# TSG Build or Buy: pre-run findings

These are results the calculator produces for its own illustrative assumptions, run in advance so they can be quoted without running the tool. All money is in Australian dollars, discounted to today’s dollars at a 3% real rate. Every figure holds only for these assumptions: the organisations are invented, and every value is a sourced or labelled assumption that can be changed. The assumptions and their sources are in [assumptions.md](assumptions.md); [the calculator](index.html) runs any other case. Built by TSG Technology, which builds software for organisations, so it has a commercial interest in the choice to build. Every value, its basis and its sources are published so the figures can be checked.

How to read the tables: each figure is a route’s median total cost over the horizon across 2,000 runs, and, for a build route, the share of runs in which it costs less than SaaS. The leading build route is the one cheaper than SaaS in the most runs. Above 80% of runs it is clearly cheaper; 60% to 80% leans cheaper; 40% to under 60% is too close to call; 20% to under 40% leans more expensive; below 20% clearly more expensive. Every value is at its base; the runs draw each value between its two outer ends.

## Public claims, tested

Each claim is restated as a cost question and run 2,000 times on a made-up organisation at the size the claim implies. Holds means 60% or more of runs agree; Too close to call, 40% to under 60%; Fails, under 40%. One made-up organisation with the model as it stands. It is not a forecast and cannot show what any organisation will save.

| Claim, in our words | Reported in | Question the model runs | Runs that agree | Verdict | Rerun on the reader’s scenario |
| --- | --- | --- | --- | --- | --- |
| Business apps are just databases, so build your own | Outlook Business, 10 January 2025 | Does CRM software built in-house cost less than SaaS, for a made-up 500-person organisation, over five years? | 99% | Holds | Any software |
| A big company can drop Salesforce and build its own CRM | Salesforce Ben, 27 September 2024 | Does CRM software built in-house cost less than SaaS, for a made-up 5,000-person organisation, over five years? | 95% | Holds | CRM only |
| A two-month AI build can replace Salesforce | Startup Fortune, 22 July 2026 | Does CRM software built in-house cost less than SaaS, for a made-up 500-person organisation, over five years? | 99% | Holds | CRM only |
| Apps are not dying | CX Today, 27 August 2026 | Does SaaS cost no more than CRM software built in-house, for a made-up 500-person organisation, over five years? | 1% | Fails | Any software |
| AI is good for helpdesk SaaS | Information Age (Australian Computer Society), 10 February 2026 | Does SaaS cost no more than helpdesk and ticketing software built in-house, for a made-up 200-person organisation, over five years? | 85% | Holds | Helpdesk and ticketing only |
| Replacing SaaS with your own build does not pay | TechCrunch, 4 March 2025 | Does SaaS cost no more than CRM software built in-house, for a made-up 500-person organisation, over five years? | 1% | Fails | CRM only |
| A small team can replace a CRM in a few months | Platform Aeronaut transcripts, 4 May 2026 | Does CRM software built in-house cost less than SaaS, for a made-up 1,000-person organisation, over five years? | 98% | Holds | CRM only |
| Teams are replacing SaaS with custom builds | Retool newsroom, 17 February 2026 | Does forms and surveys software built on an open-source or low-code base cost less than SaaS, for a made-up 200-person organisation, over five years? | 18% | Fails | Any software |
| Easy-to-copy SaaS gets replaced by AI-built apps | Information Age (Australian Computer Society), 10 February 2026 | Does forms and surveys software built by business staff with AI app builders cost less than SaaS, for a made-up 200-person organisation, over five years? | 5% | Fails | Any software |
| A vibe-coded replacement costs more than SaaS | ybrikman.com, 1 August 2026 | Does SaaS cost no more than project timesheets software built in-house, for a made-up 200-person organisation, over five years? | 25% | Fails | Any software |

The claims are our restatements as cost questions, not quotations; what each speaker said is in the sources. On the page, As claimed is the verdict above, and For you asks the same question of the organisation, software and values on screen. A claim about software in general is asked of any software; a claim about one product only when the scenario is about that product.

Claims the model cannot test, because they concern things it takes as inputs or leaves out:

- Reported in SDxCentral, 3 July 2026: A Gartner managing VP describes agentic systems as “an existential threat” for vendors with seat-based models and expects about a fifth of planned enterprise software spend to shift to AI-driven alternatives by 2030. Concerns seat counts falling or vendor hazards over time, which this model holds as inputs, not outcomes.
- Reported in Bain & Company, 23 September 2025: Bain & Company writes that seat-based pricing may not fit when AI does the work, and names helpdesk, invoice processing and time-entry approval as tasks that are easy to automate. Concerns seat counts falling or vendor hazards over time, which this model holds as inputs, not outcomes.
- Reported in Deloitte Insights, 18 November 2025: Deloitte says AI agents could reduce the number of seats organisations need, and that replacing enterprise applications with agents will take “at least five years or more”. Concerns seat counts falling or vendor hazards over time, which this model holds as inputs, not outcomes.
- Reported in IDC blog, 2 December 2025: An IDC associate vice president writes that “by 2028, pure seat-based pricing will be obsolete”, while saying SaaS is changing rather than ending. Concerns seat counts falling or vendor hazards over time, which this model holds as inputs, not outcomes.
- Reported via Yahoo Finance, 18 May 2026: A Bank of America analyst rated Salesforce Underperform, with reported reasoning that AI agents could replace seats rather than add them, which pressures the per-seat model. Concerns seat counts falling or vendor hazards over time, which this model holds as inputs, not outcomes.
- Reported in TechCrunch, 25 February 2026: Salesforce’s CEO said that if there is a SaaSpocalypse, “it may be eaten by the Sasquatch” because many companies use a lot of SaaS. Concerns whether the chance that an established CRM vendor exits or forces a migration stays low; the model takes that chance as an input (the yearly SaaS exit hazard) rather than forecasting it.
- Reported in Capacity Global, 4 February 2026: Nvidia’s CEO called the idea that the software tool is in decline “the most illogical thing in the world”, and said AI’s latest breakthroughs are about tool use. Concerns whether AI agents work through existing SaaS or replace it; the model prices each route to the same software and does not model agents doing the work.
- Reported in The Motley Fool via Yahoo Finance, 17 February 2026: A Motley Fool article reports J.P. Morgan research calling the software sell-off “broken logic” and an “overshoot at this time”. Concerns software share prices and whether vendor risk rose in 2026; the model takes the vendor exit and price-shock hazards as inputs and does not track them over time.
- Reported in UK National Cyber Security Centre blog, 24 March 2026: The UK National Cyber Security Centre’s architecture CTO does not expect organisations to “suddenly vibe code a replacement” for a CRM tool and cites risks outside most organisations’ tolerances; the model asks whether CRM SaaS costs less than a vibe-coded build over 5 years for a 500-person organisation. The model does not offer a business-staff build with an app builder for complex software such as a CRM, which is the claim’s own point, so there is no vibe-coded CRM to price.

## Patterns

- **50 scenarios:** every organisation with each use case it lists, starting greenfields (no system yet), every value at its base, over five years.
- **The leading build route against SaaS:** clearly cheaper in 20; leans cheaper in 3; too close to call in 5; leans more expensive in 9; clearly more expensive in 13. The leading route is the build route cheaper than SaaS in the most runs, as the page leads with it.
- **Which build route leads:** in-house in 8, business staff in 4, agency in 4, open-source/low-code in 34 of the 50.
- **Each build route cheaper than SaaS in more than half the runs:** in-house in 20, business staff in 8 of the 24 where it is available, agency in 15, open-source/low-code in 25 of the 50.
- **Crossover:** the leading route crosses below SaaS within the horizon in 26 of 50, at a median of month 9.
- **By organisation, smallest first, use cases where the leading build route is clearly or leans cheaper:** Quokkaloop (25 staff) 0 of 4; Wattle Lane Community Care (60 staff) 0 of 4; Harlow Pennick Lawyers (60 staff) 0 of 5; Gidgee Freight Lines (150 staff) 3 of 5; Bandicoot Creek Civil (220 staff) 1 of 5; Yarrabin Shire Council (350 staff) 3 of 6; Faculty of Applied Sciences, Tallowood University (400 staff) 3 of 5; Bightwater Resources Authority (900 staff) 4 of 5; Coralstone Mutual Bank (1,200 staff) 4 of 5; Bellbird Coast Health Network (3,000 staff) 5 of 6.

## Each organisation and its use cases

### Yarrabin Shire Council (Regional council, 350 staff, 4 in-house developers)

A regional council of about 350 staff running roads, parks, halls and community grants across a large rural shire. A small IT team of four developers maintains a handful of internal systems alongside the corporate suite. Ratepayer records fall under the Privacy Act, and the council has little tolerance for incidents.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Asset register | A$88k | A$214k, cheaper in under 1% | A$210k, cheaper in under 1% | A$222k, cheaper in under 1% | A$154k, cheaper in 1% | Open-source/low-code, clearly more expensive |
| Grants management | A$267k | A$373k, cheaper in 14% | A$350k, cheaper in 17% | A$501k, cheaper in 4% | A$329k, cheaper in 22% | Open-source/low-code, leans more expensive |
| WHS incident reporting | A$968k | A$763k, cheaper in 76% | not available | A$950k, cheaper in 52% | A$851k, cheaper in 68% | In-house, leans cheaper |
| Room and resource bookings | A$276k | A$266k, cheaper in 53% | A$325k, cheaper in 28% | A$285k, cheaper in 46% | A$202k, cheaper in 84% | Open-source/low-code, clearly cheaper |
| Regulatory obligations register | A$260k | A$227k, cheaper in 59% | not available | A$238k, cheaper in 56% | A$250k, cheaper in 52% | In-house, too close to call |
| Document approval workflow | A$525k | A$325k, cheaper in 97% | A$454k, cheaper in 69% | A$363k, cheaper in 90% | A$355k, cheaper in 92% | In-house, clearly cheaper |

The page’s answer for each:

- **Asset register:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 1% of runs. Mainly because of minimum support for a live build: across its sourced range it moves the gap by A$95k, and a higher value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$5k over the horizon).
- **Grants management:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 22% of runs), with no crossover within five years. Mainly because of grants management price: across its sourced range it moves the gap by A$170k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$7k over the horizon).
- **WHS incident reporting:** Over five years, building in-house with AI agents leans cheaper than SaaS (cheaper in 76% of runs), crossing over in month 38. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$409k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$64k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **Room and resource bookings:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 84% of runs), crossing over in month 7. Mainly because of share of staff booking: across its sourced range it moves the gap by A$185k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$13k over the horizon).
- **Regulatory obligations register:** Over five years, building in-house with AI agents is too close to call against SaaS (cheaper in 59% of runs), crossing over in month 49. Mainly because of regulatory register price: across its sourced range it moves the gap by A$408k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$22k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **Document approval workflow:** Over five years, building in-house with AI agents is clearly cheaper than SaaS (cheaper in 97% of runs), crossing over in month 22. Mainly because of share of staff in approvals: across its sourced range it moves the gap by A$284k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$6k over the horizon).

### Wattle Lane Community Care (Charity, 60 staff, no in-house developers)

A community charity of about 60 staff and many volunteers, funded largely through grants and donations. It has no developers and buys most of its software. Client and donor records are personal information under the Privacy Act.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| CRM | A$610k | A$1m, cheaper in 7% | not available | A$1.2m, cheaper in 5% | A$757k, cheaper in 27% | Open-source/low-code, leans more expensive |
| Grants management | A$191k | A$652k, cheaper in 1% | A$287k, cheaper in 13% | A$459k, cheaper in 3% | A$604k, cheaper in 1% | Business staff, clearly more expensive |
| Forms and surveys | A$69k | A$555k, cheaper in under 1% | A$185k, cheaper in under 1% | A$213k, cheaper in under 1% | A$504k, cheaper in under 1% | Agency, clearly more expensive |
| Expense approvals | A$559k | A$1.1m, cheaper in under 1% | not available | A$1.3m, cheaper in under 1% | A$908k, cheaper in 3% | Open-source/low-code, clearly more expensive |

The page’s answer for each:

- **CRM:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 27% of runs), with no crossover within five years. This route assumes hiring developers first. Mainly because of developer time kept after go-live: across its sourced range it moves the gap by A$544k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$9k over the horizon).
- **Grants management:** Over five years, every build route costs more than SaaS in most runs; the closest is business staff building it with AI app builders, cheaper in 13% of runs. Mainly because of traditional build hours (moderate app): across its sourced range it moves the gap by A$194k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$16k over the horizon). SaaS is live 8 months sooner. Until a build goes live, the model keeps the organisation on SaaS, so a slower build pays SaaS for longer.
- **Forms and surveys:** Over five years, every build route costs more than SaaS in most runs; the closest is having an agency build it, cheaper in under 1% of runs. Mainly because of traditional build hours (simple app): across its sourced range it moves the gap by A$185k, and a higher value makes this route cost more against SaaS. The largest risk is platform end of life (expected A$3k over the horizon).
- **Expense approvals:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 3% of runs. This route assumes hiring developers first. Mainly because of developer time kept after go-live: across its sourced range it moves the gap by A$533k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$12k over the horizon).

### Coralstone Mutual Bank (Mid-tier bank, 1,200 staff, 40 in-house developers)

A customer-owned bank of about 1,200 staff with a 40-strong development team. It is regulated by APRA, and CPS 234 sets how it manages information security, including for its SaaS vendors. Its tolerance for incidents is low.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Regulatory obligations register | A$408k | A$402k, cheaper in 49% | not available | A$432k, cheaper in 41% | A$557k, cheaper in 15% | In-house, too close to call |
| Contract management | A$3m | A$1.4m, cheaper in 98% | not available | A$1.8m, cheaper in 90% | A$1.3m, cheaper in 99% | Open-source/low-code, clearly cheaper |
| Procurement and purchase orders | A$2.2m | A$2.2m, cheaper in 46% | not available | A$2.6m, cheaper in 19% | A$1.7m, cheaper in 89% | Open-source/low-code, clearly cheaper |
| Helpdesk and ticketing | A$1m | A$614k, cheaper in 94% | A$665k, cheaper in 87% | A$761k, cheaper in 81% | A$467k, cheaper in 99% | Open-source/low-code, clearly cheaper |
| Client portal | A$3.8m | A$2.8m, cheaper in 85% | not available | A$3.3m, cheaper in 66% | A$2.4m, cheaper in 97% | Open-source/low-code, clearly cheaper |

The page’s answer for each:

- **Regulatory obligations register:** Over five years, building in-house with AI agents is too close to call against SaaS (cheaper in 49% of runs), with no crossover within five years. Mainly because of regulatory register price: across its sourced range it moves the gap by A$408k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$73k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not. The APRA CPS 234 regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Contract management:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 99% of runs), crossing over in month 1. Mainly because of share of staff on contracts: across its sourced range it moves the gap by A$3,186k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$21k over the horizon). The APRA CPS 234 regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Procurement and purchase orders:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 89% of runs), crossing over in month 10. Mainly because of workaround hours (complex): across its sourced range it moves the gap by A$1,360k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$15k over the horizon). The APRA CPS 234 regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Helpdesk and ticketing:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 99% of runs), crossing over in month 1. Mainly because of helpdesk agent price: across its sourced range it moves the gap by A$786k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$11k over the horizon). The APRA CPS 234 regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Client portal:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 97% of runs), crossing over in month 9. Mainly because of client portal price per internal seat: across its sourced range it moves the gap by A$2,265k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$33k over the horizon). The APRA CPS 234 regime adds yearly compliance work to every build route, and a share of it to SaaS.

### Harlow Pennick Lawyers (Law firm, 60 staff, no in-house developers)

A 60-person commercial law firm with no developers of its own. Fee earners record time in six-minute units, and clients expect a secure place to exchange documents. Client confidentiality makes its tolerance for incidents low.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Client portal | A$435k | A$965k, cheaper in 1% | not available | A$1.1m, cheaper in 1% | A$795k, cheaper in 4% | Open-source/low-code, clearly more expensive |
| Contract management | A$354k | A$898k, cheaper in 1% | not available | A$1m, cheaper in 1% | A$803k, cheaper in 2% | Open-source/low-code, clearly more expensive |
| Project timesheets | A$296k | A$738k, cheaper in under 1% | A$399k, cheaper in 11% | A$533k, cheaper in 4% | A$654k, cheaper in under 1% | Business staff, clearly more expensive |
| Document approval workflow | A$139k | A$574k, cheaper in under 1% | A$225k, cheaper in 5% | A$229k, cheaper in 7% | A$556k, cheaper in under 1% | Agency, clearly more expensive |
| Internal knowledge portal | A$175k | A$583k, cheaper in under 1% | A$243k, cheaper in 12% | A$237k, cheaper in 18% | A$529k, cheaper in 1% | Agency, clearly more expensive |

The page’s answer for each:

- **Client portal:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 4% of runs. This route assumes hiring developers first. Mainly because of developer time kept after go-live: across its sourced range it moves the gap by A$522k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$14k over the horizon).
- **Contract management:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 2% of runs. This route assumes hiring developers first. Mainly because of developer time kept after go-live: across its sourced range it moves the gap by A$500k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$19k over the horizon). SaaS is live 8 months sooner. Until a build goes live, the model keeps the organisation on SaaS, so a slower build pays SaaS for longer.
- **Project timesheets:** Over five years, every build route costs more than SaaS in most runs; the closest is business staff building it with AI app builders, cheaper in 11% of runs. Mainly because of traditional build hours (moderate app): across its sourced range it moves the gap by A$195k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$16k over the horizon). SaaS is live 7 months sooner. Until a build goes live, the model keeps the organisation on SaaS, so a slower build pays SaaS for longer.
- **Document approval workflow:** Over five years, every build route costs more than SaaS in most runs; the closest is having an agency build it, cheaper in 7% of runs. Mainly because of traditional build hours (simple app): across its sourced range it moves the gap by A$185k, and a higher value makes this route cost more against SaaS. The largest risk is platform end of life (expected A$3k over the horizon).
- **Internal knowledge portal:** Over five years, every build route costs more than SaaS in most runs; the closest is having an agency build it, cheaper in 18% of runs. Mainly because of traditional build hours (simple app): across its sourced range it moves the gap by A$186k, and a higher value makes this route cost more against SaaS. The largest risk is platform end of life (expected A$3k over the horizon).

### Faculty of Applied Sciences, Tallowood University (University faculty, 400 staff, 3 in-house developers)

A university faculty of about 400 academic and professional staff. Three developers sit within the faculty, separate from central IT. It manages research grants, lab and room bookings, and student surveys.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Grants management | A$279k | A$378k, cheaper in 15% | A$358k, cheaper in 18% | A$506k, cheaper in 5% | A$339k, cheaper in 23% | Open-source/low-code, leans more expensive |
| Internal knowledge portal | A$817k | A$400k, cheaper in over 99% | A$620k, cheaper in 80% | A$449k, cheaper in 97% | A$293k, cheaper in over 99% | Open-source/low-code, clearly cheaper |
| Room and resource bookings | A$305k | A$275k, cheaper in 61% | A$345k, cheaper in 33% | A$298k, cheaper in 53% | A$209k, cheaper in 89% | Open-source/low-code, clearly cheaper |
| Forms and surveys | A$160k | A$215k, cheaper in 22% | A$211k, cheaper in 21% | A$223k, cheaper in 21% | A$151k, cheaper in 57% | Open-source/low-code, too close to call |
| Helpdesk and ticketing | A$403k | A$384k, cheaper in 55% | A$368k, cheaper in 59% | A$513k, cheaper in 28% | A$233k, cheaper in 96% | Open-source/low-code, clearly cheaper |

The page’s answer for each:

- **Grants management:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 23% of runs), with no crossover within five years. Mainly because of grants management price: across its sourced range it moves the gap by A$170k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$7k over the horizon).
- **Internal knowledge portal:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in over 99% of runs), crossing over in month 1. Mainly because of knowledge portal price: across its sourced range it moves the gap by A$584k, and a lower value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$5k over the horizon).
- **Room and resource bookings:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 89% of runs), crossing over in month 4. Mainly because of share of staff booking: across its sourced range it moves the gap by A$214k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$10k over the horizon).
- **Forms and surveys:** Over five years, building on an open-source or low-code base is too close to call against SaaS (cheaper in 57% of runs), crossing over in month 40. Mainly because of forms and survey builder price: across its sourced range it moves the gap by A$98k, and a lower value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$5k over the horizon).
- **Helpdesk and ticketing:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 96% of runs), crossing over in month 1. Mainly because of helpdesk agent price: across its sourced range it moves the gap by A$271k, and a lower value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$7k over the horizon).

### Gidgee Freight Lines (Logistics SME, 150 staff, 1 in-house developers)

A family-owned freight business of about 150 staff, most of them drivers and warehouse workers. One developer looks after its systems. It is covered by the Privacy Act, and it accepts more risk than most.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Project timesheets | A$586k | A$555k, cheaper in 56% | A$591k, cheaper in 47% | A$687k, cheaper in 30% | A$422k, cheaper in 92% | Open-source/low-code, clearly cheaper |
| WHS incident reporting | A$476k | A$513k, cheaper in 37% | not available | A$654k, cheaper in 18% | A$500k, cheaper in 41% | Open-source/low-code, too close to call |
| Asset register | A$75k | A$203k, cheaper in under 1% | A$192k, cheaper in under 1% | A$215k, cheaper in under 1% | A$146k, cheaper in 1% | Open-source/low-code, clearly more expensive |
| Expense approvals | A$1.1m | A$1.4m, cheaper in 12% | not available | A$1.7m, cheaper in 5% | A$966k, cheaper in 76% | Open-source/low-code, leans cheaper |
| CRM | A$1.2m | A$991k, cheaper in 74% | not available | A$1.4m, cheaper in 40% | A$596k, cheaper in over 99% | Open-source/low-code, clearly cheaper |

The page’s answer for each:

- **Project timesheets:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 92% of runs), crossing over in month 1. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$340k, and a lower value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$7k over the horizon).
- **WHS incident reporting:** Over five years, building on an open-source or low-code base is too close to call against SaaS (cheaper in 41% of runs), with no crossover within five years. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$177k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$13k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **Asset register:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 1% of runs. Mainly because of minimum support for a live build: across its sourced range it moves the gap by A$95k, and a higher value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$5k over the horizon).
- **Expense approvals:** Over five years, building on an open-source or low-code base leans cheaper than SaaS (cheaper in 76% of runs), crossing over in month 31. Mainly because of workaround hours (complex): across its sourced range it moves the gap by A$462k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$12k over the horizon).
- **CRM:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in over 99% of runs), crossing over in month 1. Mainly because of share of staff on CRM: across its sourced range it moves the gap by A$696k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$9k over the horizon).

### Bightwater Resources Authority (State agency, 900 staff, 12 in-house developers)

A state government agency of about 900 staff that licenses and monitors water and mineral resources. A team of 12 developers works within a government security framework based on the ISM. Procurement, approvals and records follow formal policy, and tolerance for incidents is low.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| Procurement and purchase orders | A$1.8m | A$1.8m, cheaper in 41% | not available | A$2.2m, cheaper in 14% | A$1.4m, cheaper in 86% | Open-source/low-code, clearly cheaper |
| Regulatory obligations register | A$394k | A$423k, cheaper in 39% | not available | A$449k, cheaper in 34% | A$537k, cheaper in 14% | In-house, leans more expensive |
| Document approval workflow | A$1.3m | A$658k, cheaper in over 99% | A$1m, cheaper in 79% | A$721k, cheaper in 99% | A$814k, cheaper in 96% | In-house, clearly cheaper |
| Internal knowledge portal | A$1.8m | A$783k, cheaper in over 99% | A$1.3m, cheaper in 82% | A$859k, cheaper in over 99% | A$636k, cheaper in over 99% | Open-source/low-code, clearly cheaper |
| HR and leave | A$5.1m | A$2.9m, cheaper in 97% | not available | A$3.2m, cheaper in 96% | A$3.1m, cheaper in 95% | In-house, clearly cheaper |

The page’s answer for each:

- **Procurement and purchase orders:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 86% of runs), crossing over in month 10. Mainly because of workaround hours (complex): across its sourced range it moves the gap by A$1,094k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$14k over the horizon). The Australian Government ISM regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Regulatory obligations register:** Over five years, building in-house with AI agents leans more expensive than SaaS (cheaper in 39% of runs), with no crossover within five years. Mainly because of regulatory register price: across its sourced range it moves the gap by A$408k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$55k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not. The Australian Government ISM regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Document approval workflow:** Over five years, building in-house with AI agents is clearly cheaper than SaaS (cheaper in over 99% of runs), crossing over in month 12. Mainly because of share of staff in approvals: across its sourced range it moves the gap by A$719k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$7k over the horizon). The Australian Government ISM regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **Internal knowledge portal:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in over 99% of runs), crossing over in month 1. Mainly because of knowledge portal price: across its sourced range it moves the gap by A$1,275k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$9k over the horizon). The Australian Government ISM regime adds yearly compliance work to every build route, and a share of it to SaaS.
- **HR and leave:** Over five years, building in-house with AI agents is clearly cheaper than SaaS (cheaper in 97% of runs), crossing over in month 18. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$3,092k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$549k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not. The Australian Government ISM regime adds yearly compliance work to every build route, and a share of it to SaaS.

### Quokkaloop (Startup, 25 staff, 6 in-house developers)

A software startup of about 25 people, six of them developers. It is growing quickly and has few formal compliance obligations. It accepts a high level of risk.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| CRM | A$336k | A$694k, cheaper in 4% | not available | A$1m, cheaper in 1% | A$322k, cheaper in 56% | Open-source/low-code, too close to call |
| Helpdesk and ticketing | A$92k | A$299k, cheaper in under 1% | A$246k, cheaper in 1% | A$423k, cheaper in under 1% | A$141k, cheaper in 11% | Open-source/low-code, clearly more expensive |
| Expense approvals | A$314k | A$742k, cheaper in 1% | not available | A$1.1m, cheaper in under 1% | A$424k, cheaper in 16% | Open-source/low-code, clearly more expensive |
| HR and leave | A$225k | A$356k, cheaper in 5% | not available | A$485k, cheaper in 1% | A$241k, cheaper in 37% | Open-source/low-code, leans more expensive |

The page’s answer for each:

- **CRM:** Over five years, building on an open-source or low-code base is too close to call against SaaS (cheaper in 56% of runs), crossing over in month 51. Mainly because of traditional build hours (complex app): across its sourced range it moves the gap by A$207k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$10k over the horizon).
- **Helpdesk and ticketing:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 11% of runs. Mainly because of minimum support for a live build: across its sourced range it moves the gap by A$89k, and a higher value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$7k over the horizon).
- **Expense approvals:** Over five years, every build route costs more than SaaS in most runs; the closest is building on an open-source or low-code base, cheaper in 16% of runs. Mainly because of traditional build hours (complex app): across its sourced range it moves the gap by A$273k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$13k over the horizon).
- **HR and leave:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 37% of runs), with no crossover within five years. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$123k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$7k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.

### Bellbird Coast Health Network (Hospital network, 3,000 staff, 25 in-house developers)

A regional health network of about 3,000 staff across three hospitals and several community clinics. Its 25 developers support clinical and corporate systems. Patient and staff records are highly sensitive, and its tolerance for incidents is low.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| HR and leave | A$16.5m | A$9m, cheaper in 96% | not available | A$9.3m, cheaper in 95% | A$9.3m, cheaper in 96% | Open-source/low-code, clearly cheaper |
| WHS incident reporting | A$7.4m | A$4.2m, cheaper in 97% | not available | A$4.5m, cheaper in 95% | A$5.5m, cheaper in 88% | In-house, clearly cheaper |
| Room and resource bookings | A$1.8m | A$693k, cheaper in 99% | A$1.3m, cheaper in 76% | A$768k, cheaper in 98% | A$544k, cheaper in 99% | Open-source/low-code, clearly cheaper |
| Procurement and purchase orders | A$4.6m | A$4.4m, cheaper in 61% | not available | A$4.9m, cheaper in 35% | A$3.5m, cheaper in 92% | Open-source/low-code, clearly cheaper |
| Contract management | A$6.9m | A$2.4m, cheaper in 99% | not available | A$2.9m, cheaper in 98% | A$2.4m, cheaper in over 99% | Open-source/low-code, clearly cheaper |
| Asset register | A$235k | A$321k, cheaper in 12% | A$442k, cheaper in 1% | A$357k, cheaper in 9% | A$254k, cheaper in 39% | Open-source/low-code, leans more expensive |

The page’s answer for each:

- **HR and leave:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 96% of runs), crossing over in month 5. Mainly because of workaround hours (moderate): across its sourced range it moves the gap by A$15,734k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$1,829k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **WHS incident reporting:** Over five years, building in-house with AI agents is clearly cheaper than SaaS (cheaper in 97% of runs), crossing over in month 17. Mainly because of share of staff with WHS seats: across its sourced range it moves the gap by A$3,313k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$549k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **Room and resource bookings:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 99% of runs), crossing over in month 1. Mainly because of share of staff booking: across its sourced range it moves the gap by A$1,533k, and a lower value makes this route cost more against SaaS. The largest risk is security incident (expected A$110k over the horizon).
- **Procurement and purchase orders:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in 92% of runs), crossing over in month 12. Mainly because of workaround hours (complex): across its sourced range it moves the gap by A$3,529k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$17k over the horizon).
- **Contract management:** Over five years, building on an open-source or low-code base is clearly cheaper than SaaS (cheaper in over 99% of runs), crossing over in month 1. Mainly because of share of staff on contracts: across its sourced range it moves the gap by A$7,611k, and a lower value makes this route cost more against SaaS. The largest risk is renewal price shock (expected A$30k over the horizon).
- **Asset register:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 39% of runs), with no crossover within five years. Mainly because of workaround hours (simple): across its sourced range it moves the gap by A$186k, and a lower value makes this route cost more against SaaS. The largest risk is integration breaks (expected A$5k over the horizon).

### Bandicoot Creek Civil (Construction contractor, 220 staff, no in-house developers)

A civil construction contractor of about 220 staff working across several sites at once. It has no developers. Site safety reporting, plant and equipment, and subcontracts take much of its administration.

| Use case | SaaS | In-house | Business staff | Agency | Open-source/low-code | Leading build route |
| --- | --- | --- | --- | --- | --- | --- |
| WHS incident reporting | A$650k | A$954k, cheaper in 11% | not available | A$759k, cheaper in 33% | A$992k, cheaper in 7% | Agency, leans more expensive |
| Project timesheets | A$804k | A$1m, cheaper in 19% | A$754k, cheaper in 60% | A$815k, cheaper in 48% | A$936k, cheaper in 29% | Business staff, leans cheaper |
| Procurement and purchase orders | A$739k | A$1.1m, cheaper in 8% | not available | A$1.3m, cheaper in 6% | A$899k, cheaper in 25% | Open-source/low-code, leans more expensive |
| Contract management | A$709k | A$988k, cheaper in 20% | not available | A$1.1m, cheaper in 14% | A$903k, cheaper in 26% | Open-source/low-code, leans more expensive |
| Asset register | A$79k | A$561k, cheaper in under 1% | A$198k, cheaper in under 1% | A$217k, cheaper in under 1% | A$511k, cheaper in under 1% | Business staff, clearly more expensive |

The page’s answer for each:

- **WHS incident reporting:** Over five years, having an agency build it leans more expensive than SaaS (cheaper in 33% of runs), with no crossover within five years. Mainly because of traditional build hours (moderate app): across its sourced range it moves the gap by A$489k, and a higher value makes this route cost more against SaaS. The largest risk is security incident (expected A$27k over the horizon). The data is regulated; breach costs are in the risk line, regulatory penalties are not.
- **Project timesheets:** Over five years, business staff building it with AI app builders leans cheaper than SaaS (cheaper in 60% of runs), crossing over in month 52. Mainly because of traditional build hours (moderate app): across its sourced range it moves the gap by A$234k, and a higher value makes this route cost more against SaaS. The largest risk is security incident (expected A$36k over the horizon). SaaS is live 9 months sooner. Until a build goes live, the model keeps the organisation on SaaS, so a slower build pays SaaS for longer.
- **Procurement and purchase orders:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 25% of runs), with no crossover within five years. This route assumes hiring developers first. Mainly because of developer time kept after go-live: across its sourced range it moves the gap by A$533k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$12k over the horizon).
- **Contract management:** Over five years, building on an open-source or low-code base leans more expensive than SaaS (cheaper in 26% of runs), with no crossover within five years. This route assumes hiring developers first. Mainly because of share of staff on contracts: across its sourced range it moves the gap by A$585k, and a lower value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$19k over the horizon). SaaS is live 7 months sooner. Until a build goes live, the model keeps the organisation on SaaS, so a slower build pays SaaS for longer.
- **Asset register:** Over five years, every build route costs more than SaaS in most runs; the closest is business staff building it with AI app builders, cheaper in under 1% of runs. Mainly because of minimum support for a live build: across its sourced range it moves the gap by A$75k, and a higher value makes this route cost more against SaaS. The largest risk is losing the key maintainer (expected A$7k over the horizon).