How to connect Xero, Stripe and your other business tools to a dashboard
The four ways to feed a business dashboard, what goes wrong when data comes from several systems, how to keep access safe and the questions to ask before you build.
By Logic NI5 min read
Most dashboard problems aren't about charts. They're about data: getting the right numbers out of your accounting software, payment provider, CRM, booking system and spreadsheets, and making them agree with each other. This guide explains the ways to connect them, what goes wrong and how to keep it safe. For the wider picture, read the complete guide to business dashboards.
Where the numbers usually live
A typical Northern Ireland small or medium business has its numbers in several places:
| Area | Typical source | What you'd want from it |
|---|---|---|
| Money in and out | Accounting software such as Xero, QuickBooks or Sage | Invoices, payments, overdue amounts, cash position |
| Card and online payments | A payment provider such as Stripe, or card terminals | Takings, refunds, payouts |
| Customers and enquiries | A CRM, your website forms or a shared inbox | New enquiries, quotes sent, wins and losses |
| Bookings and jobs | A booking system or job management software | Bookings, capacity, jobs by status |
| Orders and stock | A shop platform or stock system | Orders, stock levels, returns |
| Website and marketing | Analytics, ad platforms | Visits, where enquiries come from |
| Everything else | Spreadsheets | Whatever never found a proper home |
Many of these systems offer a way for other software to read their data, often through an API (a defined way for programs to talk to each other) or through exports. What each one offers varies, so check before you plan.
The four ways to feed a dashboard
| Method | How it works | Reliability | Effort | Best for |
|---|---|---|---|---|
| Direct connection (API or native connector) | The dashboard reads from the system automatically | High, as long as the connection is maintained | Medium to high to build | Systems you rely on daily |
| Scheduled import | An export is picked up and loaded on a timetable, such as nightly | Medium to high | Low to medium | Systems with exports but no good API |
| Shared database | Your systems write to one place the dashboard reads | High | Higher, usually part of a wider build | A business building its own systems |
| Manual upload | Someone uploads a file | Low, depends on memory | Low | Last resort or one-off data |
If a number matters, avoid manual upload. Anything that relies on someone remembering will eventually be out of date.
What goes wrong when data comes from several places
| Problem | Example | What to do |
|---|---|---|
| Same thing, different names | "J. Smith Ltd", "Smith J Limited" and "Smith, J" are one customer | Decide a single identifier and match on it |
| Different definitions | One system counts a sale on order, another on payment | Agree the definition and apply it everywhere |
| Dates and time zones | A late-evening order falls on different days in two systems | Standardise on one date rule |
| VAT and currency | One system shows figures including VAT, another excluding | State clearly which the dashboard uses |
| Duplicates | The same invoice imported twice | Check for unique references on import |
| Gaps | A day missing because a connection failed | Alert when data hasn't refreshed |
| Changed fields | A supplier changes how its data is shaped, and the dashboard breaks | Have someone responsible for fixing it |
Cleaning and joining the data is usually the largest part of a dashboard project. Treat it as real work, not an afterthought.
The spreadsheet in the middle
Many businesses have a spreadsheet behind the numbers, copied from one system and tweaked before anyone sees it. That is the weak link:
| What was found | Who found it | Read it with care |
|---|---|---|
| Serious errors were found in 91% of 55 operational spreadsheets across five field audits (1995 to 2000), and about 83% of 163 spreadsheets across nine studies (1995 to 2008) | Raymond Panko, summary of spreadsheet audits | Counts errors "serious in the context of the spreadsheet's use". Thresholds varied between studies, and reviewers miss errors, so the true rate is probably higher. The audits are older, so treat this as a warning, not a current measurement |
The practical advice is to feed the dashboard from the source systems, not from a copy someone maintains by hand.
Keeping access safe
A dashboard that connects to your accounts and customer records is a sensitive thing. Good practice:
- Read-only access wherever possible. The dashboard should look, not change.
- Least privilege. Connect only to the data the measures need.
- Named, revocable access. Use credentials you can withdraw, and record who set them up.
- Role-based views. The person booking jobs doesn't need payroll.
- UK GDPR. Personal data on a dashboard is still personal data. Show names only where needed. See AI and UK GDPR for the data protection principles, which apply to dashboards too, and the ICO's guidance for organisations.
- Offboarding. Remove access promptly when someone leaves or a supplier relationship ends.
Questions to ask before you build
- Which systems feed the dashboard, and does each one offer a reliable connection?
- Who sets up and holds the access, and is it read-only?
- How often does the data refresh, and what happens if it fails?
- How are customers and other records matched across systems?
- Who fixes it when a connected system changes?
- What data leaves your systems, and where is it stored?
- How will we check the dashboard against the source numbers? Reconcile at least the first few weeks against the original reports.
Check before you trust
When the dashboard first shows a number, compare it against the source system's own report for the same period. Differences are normal at first. They tell you where a definition or a join is wrong. Do this until the numbers match and you can explain any remaining gap.
Where Logic NI fits
Joining systems is a large part of what we do. We connect your real systems, agree the definitions, test the numbers against your own reports and run it as a managed service, so the dashboard stays accurate when a connected system changes. See the integrations service and the business dashboard service, or talk to us.