Word Count: ~2,100 words | Reading Time: 9 minutes
"We're going to build something similar ourselves."
It's a response we hear regularly from RTOs exploring student engagement automation. And on the surface, it makes complete sense. You know your students better than anyone. You understand your courses intimately. Your IT team has built systems before. Why not create exactly what you need?
The answer lies in sobering research from MIT: 95% of organisations that attempt to build AI solutions internally fail to deliver measurable return on investment.
But that headline—as dramatic as it sounds—doesn't tell the full story. Let's examine what actually happens when education providers decide to build rather than buy, and why this decision often costs far more than money.
The MIT Research: Why Internal AI Builds Fail at Alarming Rates
In 2025, MIT's NANDA Initiative published "The GenAI Divide: State of AI in Business 2025", a comprehensive study analysing 300+ AI deployments, 52 organisational interviews, and surveys of 153 senior leaders across multiple industries.
The findings revealed a stark divide in success rates:
Vendor-built specialist solutions: 67% success rate
Internal builds: 33% success rate
Enterprise internal builds: Even lower, due to committee-driven development and lengthy pilot cycles
Lead researcher Aditya Challapally from MIT Media Lab identified why startups and specialist providers succeed where enterprises struggle: "They pick one pain point, execute well, and partner smartly."
RTOs attempting to build their own student engagement platforms face identical challenges to those failed enterprise projects: bloated requirements, nine-month pilot phases, and what the research termed "death by committee."
The Real Cost of Building: Beyond the Price Tag
When Australian RTOs calculate the cost of building internally, they typically focus on obvious expenses: developer salaries, infrastructure, perhaps some AI training costs. But the true cost extends far beyond the initial build.
1. Development Time and Upfront Costs
According to Australian software development industry benchmarks, a functional conversational AI platform requires:
Development time: 6-9 months minimum for an MVP
Team size: 2-3 full-time developers plus project management
Developer costs: AUD $110,000-$128,000 per developer annually (based on 2025 Australian market rates)
Project management: AUD $100,000-$130,000 annually
Infrastructure and tools: AUD $15,000-$30,000
Total initial development cost: AUD $180,000-$350,000
That's the visible cost. Finance sees it. Boards approve it. Everyone understands this number.
But there's a far larger cost that rarely makes it into the business case.
2. The Opportunity Cost: What Those 6-9 Months Actually Cost
Here's what that development timeline really means: 6-9 months of lost enrolments while you build what already exists.
Let's calculate this with real numbers for a mid-sized Australian RTO:
Current state:
200 enquiries per month
Manual response process, average 16-hour response time
Current conversion rate: 12% = 24 enrolments per month
Average course fee: $3,500
Monthly revenue from enquiries: $84,000
With instant automated response:
After-hours enquiries: 73% × 200 = 146 enquiries
Improved after-hours conversion: From 8% to 26% (based on Course Finder Group's 20 years of data)
Business hours improvement: 5-7% lift from instant response
New total monthly enrolments: 50 (up from 24)
Monthly revenue from enquiries: $175,000
The gap: $91,000 per month in lost potential revenue
Over 9 months of development: $819,000 in foregone enrolments
That's nearly a million dollars in potential course revenue you're missing while building what specialist providers have already perfected.
Now add your $250,000 development cost, and your "free" internal build has actually cost your organisation over $1 million.
3. Ongoing Maintenance: The Cost That Never Ends
Software isn't built once and forgotten. Industry-standard maintenance costs for custom software run at 15-20% of initial development cost annually.
For a $250,000 initial build, you're looking at:
Annual maintenance: $37,500-$50,000
Platform updates: Quarterly security patches
AI model retraining: Continuous improvement as conversation patterns evolve
CRM integration updates: Maintaining compatibility as HubSpot, Salesforce, etc. release updates
Compliance updates: Keeping pace with Privacy Act and Spam Act changes
Specialist providers spread these costs across hundreds of customers. You're carrying the full burden alone.
4. Compliance and Risk Management
Australian education providers operate under strict regulatory frameworks:
Privacy Act 1988 and Australian Privacy Principles
Spam Act 2003 for SMS communications
ASQA Standards for RTOs 2025
Student data protection requirements
According to the Office of the Australian Information Commissioner, data breach incidents in the education sector increased 34% from 2022 to 2024.
Building a compliant system requires legal review, security audits, penetration testing, and ongoing compliance monitoring. These aren't one-time costs—they're perpetual obligations that specialist providers build into their core product and spread across all customers.
5. The Hidden Cost: Distraction from Core Mission
Here's the question that matters most: Is your RTO in the business of education or software development?
Every hour your team spends building, debugging, and maintaining student engagement software is an hour not spent on:
Improving course content and delivery
Supporting current students through their learning journey
Developing trainer capability and industry currency
Building industry partnerships that create job pathways
Ensuring compliance with ASQA's Standards for RTOs 2025
Strategic planning for your organisation's future
As the MIT research concluded: successful organisations "pick one pain point, execute well, and partner smartly." They don't try to become software companies while running their core business.
Why Specialist Solutions Succeed Where Internal Builds Fail
The MIT study found specialist tools succeed at twice the rate of internal builds. Why such a dramatic difference?
1. Focused Expertise and Learning at Scale
Specialist providers do one thing exceptionally well. They've invested years refining conversational AI specifically for education enquiries. They've processed millions of student conversations and learned what actually works versus what sounds good in planning meetings.
Your internal development team—no matter how talented—starts from scratch. They're learning lessons that specialist providers learned (and solved) years ago. You're essentially paying for their education in a field that isn't your core competency.
2. Continuous Innovation Funded at Scale
When a specialist provider improves their AI model, adds a feature, or solves a compliance challenge, every customer benefits immediately at no additional cost.
When you build internally, every improvement requires your team's time and your budget. You're choosing between:
Shared innovation: R&D costs distributed across hundreds of customers
Solo innovation: Your budget alone funding all development and improvement
3. Proven Integration and Reliability
Specialist tools come with battle-tested integrations to major CRM platforms (HubSpot, Salesforce, GoHighLevel, Zoho), established uptime SLAs, professional support teams, and comprehensive documentation.
Internal builds require you to:
Build and maintain all integrations yourself
Handle all technical issues with your own resources
Provide your own support infrastructure
Document everything for future team members
Train new staff when developers leave
The "Nine-Month Pilot Cycle" Death Spiral
The MIT research specifically identified "nine-month pilot cycles" as a failure pattern in enterprise AI projects. This is exactly what happens when RTOs decide to build internally:
Month 1-2: Requirements gathering, stakeholder meetings, endless debates about features
Month 3-5: Initial development, discovering the hard problems
Month 6-7: Testing with internal team, finding bugs
Month 8-9: Pilot with small user group, collecting feedback
Month 10-12: Major revisions based on real-world usage
Month 13+: Finally ready for broader deployment
Meanwhile, a specialist solution can be:
Implemented in 15 minutes
Refined based on real production data immediately
Adjusted based on actual student conversations, not theoretical requirements
Generating ROI from day one
Which approach allows you to adapt faster to actual student needs?
When Building Might Make Sense (It's Rare)
To be fair and balanced, there are rare circumstances where building internally could be justified:
You're a major university with dedicated software development teams and budgets exceeding $500,000 for this single project
You have genuinely unique requirements that no existing solution can address (though this is rarer than most organisations think—usually it's "preferred" features, not actual requirements)
You're in the software business anyway and this aligns with your existing technical capabilities and mission
You have unlimited time and aren't concerned about competitors capturing market share while you develop
For the vast majority of RTOs, TAFEs, and private training providers, these conditions don't apply.
The "We Can't Afford the Subscription" Paradox
This objection reveals a fundamental misunderstanding of opportunity cost.
"We can't afford $1,800 per month for a subscription."
"But we can afford to lose $91,000 per month for 9 months while we build?"
The subscription isn't a cost in isolation. It's a fraction of the revenue it enables. Based on the worked example above:
Monthly subscription cost: ~$1,800
Monthly additional revenue enabled: $91,000
Net monthly benefit: $89,200
ROI: 4,956%
When viewed correctly, the question isn't "Can we afford the subscription?" It's "Can we afford NOT to subscribe while we slowly build an inferior alternative?"
Calculate Your Real Build vs. Buy Decision
Before committing to an internal build, answer these questions honestly:
Development cost: Can you build a production-ready, compliant solution for under $200,000?
Timeline: Can you afford to wait 9+ months for functionality you could have today?
Opportunity cost: How many enrolments will you miss during development? (Use the framework above to calculate)
Maintenance: Can you commit $40,000+ annually for ongoing maintenance, updates, and compliance?
Expertise: Do you have conversational AI specialists on staff, or will developers be learning as they go?
Focus: Does building software strengthen or distract from your core educational mission?
Risk tolerance: Can you absorb a 67% chance of failure, or the opportunity cost even if you succeed?
Competitive pressure: Can you afford to fall further behind while competitors implement faster?
If you answered "no" or "unsure" to any of these questions, you're not in the 5% that should build internally.
What the Successful 5% Do Differently
The organisations that succeed with AI—whether in education or any other sector—follow a different playbook:
Identify specific pain points: "Students enquire after hours and we're not responding"
Partner with specialists: Use proven solutions from teams focused solely on solving this problem
Execute on core business: Focus resources on being an excellent education provider
Measure and iterate: Use data from specialist tools to continuously improve student experience
This isn't about lacking ambition or capability. It's about strategic resource allocation that maximises your impact on students' lives.
The Question That Actually Matters
Would you rather invest $250,000 and 9 months building a solution with a 33% success rate, or invest in:
Developing a new high-demand course in AI, data analytics, or cyber security
Comprehensive trainer professional development in your key delivery areas
Student support services that improve completion rates
Marketing that drives more high-quality enquiries to your newly responsive system
Facility improvements that enhance the learning environment
And still have automated student engagement running within 15 minutes, capturing every after-hours enquiry while you focus on what you do best?
The Bottom Line
The MIT data is unambiguous: 95% of organisations building AI internally fail to achieve ROI. Not because they lack intelligence, capability, or commitment—but because they're solving the wrong problem.
The problem isn't "we need custom software built to our exact specifications."
The problem is "students are enquiring and we're not responding fast enough to convert them."
Solving that problem doesn't require you to become a software company. It requires you to partner smartly with specialists who've already solved it thousands of times, while you focus your resources on your actual mission: transforming students' lives through education.
Your students don't care whether you built the response system yourself or partnered with experts. They care that someone answered their question at 9pm on a Tuesday when they were ready to commit to their future.
They care that you made it easy for them to take the next step.
They care that you were there when they needed you.
Be in the 5% that succeeds: focus on education, partner for technology, and start capturing those enquiries today instead of twelve months from now.
Ready to Focus on What You Do Best?
See how StudentIgnite can be live and responding to enquiries in 15 minutes—proven technology from a team with 20 years in Australian education—while you focus on delivering exceptional training outcomes.
Sources: This article draws on research from MIT NANDA Initiative's "The GenAI Divide: State of AI in Business 2025," Australian software development cost data from industry benchmarks, salary data from Glassdoor and Indeed Australia, and 20 years of student enquiry data from Course Finder Group's Australian education marketing platform.