Custom Software or an
Off-the-Shelf Tool? How to Decide

Compare a ready-made tool and a custom build by the rule your business depends on, the full cost over the same period and who will own the result.

8 min read

Weighing custom software vs off-the-shelf tools comes down to one question: can a ready product hold the rule your business depends on, at a cost and level of control you can live with for years? Answer it with your own data and a full cost comparison before anyone writes a line of code or signs a subscription.

The short answer: buy when a ready tool covers your core workflow through configuration and the supplier can support the rest. Build when the gap is a rule that carries your risk or sets you apart, no setting or connector can hold it, and you can name who will maintain and support it after launch. Many good answers combine the two: standard parts where they fit, custom work where the rule lives.

Custom software vs off-the-shelf: where do ready tools win?

In any build vs buy software decision, ready tools usually win where the need is common. A post on the UK Government Digital Service technology blog puts it simply: a commodity such as email is better bought, while a capability that is your core business may be worth building (GDS).

The UK government's purchasing guidance lists the conditions that favour buying: a commercially available product meets most of your needs, suppliers can configure it to fit, you need little customisation, specialist support is available on the market, and your organisation can support what it buys (GOV.UK). When those conditions hold, the supplier's ongoing development and support work in your favour.

The same guidance carries a warning worth repeating: even small modifications to off-the-shelf software can remove most of the benefit of using it. Customisations and workarounds can raise costs, make maintenance harder and restrict future upgrades. Configure the tool and use it as intended, or treat the gap as a reason to look further.

Practical note

Before you buy, trial the product on one small but hard problem from your own work, as the GOV.UK guidance suggests. Use your own data, test how it connects to the tools you already run, and check whether the people who will use it can finish their tasks in it.

What are the signs you have outgrown a tool?

A tool rarely fails all at once. These signals tend to appear first:

  • A second system holds the real rule. A spreadsheet, a shared inbox or one person's memory carries the logic the tool cannot.
  • One question needs several tools. Answering a simple question means exports, copies and manual joins.
  • The risk sits in the gap. The step the tool leaves to people is the step where a mistake costs money, a client's trust or compliance.
  • Customisations hold back upgrades. Each supplier release means re-testing your workarounds.
  • Volume outgrew the workaround. Manual steps that were fine at low volume now take a team.
  • You need to own the change. The rule changes on your schedule, and the supplier's roadmap does not include it.

Not every signal calls for a build. If the gap is moving information between tools you already like, an integration may close it; our guide to native integration, connector or custom build covers that choice.

Can you combine standard parts with custom work?

Yes, when the gap is one rule rather than the whole workflow. Build the part where your rule lives, and use standard parts for everything around it.

The legal research platform shows the split. A law firm needed its archive of 130 TB of documents, many of them scans, to be searchable with a source behind every answer. A general chatbot was ruled out: lawyers cannot use an answer they cannot trace, and nothing from one matter may reach someone who is not on it.

The team used a managed cloud model and a standard vector database, which kept the system simple to run. The engineering time went into ingestion with OCR, permissions per matter and citations, because that is where the firm's risk sat. Matter permissions are applied before passages are retrieved for the model, and answers cite the paragraphs they draw on, so the lawyer can open each source and assess the conclusion. The system went live in two months; finding and confirming a source now takes about five seconds instead of hours.

Example: one question, more than fifty sources

The research data platform is the same pattern at data scale. A firm that researches markets, finances and competitors had analysts spending much of their time collecting and reconciling data before any analysis started. The same company appeared under different names in different sources, and many questions needed several systems and manual joins.

The platform is assembled from widely used components: Kafka for data that changes constantly, Airflow for scheduled loads, Spark for processing and ClickHouse as the analytical store, all on AWS. The custom work is the rule that mattered: records from more than 50 sources, about 15 million new ones a day, are cleaned and linked during processing, so the store already knows which records describe the same company when an analyst asks. On top sit tools for analysts, including plain-language questions that become SQL they can read and check.

The client's project account reports queries four times faster on average and 40 percent more tasks completed by the same analytics team.

Two client projects from the case studies, split into standard parts and custom work.

Standard parts where they fit
Standard partsCustom work, where the rule lives
Legal research platform
  • Managed cloud language model
  • Standard vector database
  • Ingestion with OCR for 130 TB of documents
  • Permissions per matter, checked before retrieval
  • Answers cited to the paragraph
Research data platform
  • Kafka for streaming data
  • Airflow for scheduled loads
  • Spark for processing
  • ClickHouse store on AWS
  • Connecting more than 50 sources, each with its own format and schedule
  • Cleaning and linking the same company across sources before storage
  • Analyst tools, including plain-language questions that become readable SQL
  • Legal research platform

    Standard parts

    • Managed cloud language model
    • Standard vector database

    Custom work, where the rule lives

    • Ingestion with OCR for 130 TB of documents
    • Permissions per matter, checked before retrieval
    • Answers cited to the paragraph
  • Research data platform

    Standard parts

    • Kafka for streaming data
    • Airflow for scheduled loads
    • Spark for processing
    • ClickHouse store on AWS

    Custom work, where the rule lives

    • Connecting more than 50 sources, each with its own format and schedule
    • Cleaning and linking the same company across sources before storage
    • Analyst tools, including plain-language questions that become readable SQL
In both projects, the custom work sat where the client's rule lived.

What does each option really cost over time?

Count the total cost of ownership for both options over the same period and the same expected volume. GOV.UK asks teams to understand as much of the full cost of building or buying as possible, across the product's whole lifecycle: building or buying, upgrading, continuous improvement and retirement (GOV.UK). Its Service Standard adds a second test: keep the ability to make a different choice later (GOV.UK Service Manual).

What each option costs over the same period, from getting started to leaving it
CostOff-the-shelf toolCustom software
Getting startedSetup, configuration and data import.Discovery, design, build, testing and data import.
Running itSubscription fees, often per user or by usage, and price changes at renewal.Hosting, monitoring, security updates and support.
Changing itLimited to the supplier's settings and roadmap; workarounds add manual work.Each change designed, built and tested by your team or a partner.
Connecting itThe integrations on offer and their limits.Interfaces you build and maintain.
Leaving itData export, contract terms and the effort of switching.Contract terms on who owns the code, data and documentation, and the effort of handing over to a new team or moving to another system.
People's timeTraining, and the manual steps around the tool's gaps.Training, and the manual steps for cases the system does not cover.
  • Getting started

    Off-the-shelf tool
    Setup, configuration and data import.
    Custom software
    Discovery, design, build, testing and data import.
  • Running it

    Off-the-shelf tool
    Subscription fees, often per user or by usage, and price changes at renewal.
    Custom software
    Hosting, monitoring, security updates and support.
  • Changing it

    Off-the-shelf tool
    Limited to the supplier's settings and roadmap; workarounds add manual work.
    Custom software
    Each change designed, built and tested by your team or a partner.
  • Connecting it

    Off-the-shelf tool
    The integrations on offer and their limits.
    Custom software
    Interfaces you build and maintain.
  • Leaving it

    Off-the-shelf tool
    Data export, contract terms and the effort of switching.
    Custom software
    Contract terms on who owns the code, data and documentation, and the effort of handing over to a new team or moving to another system.
  • People's time

    Off-the-shelf tool
    Training, and the manual steps around the tool's gaps.
    Custom software
    Training, and the manual steps for cases the system does not cover.

Lock-in applies in both columns. GOV.UK's guidance on technical lock-in in the cloud notes that it can never be avoided completely, and that accepting some can help a team deliver faster; the advice is to watch for switching costs that grow out of proportion (GOV.UK). A tool can lock in your data model; a build can lock you to one supplier unless you hold the code, the data and the documentation.

A decision checklist

Four questions, asked in order, before you compare the full cost.
  1. Does a ready tool cover the core workflow with configuration alone?

    Yes

    Buy and configure

    After a trial on one small but hard problem.

    No: continue to question 2

  2. Can an integration or connector close the gap?

    Yes

    Buy and integrate

    Keep the tool, connect it to the rest.

    No: continue to question 3

  3. Is the gap a rule that carries your risk or sets you apart?

    No

    Adapt the process

    Or accept the gap for now.

    Yes: continue to question 4

  4. Can standard parts carry everything around that rule?

    Yes

    Hybrid build

    Standard parts around a custom core.

    No

    Custom build

    With a named owner for maintenance and support.

In every branch: compare the full cost over the same period, and plan the exit.

Use these steps to decide when to build custom software and when a ready tool is enough. Work through them with the people who do the work today:

  1. Write down the one rule the business depends on, in a sentence a colleague could test.
  2. Trial the best ready tool against that rule, using your own data.
  3. Check whether a connector or integration closes the gap.
  4. Count the full cost of each option over the same period and volume.
  5. Name who will own the tool or the code after launch.
  6. Check the way out: data export, contract length and who holds the intellectual property.

If you build, how do you start small?

Keep standard tools for the work they already do well, and build only the part where your rule lives. Scope the first release around one task and write acceptance criteria for it, so you can see quickly whether the custom part earns its place. Run it alongside the current tool for one team or one workflow before moving everything over.

Use open standards for the interfaces between parts. GOV.UK's technology guidance notes that this makes later changes easier (GOV.UK Service Manual), which matters when you swap one standard part for another. Agree in writing who owns the code, the data and the documentation.

If the custom part is a web application, such as a customer portal, dashboard or internal tool used in a browser, scope its first release by the accounts, screens and data connections that one task needs.

Sources

Frequently asked questions

  • Not necessarily. A build's upfront cost is easy to see; its hosting, security updates and support are easier to miss. A tool's subscription is easy to see; its workarounds and price changes at renewal are easier to miss. Compare both over the same period and volume, using the cost table above, before deciding which is cheaper for you.

  • Yes, if you keep the way out open. Export your data regularly in a format another system can read, avoid contracts longer than your plans, and connect tools through open standards so one part can be replaced without rebuilding the rest.

  • Whoever the contract says, so make it explicit. GOV.UK requires government contracts to state who owns the data and the intellectual property, including code and business rules, which is good practice for any buyer. Ask for access to the code repository and design files from the start, and for documentation that another team could work from.

Weighing a tool against a build?

Tell us which tools you use now, where your team works around them and which rule matters most. We can help you compare the options and scope the smallest build that closes the gap, or confirm that you do not need one.

Compare your options

Let’s build your next advantage

Tell us what you want to launch, improve or automate. Let’s discuss the scope and the next step.

Get in touch

Prefer email?

sales@cursopic.com

Cursopic

Tell us about your project

What would you like to launch, improve or automate? A few lines are enough to start.

We use what you send only to reply to you about your request. Privacy policy (opens in a new tab)

Prefer email?

sales@cursopic.com