Build vs buy: when a custom internal tool is actually the right call
By Chris Greeff
Most teams should not commission custom software. If an off-the-shelf product fits your process, buy it. If a low-volume automation flow covers the gap, keep it. A custom build earns its cost in three situations: the process is specific enough that no product matches it, per-task or per-seat pricing has started to hurt, or you have hit a hard limit in a no-code tool. This guide is about telling those situations apart.
The five real options
Every internal-tooling decision comes down to five choices. Most comparisons skip the first one, which is a mistake — it is often the right answer.
- 1. Keep doing it by hand
- A spreadsheet and a person. Genuinely correct when the volume is low and the process is still changing shape every month. Automating a process you have not settled yet just locks in the wrong version of it.
- 2. Buy off-the-shelf SaaS
- Someone has already built your process as a product. Cheapest and fastest by a wide margin when it fits — and you should look hard before concluding it does not.
- 3. Glue it together with an automation tool
- Connect the tools you already pay for. Excellent for linear, low-volume flows: when this happens, do that. Costs rise with task volume rather than with value.
- 4. Build on a low-code platform
- Assemble internal screens over your existing databases and APIs, quickly. Powerful, and priced per builder seat for as long as you use it.
- 5. Commission a custom build
- Software shaped to your process rather than to a product's assumptions. Highest upfront cost, and the only option with no per-seat or per-task ceiling.
Side by side
Figures are typical USD list prices for each category as published in August 2026. Avokodo prices are in New Zealand dollars.
| Option | Cost to start | Cost at scale | Who can change it | Where it breaks |
|---|---|---|---|---|
| Spreadsheet + manual | Nothing | Staff hours, rising | Anyone | On volume, and on human error |
| Off-the-shelf SaaS | Nothing | Per seat, forever | Nobody — the vendor decides | When your process differs from the product’s |
| Automation tools | Free tier | From about USD $20/mo; per task | A technical colleague | On task volume, and on branching logic |
| Low-code platforms | Free tier | From about USD $50/builder/mo | A technical colleague | At the platform’s ceiling, and on seat count |
| Custom build | NZ$1,000–NZ$5,000 | NZ$50–NZ$250/mo hosting | Any developer | On upfront cost, and on needing a developer |
Automation platforms generally count a task as each successful action step, with triggers and filters free and overages billed above your plan’s included volume. Low-code platforms generally count builder seats separately from end users, so the bill tracks how many people edit the tool rather than how many use it.
When off-the-shelf wins — and you should not hire anyone
If a product already does 80% of what you need and the missing 20% is preference rather than necessity, buy the product. It will be cheaper than any custom build, it is maintained by a company whose entire business is maintaining it, and it works on Monday.
The trap is the reverse: bending a genuinely specific process to fit a product because the product was cheaper, then paying for that mismatch in staff time every week for three years. The question is not "does this tool do the job" but "what does my team have to do differently to make it fit, and what does that cost annually".
When an automation tool wins
Automation platforms are the right answer for linear, low-volume connective work: a form submission creates a CRM record, a payment posts to a channel, a signed document files itself. Set up in an afternoon, no developer needed, and on a free or entry tier it costs less than lunch.
They stop being the right answer at volume and at complexity. Pricing runs per task — each successful action step — so cost scales with how much work the tool does rather than how much value it creates. Once a flow needs real branching, retries that mean something, or a decision that depends on three systems agreeing, you are building software in a tool not designed for building software. That is usually where the maintenance pain starts.
The honest rule: if your monthly task count is stable and modest, keep what you have. Avokodo will say so.
When low-code wins
Low-code platforms are genuinely good, and they win when you have a technical person in-house who will own the tool long-term. If someone on your team can and will maintain it, low-code gets you internal screens fast and keeps the changes in your hands.
Two things eventually bite. The first is seat pricing — the business tiers run around USD $50 per builder per month billed annually, and that is a bill with no end date. The second is the ceiling: low-code is excellent right up to the point where you need something the platform does not do, and then you are stuck in a way that is expensive to undo.
If you have no in-house technical owner, low-code tends to become the thing nobody maintains — which is worse than either buying or building.
When a custom build wins
Custom is right when the process is the point. If how your business handles orders, claims, jobs or approvals is genuinely specific — and especially if that specificity is part of why customers choose you — bending it to fit a product costs you the thing that made it worth having.
It also wins on arithmetic once subscription pricing compounds. The build is a fixed number agreed before work starts rather than a line item that grows, and it does not get more expensive because you hired three more people or because last month was busy.
It is the wrong call when your process is ordinary, your volumes are small, or nobody on your side can make decisions about scope. Custom software needs someone who knows how the work actually happens.
The three-year arithmetic
Subscription pricing looks small monthly and large in aggregate. Three years is a fair window for an internal tool. The figures below are typical USD list prices converted at roughly 1 USD ≈ 1.7 NZD, rounded, for comparison against Avokodo's New Zealand dollar pricing.
| Option | Three-year maths | Total |
|---|---|---|
| Automation tool, 2,000 tasks/mo | USD $69/mo × 36 | ≈ NZ$4,200 |
| Low-code platform, 5 builder seats | USD $50 × 5 × 36 | ≈ NZ$15,300 |
| Avokodo custom build | NZ$3,000 build + NZ$150/mo × 36 | ≈ NZ$8,400 |
Read that honestly: an automation tool is genuinely cheaper over three years, and if a simple flow does the job you should keep it. The custom build beats the low-code seat model at five builders, and it is the only line here where the number is set by scope agreed upfront rather than by how many people you employ or how busy you get. Task volume and seat count are what move these figures, so run them with your own.
How to decide in ten minutes
Answer these six honestly. They are ordered so the cheapest answers come first.
- 01Has the process stopped changing shape? If not, do not automate it yet.
- 02Have you actually trialled two off-the-shelf products? If not, do that first.
- 03Could a linear "when this, then that" flow cover it? If yes, try an automation tool.
- 04Is your monthly task volume small and stable? If yes, stay on the automation tool.
- 05Do you have someone in-house who will own a low-code tool for years? If yes, low-code is viable.
- 06Is the process specific to how your business works, and does per-seat or per-task pricing already hurt? If yes, build.
If you answered your way to the bottom, a custom build is probably right. If you stopped earlier, the cheaper answer is the correct one, and Chris Greeff would rather tell you that on a call than sell you a build you did not need.
Common questions
Is it cheaper to build or buy internal software?
Buying is almost always cheaper upfront and often cheaper over three years for small teams. Building wins when per-seat or per-task pricing compounds, or when the process is specific enough that no product fits. A custom internal tool from Avokodo is NZ$1,000–NZ$5,000 to build and NZ$50–NZ$250 a month to run, against roughly NZ$15,000 over three years for five builder seats on a low-code platform.
When should we move off an automation tool?
When the task bill has become a real line item, when flows need branching and retries rather than a straight line, or when a broken flow causes a problem nobody notices for a day. Below that, an automation tool is the cheaper and better answer and you should stay on it.
Is low-code a good alternative to custom development?
Yes, if you have an in-house technical person who will own it. Business tiers run around USD $50 per builder per month billed annually, which is a permanent cost, and low-code platforms have a ceiling you may eventually hit. Without an internal owner, a low-code app usually becomes the tool nobody maintains.
What does a custom build cost to keep running?
NZ$50–NZ$250 a month, which covers hosting, backups, monitoring, security patches and small changes as you use it. Nothing is metered and nothing is charged per seat, so the figure does not climb as your team grows or as the tool gets busier — you move tiers only if usage changes materially.
How long does a custom internal tool take to build?
Avokodo builds typically go live two to four weeks from the first call: week one maps the process and fixes scope and price, weeks two and three build in the open, and weeks three and four launch and hand over.
Full pricing detail is on the pricing page.