first, then execution time when posting time is absent. For a posted-week view, flag posted rows missing a posting date for review. Keep pending activity visible separately. Get Transactions · Banking data and transaction dates · Signed amounts and record identity.
Follow one A$600 transfer through the dashboard
Sandbox dataset. These seven fictional records represent 2–8 November 2026. The accounts, descriptions and transfer match were made for this tutorial. No customer records or Fiskil sandbox credentials are needed.
| Day | Account | Movement | AUD | Counted in the two-account view |
|---|---|---|---|---|
| 2 Nov | Operating | Client receipt | +2,400 | Inflow |
| 3 Nov | Operating | Rent | −900 | Outflow |
| 4 Nov | Operating | Transfer to Reserve | −600 | Internal transfer |
| 4 Nov | Reserve | Transfer from Operating | +600 | Internal transfer |
| 4 Nov | Operating | Supplier invoice | −600 | Outflow |
| 6 Nov | Reserve | Interest received | +10 | Inflow |
| 8 Nov | Operating | Pending card authorisation | −45 | Pending |
The supplier payment and the transfer leave the same account on the same day for the same amount. Dropping every A$600 debit would hide a real bill. In this dataset the business has confirmed which two records belong together. In your app, retain the two source IDs and the basis for each match. Use an equal amount and nearby dates to suggest a match for review. If you have only a possible match, keep the rows visible and mark the affected gross totals provisional.
Before removing the confirmed pair, the posted records add up to A$3,010 in and A$2,100 out. Both totals are A$600 too high for the combined view, yet their difference is still A$910. Checking the net alone would miss the error.
Run the dashboard and change the account selection
Save the following as cash-flow.html and open it in a browser. The example takes a single week of already-normalised rows as input. Transfer links belong to this app’s fixture. The arithmetic uses integer cents and formats dollars only for display.
<!doctype html>
<html lang="en-AU">
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Weekly cash flow · Sandbox</title>
<style>
body { max-width: 48rem; margin: 2rem auto; padding: 0 1rem; font: 1rem/1.5 system-ui; }
table { border-collapse: collapse; width: 100%; }
th, td { text-align: left; padding: .45rem; border-bottom: 1px solid #ddd; }
td:nth-child(2), th:nth-child(2) { text-align: right; }
</style>
<h1>Weekly cash flow</h1>
<p>Sandbox example · AUD · 2–8 November 2026 · Sydney time</p>
<label>Accounts
<select id="scope">
<option value="both">Operating and Reserve</option>
<option>Operating</option><option>Reserve</option>
</select>
</label>
<p id="totals" aria-live="polite"></p>
<table>
<thead><tr><th>Movement</th><th>AUD</th><th>Treatment</th></tr></thead>
<tbody id="details"></tbody>
</table>
<script>
// One business, one week. Signed amounts are integer AUD cents.
// ID, account, description, cents, status
const rows = [
['r1', 'Operating', '2 Nov · Client receipt', 240000, 'POSTED'],
['r2', 'Operating', '3 Nov · Rent', -90000, 'POSTED'],
['r3', 'Operating', '4 Nov · Move to Reserve', -60000, 'POSTED'],
['r4', 'Reserve', '4 Nov · From Operating', 60000, 'POSTED'],
['r5', 'Operating', '4 Nov · Supplier invoice', -60000, 'POSTED'],
['r6', 'Reserve', '6 Nov · Interest', 1000, 'POSTED'],
['r7', 'Operating', '8 Nov · Card authorisation', -4500, 'PENDING']
];
// A reviewed pair in this fixture. The supplier payment is unrelated.
const counterpart = new Map([['r3', 'r4'], ['r4', 'r3']]);
const money = new Intl.NumberFormat('en-AU', {
style: 'currency', currency: 'AUD'
});
const format = cents => money.format(cents / 100);
const scope = document.querySelector('#scope');
const details = document.querySelector('#details');
function draw() {
const selected = rows.filter(([, account]) =>
scope.value === 'both' || account === scope.value);
const postedIds = new Set(selected.filter(row => row[4] === 'POSTED')
.map(row => row[0]));
let inflow = 0;
let outflow = 0;
details.replaceChildren();
for (const [id, account, label, cents, status] of selected) {
let treatment;
if (status !== 'POSTED') {
treatment = 'Pending';
} else if (postedIds.has(counterpart.get(id))) {
treatment = 'Internal transfer';
} else if (cents > 0) {
inflow += cents;
treatment = 'Inflow';
} else {
outflow -= cents;
treatment = 'Outflow';
}
const tr = details.insertRow();
for (const value of [account + ' · ' + label, format(cents), treatment]) {
tr.insertCell().textContent = value;
}
}
document.querySelector('#totals').textContent =
'In ' + format(inflow) + ' · Out ' + format(outflow) +
' · Net ' + format(inflow - outflow);
}
scope.addEventListener('change', draw);
draw();
</script>
</html>
Switch from both accounts to Operating. The A$600 now leaves the selected account, so its outflow belongs in that view. Select Reserve and the same transfer is an inflow. “Inflow” describes movement into the selected accounts. A transfer into Reserve does not become sales revenue.
| Selected accounts | Money in | Money out | Net movement |
|---|---|---|---|
| Operating and Reserve | A$2,410 | A$1,500 | +A$910 |
| Operating only | A$2,400 | A$2,100 | +A$300 |
| Reserve only | A$610 | A$0 | +A$610 |
The two individual net movements add back to A$910. For a second check, give the fixture a combined opening balance of A$5,000 and a closing balance of A$5,910. The difference should match the posted net movement. When checking real data, compare balances for the same accounts, currency and cut-off. Separately fetched balances and transactions can reflect different update times. Fiskil’s balance calculation guide.
Show how current the totals are
Check GET /v1/data-freshness with the customer’s end_user_id, the banking.transactions dataset and each selected account_id. Put the reported transaction freshness beside the account name. If one account was updated at 09:00 and the other at 11:00, a single “Updated 11:00” label would overstate the coverage. If freshness is missing, say “Update time unavailable”. A refresh request queues work asynchronously. Keep the last complete calculation labelled with its original times until the new transaction pages are ready. Get Data Freshness · Refresh Banking Transactions.
Three cases are worth testing when you replace the fixture with a feed:
- A pending purchase posts. Recalculate from posted rows. The pending authorisation should never contribute a second debit.
- A transfer spans a week boundary. If the debit posts on Sunday and the credit on Monday, show the confirmed amount in transit separately and explain the difference from the change in balances. The demo assumes both legs settle in the same week.
- One account has only partly loaded. Mark the combined view incomplete. An unavailable account and an account with zero movement need different labels.
This dashboard reports observed cash movements. Its A$910 result says how much net cash entered the selected accounts that week. It tells the reader neither how much they can safely spend nor which financial product to choose. ASIC’s Moneysmart explains the distinction between factual information and financial advice. General and personal financial advice.
If the next question is “What will our balance be next Friday?”, start with the reconciled actuals and add expected movements separately. Cash-flow forecasts from bank data.


