ABC Ltd · Architecture decision brief

Reaching Xero: three routes for the practice build

How ABC connects to Xero and Xero Practice Manager for the Fletchers build, compared on cost, terms, and fit, after Xero's March 2026 pricing change reshaped the ground.

Prepared for Joseph, Scott & Mike Date 1 July 2026 Status For decision
The shared ground

What changed on 2 March 2026

Xero retired its old revenue-share model and moved to five usage tiers priced on two things: connections (each connected client Xero org) and data egress (GB pulled from the API each month). There is no grandfathering. Every route below sits on top of this, so it pays to understand it before choosing.

Starter
$0
5 connections cap · unlimited egress · 1,000 calls/day/org · no App Store listing
Core
~$22/mo
10 GB egress · 5,000 calls/day · card required
Plus
~$155/mo
50 GB egress · listing optional
Advanced
~$895/mo
~£8.4k/yr · up to 10,000 connections · 250 GB egress
Enterprise
POA
Negotiated on application
XPM lives on Advanced only
The Xero Practice Manager API, the Journals endpoint, and Bulk Connections are only available from the Advanced tier, and each needs a security assessment (initial and annual) plus use-case approval. This is the single fact that shapes the whole decision, because your build depends on XPM.
The exemption
Bespoke integrations built by an accountant or bookkeeper for their own practice or a single client are excluded from the new pricing model. Custom Connections keep their existing commercial terms.
Egress overage
$2.40 AUD per GB (~£1.20) above the tier allotment on Core, Plus and Advanced. Chatty polling is now a line item, so webhooks, caching and delta-sync matter commercially, not just technically.
No AI/ML training on Xero data
The revised developer terms prohibit using data obtained via Xero's APIs to train or contribute to any AI or ML model. Applies to every route and must hold in the architecture.

Xero bills in AUD; USD and GBP figures are approximate conversions and exclude tax. Confirm live numbers in the developer portal.

Three ways to get Xero (and, where possible, XPM) data into the build. The score out of ten is a judgement of fit for ABC's actual need, a bespoke, multi-tenant tool for accountancy firms that leans on XPM, weighing cost at current scale, terms cleanliness, control, and how it scales. It is an opinion to argue with, not a verdict.

OPTION A

ABC holds the Xero app

One standard OAuth 2.0 app registered to ABC; each client org connects by consent, ABC holds tokens per tenant. The textbook multi-tenant architecture.

7/10
Best practice
Right shape, heavy floor

What it is

ABC registers a single public OAuth app under an ABC-owned shared address (a dev@ or developers@, not a founder's personal Xero login), and every client's Xero org authorises it. This is the "one dashboard, many client orgs" model, the correct destination for a product company.

Potential pricing

  • ABC building for multiple clients is a vendor app, so the exemption does not apply, full new pricing is in play.
  • Because the build needs XPM, that forces the Advanced tier: ~$895/mo (~£700/mo, ~£8.4k/yr), plus a security assessment and use-case approval before XPM is switched on.
  • Egress overage on top if the sync design is careless.

Gotchas

  • The XPM-on-Advanced cost floor lands from day one, with a single live client, that is a lot of fixed cost to carry.
  • The security assessment is an annual commitment, not a one-off.
  • No grandfathering, and the AI/ML training prohibition constrains what you can do with the pulled data.

Limitations

  • Cost scales with connections and egress, so egress discipline (webhooks, caching, delta-sync) is non-negotiable.
  • You wear Xero's full commercial exposure directly as the app owner.

Why it still scores well

It is the only route that gives ABC one centrally-controlled, multi-tenant product it owns outright and can sell to many firms. The problem is purely the cost floor, which the exemption question (below) could remove.

OPTION B

Fletchers holds the app (their practice)

The app is registered under Fletchers Accountants as a bespoke integration for their own practice, connecting the firm's own client Xero orgs.

7/10
Best practice
Cheap only if XPM clears

What it is

Because Fletchers is a practising firm, an integration built for its own practice reads directly onto the exemption. The developer app lives under Fletchers; ABC's software plugs into it. A future client firm would hold its own exempt app the same way, that per-client pattern is how it scales.

Potential pricing

  • Qualifies for the exemption (bespoke integration for own practice), confirmed in Xero's official FAQ. This removes tier fees and egress charges on the accounting-side data, cheap and terms-clean for that half.
  • But the exemption is a pricing exemption only, and the practice-management (XPM) half your build leans on carries its own access gates the exemption does not clearly lift. See the examined caveat below, it is the deciding factor for this option.
The XPM caveat, examined

Fresh research on the exemption-versus-XPM interaction, from Xero's own FAQ and specialist integrators. The short version: the exemption is real but narrow, and XPM sits behind three separate gates, only one of which the exemption might touch.

  • Gate 1, security assessment (initial and annual). The Xero API Consumer Security Assessment, around 21 questions, is mandatory for XPM. Specialist integrators confirm there are no exemptions, even for internal or single-practice tools. The exemption does not remove this.
  • Gate 2, use-case approval. XPM scopes are not self-serve. You submit the XPM API Access Form, Xero reviews it manually, and the scope is granted only on approval, days to weeks, outside your control. The exemption does not remove this either.
  • Gate 3, Advanced-tier availability. Xero's FAQ states XPM is "only available starting at the Advanced tier" (AUD $1,445/mo, ~£750/mo, the same Advanced tier as Option A). The exemption is written against "this new pricing model", so whether an exempt app counts as being at Advanced for feature access, or sits outside the tier ladder and is simply blocked, is genuinely unresolved in the public docs. This is the crux.

What it means. If the exemption lifts Gate 3, Option B reaches XPM at effectively no Xero fee and is clearly the best route. If it does not, an exempt Fletchers app may still be pushed onto Advanced to unlock XPM, at which point B's headline cost advantage for the practice-management features disappears and it sits level with Option A. Only Xero can settle this, so it must be asked directly before committing.

Roadmap risk, either way. XPM API v3.0 retires 30 April 2027 (migrate to v3.1), some endpoints (Leads, Purchase Orders, Suppliers) switch off from August 2026, and Xero's announced "Partner Hub" will fold XPM, Xero HQ, Workpapers and Tax into one product, so XPM as a standalone target is moving. Build against v3.1 and keep the XPM data-access layer swappable.

Gotchas

  • Dev-account ownership sits with the client, not ABC, your product runs on Fletchers' developer account.
  • Connected-party governance, the ABC/Fletchers relationship is already documented as non-arm's-length; putting the Xero app on Fletchers' side adds to that and should be papered.
  • It does not centralise into one ABC multi-tenant app.

Limitations

  • IP and control friction, ABC's build depends on infrastructure it does not own.
  • Every new client repeats the registration and assessment, no single pane of glass across all connections.

Why the score moved from 8 to 7

On first pass this looked like the clearly cheapest, cleanest route. The research narrows that: the exemption is confirmed but covers pricing only, and the XPM half the build depends on is gated independently, with the tier interaction unresolved. So B is excellent if Xero confirms XPM access under the exemption, and merely level with A if not. That contingency, on the exact feature the build needs, is what pulls it back to 7 and puts the written question to Xero on the critical path.

OPTION C

Microsoft Power Platform connectors to Xero

Use Power Automate's premium Xero connector (plus the standard M365/SharePoint connectors clients already have) as the integration layer, rather than holding a custom Xero app.

4/10
Best practice
Glue, not foundation

What it is

Rather than ABC or Fletchers holding a Xero app, you lean on Microsoft's low-code automation layer, the premium Xero connector for accounting data, and the free standard connectors for the M365/SharePoint side clients already live in.

Potential pricing

  • Power Automate Premium: $15/user/month (billed annually, ~£12) unlocks premium connectors including Xero.
  • Standard M365 connectors (SharePoint, Outlook, Teams) are already in clients' existing licences, no extra cost.
  • Pay-as-you-go alternative ~$0.60 per premium cloud-flow run via Azure; Dataverse 250 MB/user included, extra ~$40/GB/mo.

Gotchas

  • No XPM. The Power Automate Xero connector targets Xero accounting data; XPM is a separate API and is not exposed by it, so this route likely cannot reach the practice-management data the build depends on. Confirm before relying on it.
  • The Xero pricing question does not vanish. Whoever's Xero app sits behind the connector still bears Xero's connection/egress model, worth confirming exactly whose registration that is.
  • Per-user licensing bites. Everyone who triggers or is processed by a shared flow may need a licence, so cost scales with the client's whole team, not one seat.

Limitations

  • Power Automate is an automation layer, not an application platform, it complements ABC's Astro/Cloudflare build, it cannot be the bespoke product itself.
  • Reaching many external client Xero orgs through one Power Platform tenant is awkward for a multi-tenant tool.
  • Microsoft ecosystem lock-in, and a thin code/version-control story compared with your existing stack.

Where it does earn its place

As tactical glue on the M365/SharePoint side, document routing, notifications, list sync, it is cheap and native, given clients' deep Microsoft usage. Keep it in the toolbox for that, not as the Xero+XPM foundation.

Side by side

The comparison at a glance

  A · ABC holds the app B · Fletchers holds the app C · Power Platform connectors
Best-practice score 7 / 10 7 / 10 4 / 10
Likely Xero cost Advanced tier, ~£8.4k/yr (XPM forces it) Exempt on fees; XPM may still force Advanced (~£750/mo) — unresolved $15/user/mo Premium; Xero cost sits behind the connector
Reaches XPM? Yes, at Advanced + assessment Unresolved — pricing-exempt, but 3 access gates remain No, connector is accounting-only
Exemption applies? No (multi-client vendor app) Yes (bespoke, own practice) N/A, routes around it
Who owns it ABC, full control The client (Fletchers) Split across Microsoft + connector owner
Fits a bespoke app? Yes, ideal architecture Yes, per-client pattern No, automation layer only
Scales to many firms Best, one central product Per-client, repeats setup each time Poor for multi-tenant
The read

Recommendation

The whole decision pivots on two answers ABC does not yet have: whether the exemption reaches ABC's structure, and whether an exempt app can touch XPM without paying for the Advanced tier. Until those are settled, committing an architecture is guessing.

Do this first, in writing to Xero
  1. Does the ABC/Fletchers two-company structure qualify for the bespoke-practice / single-client exemption? Describe it plainly, ABC builds the software, Fletchers is the practice, and ask directly.
  2. Does the pricing exemption also lift the "XPM only at the Advanced tier" gate? The research settled that the exemption covers pricing only, and that XPM carries three separate gates, a security assessment (initial and annual), manual use-case approval, and Advanced-tier availability. The first two apply regardless; the open question is purely whether an exempt app reaches XPM without paying for Advanced. Ask that exact question, and start the XPM Access Form and security assessment now, since neither is waived and both take time.
1
If the single-client exemption is confirmed for ABC, Option A becomes the strategic destination at little or no Xero cost for the pilot, you get the central product without the cost floor.
2
If it is not, Option B (Fletchers-owned, exempt on pricing) is the fallback, but only a true cost saving if Xero also confirms XPM access under the exemption. If XPM still forces Advanced, B and A cost much the same for the practice-management features, so decide B on control and the per-client pattern, not on price alone.
3
Option C stays in the toolbox for M365/SharePoint-side automation only, not as the Xero+XPM foundation.
4
Meanwhile, you can develop the non-XPM parts now on the free Starter tier (5 connections), just don't hard-wire the architecture around XPM until the access route is confirmed.