How to automate invoice data entry into Xero (3 ways, honestly compared)
Hubdoc, Dext, or a custom pipeline? How to stop typing invoices into Xero, what each option costs, and when custom automation pays for itself.
If someone in your business spends hours each week typing supplier invoices into Xero — vendor, date, line items, GST, account code — you’re paying a person to be a scanner. In 2026 this is one of the most solved problems in small-business admin, and for a lot of businesses the solution is included in the Xero subscription you already pay for.
We build custom invoice automation, so it would suit us to tell you the built-in tools are rubbish. They’re not. Here are all three options, when each one wins, and the specific situations where custom is genuinely worth it.
What invoice automation actually does
Every option follows the same pipeline:
- Capture — invoices arrive (email, photo, folder, supplier portal) and land in one place.
- Extract — OCR reads vendor, ABN, date, invoice number, totals and GST off the document.
- Code — the system picks the Xero contact and account code, usually by learning from history (“Telstra → Telephone & Internet”).
- Push — a draft bill appears in Xero with the source document attached, waiting for a human to approve.
That last word matters: good automation drafts, humans approve. You’re removing typing, not oversight.
Option 1: Hubdoc — free with your Xero plan
Hubdoc is Xero-owned and included free with most Australian Xero subscriptions, which makes it the obvious starting point.
How it works: each supplier emails invoices to your unique Hubdoc address (or you snap photos in the app). Hubdoc extracts the data, learns your coding preferences per supplier, and publishes draft bills to Xero with the PDF attached.
Good: the price; supplier-level rules (set Telstra’s coding once, every future bill follows); documents stored against transactions, which your accountant and the ATO will both appreciate.
Limits: extraction is reliable for standard one-total invoices but doesn’t really do line items; multi-page or unusual layouts confuse it; processing can lag at month-end; no real approval workflow beyond Xero’s own.
Verdict: if you process under ~50 invoices a month from mainstream suppliers, set up Hubdoc this week and stop reading. Genuinely.
Option 2: Dext (and similar paid tools)
Dext (formerly Receipt Bank) is the most common paid step up; EzzyBills and Lightyear play in the same space, with Lightyear strongest on multi-step approvals.
How it works: same pipeline as Hubdoc, with materially better extraction — line items, more layouts, fewer misreads — plus supplier rules, approval flows, and expense claims for staff receipts.
Cost: from roughly $35–$80+/month depending on plan and volume.
Good: accuracy is noticeably better than Hubdoc; line-item extraction matters if you track spending by category or job; approval workflows suit businesses where someone other than the bookkeeper signs off.
Limits: another subscription forever; per-document pricing tiers sting at volume; still assumes invoices arrive as reasonably normal documents — it won’t log into a supplier portal or untangle a weekly consolidated statement.
Verdict: the right answer for most businesses processing 50–300 invoices a month, or anyone who needs line items and approvals. Payback is immediate against even one hour a week of typing.
Option 3: a custom pipeline — when the off-the-shelf tools hit their ceiling
Some invoice workflows are shaped wrong for any subscription tool. The tells:
- Invoices arrive weirdly. Locked supplier portals you must log into, consolidated weekly statements covering dozens of deliveries, CSV exports pretending to be invoices, faxes (still a thing in some industries).
- The coding needs business logic. Splitting one invoice across jobs or cost centres by rules (“fuel levy to account X, but only for deliveries to site Y”), matching against purchase orders, flagging price rises against contracted rates.
- Volume makes per-document pricing silly, or the workflow involves steps no tool offers — cross-checking invoices against delivery dockets, notifying a Slack channel, holding anything over a threshold for the owner.
This is what we built for one client: a system that watches a folder for incoming invoices, OCRs vendor, date and total, files each one into the right Xero account, and posts a Slack notification — with anything ambiguous routed to a human. It reclaims about 3 hours a week, every week, with no per-document fees.
Cost: typically $4,000–$12,000 to build, depending on how messy the inputs are, plus modest hosting (often under $30/month). Maintenance is minimal once the supplier formats are learned.
The maths: 3 hours a week at a $45/hr loaded cost is about $7,000 a year. Against a (say) $8,000 build, that’s payback in 14 months and pure savings after — versus a subscription that never finishes paying.
Verdict: worth it when your invoices are too weird, too many, or your rules too specific for Hubdoc/Dext. Not worth it below ~$5,000 of annual typing time — use Option 1 or 2.
Which option fits you?
| Your situation | Start with |
|---|---|
| < 50 standard invoices/month | Hubdoc (free with Xero) |
| 50–300/month, want line items or approvals | Dext or Lightyear |
| Portals, statements, PO matching, job-splitting rules | Custom pipeline |
| Not sure | Hubdoc for a month — where it fails tells you exactly what you need |
That last row is the same advice we give paying clients: a month of Hubdoc costs nothing and produces a precise list of your weird cases. If the list is empty, you’re done. If it’s long, you’ve just written the spec for the custom build.
Setup tips that apply to every option
- Give suppliers the capture address. The system only sees invoices that reach it; update your accounts-payable email with suppliers once, properly.
- Keep humans on approval for the first month, checking every draft. Accuracy issues show up early and per-supplier; fix the rules, then relax.
- Standardise your chart of accounts first. OCR can’t fix ambiguity about which of your three “Repairs” accounts a bill belongs in. Clean coding rules are 80% of automation accuracy.
- Watch the GST. Australian invoices mix GST-inclusive, GST-free and mixed supplies; whichever tool you use, spot-check GST treatment per supplier early. Your BAS depends on it.
Frequently asked questions
Does Xero have built-in invoice OCR without Hubdoc? Yes — emailing bills to your Xero bills inbox extracts basic fields on most plans, but it’s more limited than Hubdoc. Since Hubdoc is bundled free, there’s rarely a reason to stop at the built-in version.
Is OCR accurate enough to trust? For standard invoices, extraction of totals and dates is reliably high; coding accuracy depends on your rules. The right mental model is “a fast assistant whose drafts you approve”, not “an employee you never check.” Keep approval on.
We’re on MYOB, not Xero — same options? Broadly yes: MYOB has its own capture (formerly via its In Tray), Dext supports MYOB, and a custom pipeline can push to MYOB’s API just as it does Xero’s.
Can automation handle invoices we send to customers too? Different problem, also automatable — recurring invoices, invoice-on-job-completion, and automatic payment reminders. See our small business automation ideas for that side of the ledger.
Still typing invoices?
Tell us how your invoices arrive and how many you handle a month, and we’ll tell you straight whether the answer is the free tool, the $35/month tool, or a custom pipeline — and what that pipeline would cost. Start the conversation; the advice costs nothing either way.