Built to RUN — The accountability gap is closing. Most practices aren’t ready for it.
The accountability gap is closing. Most practices aren’t ready for it.
Two weeks of cases, data and policy signals all point at the same structural problem: providers carry the attribution, students carry the consequence, and agents sit in the middle with neither the oversight nor the documentation to defend their position when it matters. This issue maps what changed and what to do about it.
Welcome back to Built to RUN.
The last fortnight has produced more usable signal for migration practice than any two-week stretch this year. A CPD session on MARA complaints. A real fraud case where an agent ignored a college’s withdrawal request. Refusal data spilling from Level 3 to Level 2 and now Level 1 source markets. A government collecting nearly three billion dollars in non-refundable visa fees while pushing the consequence of refusal onto everyone except the lodging party.
None of these stories are about agents in isolation. But every one of them lands on the agent’s desk eventually.
This issue connects them.
01 · What the MARA CPD room actually revealed
Documentation discipline is the floor
I sat in on a CPD session two weeks ago. The advertised topic was how to handle a MARA complaint and investigation. The actual content was more useful: a candid picture of where Australian migration practice is in 2026.
Six signals came out of that room. Three matter most for this issue.
The complaints process runs on documentation.
When OMARA arrives, it asks for your records. Contracts. Correspondence. File notes. A timestamped timeline of who said what to whom and when. Most practices cannot produce this on demand. They have the work product, but it lives in inboxes, WhatsApp threads, head-knowledge of the principal agent, and a filing system designed by accident over the years.
A complaint is rarely about the technical lodgement.
It is about how the relationship was managed. Clients who feel heard rarely escalate. Clients who feel ignored almost always do. The compliance failure is usually downstream of a communication failure.
The CPD market is filling slots, not gaps.
Most agents in the room had done their points. Almost none felt better prepared for an actual investigation than they had been a year ago.
The full write-up is on the blog. The point for Issue 4 is that documentation discipline is the floor — and most practices are still living below it.
02 · A real case — and why the attribution model is broken
Least information, most consequence
Late April: a selective, well-regarded CRICOS provider receives what looks like a single ELICOS application for 24 weeks. Student affirms it in writing. College accepts. CoE issued.
Behind the scenes, an education agent had lodged the visa as a family of three. The financial capacity evidence was AI-generated. The application was refused for fraud.
The agent’s response to the s57 request was thin and late. The college only learned the truth on the day the response was due. They requested immediate withdrawal. The agent ignored them and submitted anyway.
Result
- Provider grant rate: down 0.25 percentage points
- Provider exposure: potential full refund under TPS
- Agent: unaffected
- Student: fraud finding on record
This is not a one-off case. It is the structural shape of the problem. The party with least information about the underlying conduct carries the most consequence. The party with most control over the application carries no formal accountability for outcomes.
Providers are now asking publicly for what should have existed years ago: visibility into lodgements linked to their CoEs, notification when a request for further information is issued, the ability to flag concerns before grant. PRISMS and ImmiAccount sit in parallel. They could be connected. They have not been.
For agents reading this, the implication is direct.
Provider patience is finite. The next contract review you do will likely include clauses requiring earlier visibility, faster communication on s57 requests, and consequences for ignoring withdrawal instructions.
The agents who have already built that operational discipline will sign new agreements without friction. The agents who have not will lose channel partners.
03 · The refusal data has moved — and so has the pipeline
Your case mix is shifting
Reports from agents and colleges across the past fortnight are converging on the same picture. Roughly one in three student visa applications is now being refused, weighted heavily toward private providers. Historically Level 3 source countries took the brunt. The pattern has now spilled into Level 2 and even Level 1 markets that used to be considered low-risk.
The financial geometry behind this is worth holding in mind:
→ Visa application charges generated approximately $2.6 billion for government in 2024-25
→ Roughly 30 per cent of that comes from the non-refundable $2,000 component on the student visa
→ Students from LATAM and Europe — historically reliable cohorts — are openly looking at Malta, Germany, Ireland as alternatives
Australia, the UK and Canada have all closed doors at roughly the same time. The genuine students who built our pipeline over twenty years are not stuck deciding between us. They are choosing somewhere else.
For migration practice this is a caseload composition issue, not a sympathy issue. The mix of cases on your desk over the next 12 months will look different from the mix you have been running for the last five years. Source markets you have not worked in will become your bread and butter. Markets you built your business on will thin out.
The agents who treat this as a temporary fluctuation will get caught flat. The ones who treat it as a structural shift and rebuild their intake processes accordingly will be in business in 2028.
04 · The operating posture — what good looks like now
Three changes, all non-negotiable
Pulling these threads together, the operating posture for a migration practice in mid-2026 has tightened in three specific ways.
Documentation is no longer optional.
Every lodgement, every variation, every s57 response, every conversation with a provider needs to leave a timestamped record in a system you control. Not in your inbox. Not in your head. Not in a spreadsheet maintained by one staff member. In a system that survives staff turnover, software changes, and a regulator asking for files three years from now.
Transparency with providers is no longer goodwill.
It is the price of doing business. Notify when a visa is lodged. Disclose family composition. Confirm the GS letter is on file. Pass on s57 requests inside 24 hours, not at day 27 of a 28-day window. Providers asking for this are not being unreasonable. They are managing exposure that is now formally theirs to manage.
Your case mix is shifting and your knowledge base needs to shift with it.
The agents who can speak credibly about LATAM evidence requirements, European GS framing, or onshore transfer rules under the new ESOS framework are positioned to absorb the realignment. The agents whose practice is built around one or two source markets and one or two visa subclasses are exposed.
This is what RUN-Ready actually looks like in 2026. Not a theoretical maturity score. The concrete capacity to operate transparently, document defensibly, and absorb the case mix shift without losing margin.
05 · The diagnostic
Find your actual constraint
If any of the above made you uncomfortable in a recognisable way, that is the diagnostic working.
The RUN Readiness Assessment maps where your practice sits across the four STKN layers — Systems, Technology, Knowledge, Networks — and tells you which one is your actual constraint. Five minutes. No login required. Honest answers only.
The agents using the assessment most usefully right now are the ones who already suspect they have a documentation problem, a provider communication problem, or a case mix problem, and want to know which one to fix first.
Most practices try to fix all three at once. That is why most plateau.
Closing
The accountability gap that has run through this sector for fifteen years is closing. Slowly, untidily, with the wrong incentives in places, but it is closing. Providers are asking for visibility. Regulators are asking for documentation. Students are asking for outcomes that match what they were sold.
The practices that built systems for this in advance will treat the next two years as a sorting event in their favour. The practices that did not will spend the next two years explaining why.
Issue 5 will go deeper into the documentation layer specifically — what a defensible file looks like, what regulators actually open first, and how the PROOF Engine inside Educli captures evidence as work happens rather than as an end-of-month exercise.
If you have a question about any of the cases or signals in this issue, reply directly. I read every one.
— Jan