---
title: "When to bring in a software house, and what to prepare · Volcanic Minds"
description: "The right time to call a software house, when you do not need one, what to bring to the first meeting and when to pay for a feasibility study."
url: "https://volcanicminds.com/en/guides/when-to-hire-software-house"
lang: "en"
type: "second_level_page"
updated: "2026-10-31"
alternate: "https://volcanicminds.com/guides/when-to-hire-software-house"
summary: "It pays to bring in a software house when a problem costs you time or money every month and no ready-made tool solves it without workarounds: you do not need a project yet, you need a problem described well. To the first meeting you bring today's work, not tomorrow's screens: where time is lost, who will use the software, which systems it must talk to, the real spreadsheets and an order of magnitude for the budget. If a ready-made product is enough for you, a serious software house tells you before the quote."
---

# When to bring in a software house and what to bring to the first meeting

From the idea to the prototype built with AI, from the Excel sheet to the management system that is no longer enough: when to call a supplier, when you do not need one and what to prepare.

## When is the right time to bring in a software house?

When a problem costs you time or money every month, and the tools you have, or could buy, solve it only with workarounds. Where you start from changes the first step, not the rule.

| Where you are today | It is time when | Before you call |
| --- | --- | --- |
| You only have an idea | you can say what problem it solves, for whom and what it costs today | write the problem on one page; if the idea uses AI, start from a process |
| You work with Excel and email | holding the spreadsheets together costs more than a system that replaces them | try a ready-made management system |
| You built a prototype with AI | others use it, it touches real data or it must talk to other systems | have it reviewed before you use it |
| The management system is there but is not enough | the work it does not cover is done by hand, every day | ask the vendor whether a configuration or a module covers it |
| You built your SaaS on your own | the paying customers depend on the product | put code, access and data in order |

For each starting point there is a guide that helps you take the step before you call: [where to start with AI](/guides/start-with-ai-in-business) if you only have an idea, [Excel or business software](/guides/excel-or-business-software) if you work with spreadsheets, [how to have a prototype built with AI reviewed](/guides/ai-prototype-to-production) if you have one, [how to choose between custom and off-the-shelf](/guides/custom-software-pros-cons#how-do-you-choose-between-custom-and-off-the-shelf) if the management system is not enough. For a SaaS the answer is [further on](#if-the-product-is-your-saas-when-do-you-stop-doing-it-alone).

**You do not need to arrive with a project: you need a problem you can describe, and whose cost you know.** The UK government's manual for digital services asks the same before building any service: [turn the solution you have in mind into the problem to be solved](https://www.gov.uk/service-manual/agile-delivery/how-the-discovery-phase-works), and understand what that problem costs today.

**Enterprises using management software**

56% out of 100

Italian enterprises with at least 10 employees, in 2025.

Source: [ISTAT, “Imprese e ICT, anno 2025” (Enterprises and ICT, 2025), 15 December 2025](https://www.istat.it/comunicato-stampa/imprese-e-ict-anno-2025/)

For a company with at least 10 people, then, the starting point is often a management system it already has. The question is not whether to have software, but what to do with the work the management system does not cover: sometimes a module or [a connection with another system](/erp-integration) is enough, sometimes a new piece built alongside is needed.

## When do you not need a software house?

When the work is the same in many companies, when it is a list or an analysis to do once, or when the tool serves you alone and losing it would cost nothing.

| What you need | What is enough |
| --- | --- |
| A list, a calculation, an analysis to do once | a spreadsheet |
| A process that is the same in many companies: invoices, payroll, customer contacts, tickets | a ready-made product, configured |
| A tool for you or a few colleagues, with no personal data and no connections | a tool you build yourself, with AI too |
| Two systems that do not talk to each other | a connection between the two, which is a smaller job than new software |

Google and the AI assistants say so too: asked by a company with 15 technicians and about 30,000 €, Perplexity and Google's AI Mode suggested trying ready-made field service software first.

**If a ready-made product is enough for you we tell you before the quote, even when it is a product we do not make.** The same goes for the rest: [we do not build before the problem has been validated](/frequently-asked-questions#cosa-non-facciamo).

## How much does the problem cost you today?

You count it over a normal week: the hours spent looking for information, copying data and checking errors, person by person; then the rework, the missed deadlines and the work you put off because you cannot keep up. Taken over a year, it is the figure to compare any solution against, ready-made or custom.

| What you count | How you measure it |
| --- | --- |
| Hours lost looking, copying, checking | one week of counting, person by person |
| Errors and rework | how many times a month, and what each one costs |
| Missed deadlines | penalties paid, customers lost or unhappy |
| Work turned down or put off | requests you did not answer in time |

An example with made-up numbers, to replace with yours: three people losing five hours a week each make 15 hours; if an hour of work costs you 30 €, that is 450 € a week, over 20,000 € a year across 46 working weeks. The three-year sum, with licenses, maintenance and the time to change habits, is in the guide [custom or off-the-shelf](/guides/custom-software-pros-cons#how-do-you-do-the-three-year-sum).

**Before we start we define with you [the indicators to measure the result by](/frequently-asked-questions#roi), starting with the time recovered.** It is the sum in this section, written before instead of after.

## What should you prepare for the first meeting?

Today's work, not tomorrow's screens: what does not work and what it costs, the path of a request with its exceptions, who will use the software, which systems it must talk to, the real spreadsheets and an order of magnitude for the budget.

| What you bring | Why it matters |
| --- | --- |
| The problem on one page, with its cost | it says what must change, and what changing it is worth |
| The path of a request, from arrival to closure, with the exceptions | the exceptions, like the customer who changes their mind or the second visit, are handled by word of mouth today and software must handle them in writing |
| Who will use the software, with which roles and how many people | roles and permissions decide what each person sees and can do |
| The systems it must talk to | management system, invoicing, email, warehouse: each connection is a piece of work |
| The real spreadsheets and documents, messy ones too | they show what the data look like, and what moving them into the new system costs |
| An order of magnitude for the budget, and a date that matters | they separate the possible paths before anyone designs them |
| The prototype, if there is one, with what works and what is missing | it shows what you need better than a document |
| Who decides, and who follows the project on your side | without a person who answers, every question waits |

The budget is not there to make you raise the price: it is there to choose the path. Our thresholds are 10,000 € for an MVP, about 15,000 € for an essential management system for a garage and 45,000 € for an Enterprise platform: [they are entry thresholds, not a price list](/guides/custom-software-cost#how-much-does-custom-software-cost-in-figures).

On drawing the screens, Google and the AI assistants disagree. Google's AI Mode says not to draw them, or only drafts, because design and user experience are the software house's job; ChatGPT recommends them, on paper too; Perplexity asks for sketches of 5-8 key screens. **Sketches are fine if they help you explain; the real design comes after the analysis**, when it is known what each screen must do: that is the work of [Design Strategy](/services/design-strategy).

With us, less than this list is enough: in the introductory call you tell us what takes you the most time today, who will use the software and with which roles, which systems it must talk to, where the data sits today. We ask the questions that matter. If you talk to several suppliers, send them all the same document, so the answers can be compared: how to compare them is in the [guide on costs](/guides/custom-software-cost#how-do-you-compare-two-quotes), what the contract must say about the code in [the one on choosing the supplier](/guides/choose-software-house#what-must-the-contract-say-about-the-code).

**Would you rather talk it through?**

Book a call with people who work on these topics every day: we will look at your case together.

- [Book a call](https://vms.srl/booking)

## Do you need a feasibility study before the quote?

For a project of medium complexity, no: the analysis that leads to the quote is enough. You need one when the project is complex, the timeline is tight, or you must choose between different paths before spending: a ready-made product, custom software, a connection between systems.

Ask Google or an AI assistant and you get different answers, and none cites a source for the figures.

| Who answers | What it says about the study |
| --- | --- |
| Google AI Mode | “it absolutely makes sense”, about 1,500-3,000 €, no source |
| ChatGPT | a preliminary phase of 1,500-3,000 €, “a rough indication, not a market price list”; not a traditional study of 5,000-10,000 € |
| Perplexity | a discovery of 5-10 working days, no amount: “Do not pay for an ‘academic’ feasibility study costing thousands of euros” |
| **Volcanic Minds** | free analysis for a project of medium complexity; for a complex one, a study of 1 to 2 weeks, up to 2,500 €, deducted from the project if you sign |

Answers collected on 31 October 2026 to the question, asked in Italian, of a company with 15 technicians and a budget of about 30,000 €; the quotes are our translation.

With us the analysis that leads to the quote is free for a project of medium complexity. For a complex project or a tight timeline we run a feasibility study: it takes 1 to 2 weeks, costs up to 2,500 € and, if you sign, is deducted from the project. It includes a meeting with the people who decide, meetings with whoever runs the process, a reading of the data and a final document with the recommended path and the quote, with a range and a maximum. **The study stays yours even if you do not sign, and it can conclude that building nothing is the better choice.**

The UK government calls this phase discovery: for its services it typically takes 4 to 8 weeks, nothing is built in it and, the manual says, [stopping at the end is not a failure](https://www.gov.uk/service-manual/agile-delivery/how-the-discovery-phase-works) if the research shows it is the best thing to do. Our study is shorter, but the rule is the same: if the answer is not to build, you spent little to find out.

## What if you already have a prototype built with AI?

Bring it: it shows what you need better than any document. But a prototype that works is not software ready for everyday work, and on this the answers of Google and the AI assistants agree; Google's AI Mode adds that its code will probably not be reused.

To the first meeting bring the demo, the list of what works and what is missing, and where the data you put in it sits. **The prototype is the best way to explain to us what you need; whether it becomes the real software is decided by a code review.** With us the review costs 2,500 € at a fixed price and takes one week; if you then give us the work, we deduct it from the quote. What gets checked, and what fixing or rebuilding it costs, is in the guide on the [prototype built with AI](/guides/ai-prototype-to-production#should-you-fix-it-or-rebuild-it).

## If the product is your SaaS, when do you stop doing it alone?

When the paying customers depend on the product and every fix takes away the time to sell it. The other signals: a customer asks where their data is and who can see it, and you cannot answer in writing; only one person knows how to release; every new customer means a copy or a change of the code.

The last one is the most expensive. A platform born for one client and then copied for every new contract holds up while the clients are few; after that, every fix has to be repeated on every copy. That is why [we design a SaaS multi-tenant](/saas-development), one installation for all customers with their data kept apart. In the multi-tenant auditing platform the first working MVP arrived within six to eight weeks, and the whole platform, analysis and infrastructure included, in about four months.

**You do not need to hand us the whole product: [we also take on only one part](/frequently-asked-questions#intero-ciclo-o-parte), and the rest stays with you.**

## An example: 15 technicians and 30,000 €

A service company with 15 technicians receives customer requests by phone and email, assigns the jobs on an Excel sheet and keeps track of deadlines from memory. The owner has a budget of about 30,000 € and wonders whether it is time to call a software house. It is the case we put to Google and the AI assistants; the numbers in the sum are examples.

The first step is the sum. If each technician loses half an hour a day between phone calls to find out where to go and job reports to copy out, over 220 working days that is 1,650 hours a year, close to the 1,760 of a full-time technician, 8 hours for 220 days.

The second is to try ready-made field service software, as Perplexity and Google's AI Mode suggested, with a demo on your own cases and not on the vendor's. If it covers the work, you stop there: a case where it is enough is in the [example of the 25 field technicians](/guides/custom-software-pros-cons#an-example-25-field-technicians).

The third comes if the product leaves out exactly the point that costs: for example a job report that must follow the checks set by each customer's contract and reach invoicing without being copied out. At that point the owner calls, with the problem on one page, the path of a request with the exceptions, the office and technician roles, the Excel sheet and the budget.

**With 30,000 € you do not start from the whole software: you start from the point that costs most, and use decides the next priorities.** With us an MVP, one function only in production and used by your people, starts at 10,000 € and takes 6-8 weeks; the rest of the budget goes to the parts that use proves useful. Google's AI Mode called 30,000 € a “middling” figure, “potentially tight” for custom software from scratch: it is, if you want all of it at once.

## Frequently asked questions

### Do I need a defined project before calling a software house?

No. You need a problem described well: what does not work, what it costs you, who will use the software and which systems it must talk to. The project is written by the analysis, together with you.

### What does the first meeting with you cost?

Nothing, it is an introductory call, free and with no commitment. The analysis that leads to the quote is free too for a project of medium complexity. For a complex project or a tight timeline we run a feasibility study of up to 2,500 €, deducted from the project if you sign and yours to keep if you do not.

### Should I draw the screens before calling you?

No. Sketches are fine if they help you explain, but the real design comes after the analysis, when it is known what each screen must do. Bring instead the spreadsheets and documents you use today.

### Is it worth asking for several quotes?

Yes, if you send them all the same document: otherwise you get offers that cannot be compared. Ask each one for the range with the maximum, what it includes, the recurring costs and who will own the code.

### Can I reuse the prototype I built with AI?

Sometimes yes, sometimes no, and a code review decides. With us it costs 2,500 € at a fixed price, takes one week and is deducted from the quote if you then give us the work. The prototype is useful anyway, because it shows what you need better than a document.

### And if a ready-made product is enough for me?

Better to know before spending. If the analysis shows that a ready-made product is enough for you we tell you before the quote, even when it is a product we do not make.

## Want to know whether it is time?

Tell us about the problem and how you work today, in an introductory call, free and with no commitment: we tell you whether you need a software house and, if not, what is enough for you.

- [Let's talk about your project](/en/contacts)

## Articles on the same topics.

*Keep reading*

- [Green software: ethics with profit](/en/insights/ai-roi-green-software-profit-strategy.md): Cut your cloud bill and raise operating margins with green software: why energy efficiency is the ROI metric that turns sustainability into an edge.
- [This website for people and agents](/en/insights/static-website-for-humans-and-ai-agents.md): The Volcanic Minds website is static, built with Astro, managed with a Git-based CMS and readable by AI agents via Markdown, llms.txt, JSON-LD and WebMCP.
- [Web Bot Auth: visitors with no eyes](/en/insights/web-bot-auth-agentic-traffic.md): AI agents read your site and convert better than organic, yet they never show up in analytics. How to measure them and verify them with Web Bot Auth.