Is it cheaper to keep maintaining our old software or replace it with an off-the-shelf product?

Neither is cheaper as a rule: compare the two over five years, with maintenance, hosting, upgrades and risk on one side and licence per user, implementation, customisation, data migration, integration and the processes the product will not do on the other, and the usual answer is a mix of the two. This page is for a finance, operations or IT manager at a medium or large organisation deciding whether an old custom system is worth keeping.

By Graham Church, founder and managing director. Updated 15 October 2026.

Compare over five years, not at the quote

A replacement decision usually goes wrong at the first comparison: this year’s maintenance bill against the product’s licence per user. Both numbers are real and neither is the cost. Put both options on the same five-year sheet.

Keeping the custom system. The support or maintenance fee, hosting and its eventual replacement, licences for the database and components, any extended security updates, the upgrades needed to stay on supported versions over the five years, and a line for risk: what it would cost if the one person who knows it left, or if a security fault were found in an unsupported component. Add the staff time spent on workarounds the system makes necessary.

Moving to a product. The licence or subscription for every user who needs access, the implementation partner’s fee, customisation and configuration, moving and cleaning the data, integration with the systems the custom software currently talks to, training and the dip in productivity after go-live, and then the line that decides most cases: the processes the product will not do, and what the business will do about them. Those become spreadsheets, manual steps or a small bespoke add-on, each with its own cost.

Put the numbers side by side and the answer is often closer than either camp expected. Where it is not close, the decision is easy.

Signs that keeping it is cheaper

  • The process is specific to your organisation and is where you compete or comply. Products are built for the common case, and bending one to an uncommon process costs more each year.
  • The system is on supported versions, or can be brought onto them for a known one-off cost, and small changes reach users in days.
  • It sits in the middle of other systems. Every integration the product has to replicate is a project in itself.
  • The number of users is large. Per-user subscriptions scale with headcount; a maintenance fee for a bespoke system does not.
  • The system is a few years from being retired for other reasons, so a replacement would never pay back.

Signs that replacing it is cheaper

  • The system does a standard job: accounts, payroll, HR, CRM, helpdesk, document management. Products in these areas are mature, and bespoke software earns nothing by being different.
  • Keeping it supported needs a large one-off spend: an upgrade from a platform with no upgrade route, or a rewrite of a part that cannot be changed safely.
  • Nobody who understands it is available, and the source code is missing or cannot be built. There is nothing to maintain.
  • Most of the real work already happens outside it, in spreadsheets and workarounds, so the business has in effect replaced it already.
  • The vendor’s roadmap covers things the business will need next and would otherwise pay to build.

The hybrid is usually the real answer

Few systems are all one thing. A custom system that runs the order process also, somewhere, produces invoices and keeps a contact list, and those parts are standard. The usual outcome is to keep the core and replace the edges: a packaged finance or CRM product takes over the standard parts, connected to the bespoke core by an integration, and the custom code shrinks to the process that is specific to you.

The reverse also happens. A product takes over the main process and a small bespoke piece survives for the one thing it cannot do, such as a pricing rule or a regulatory report. Either way, the comparison is between two combinations, not two systems, and the integration between them belongs on the five-year sheet.

Do the moves in stages. Replace one edge, prove it, retire the old code for that part, then the next. Nothing is switched off until its replacement has carried the load.

Doing the comparison honestly

Ask the product vendor for a reference customer of similar size who moved from a custom system, and ask that customer what the product did not do. Ask whoever maintains the custom system for the list of changes and fixes in the last year, so the maintenance fee has something to be compared against. For the state of the custom system itself, an independent view settles arguments.

The maintain, modernise or replace check asks ten questions and gives a first recommendation with its reasons. Where the answer points to replacement with new software rather than a product, our answer on what it costs to replace legacy software covers that route. CodeFirst, a UK software company that maintains and modernises custom business systems, provides the independent view as a fixed-price code audit; if a product would do the job, the report says so.

Related questions

How do we price the risk of keeping the old system?

Ask what it would cost if the system were unavailable for a week, and what it would cost to get a change made if the one person who knows it left. Then ask how likely each is over five years. Those two figures, weighted by likelihood, are the risk line on the keep side of the sheet, and they are usually larger than people expect for a system on unsupported software.

Why do product implementations cost more than the licence suggests?

The licence is the one number the vendor can quote without looking at you. Implementation, data migration, integration and the processes the product does not do are priced by a partner after discovery, and over five years they are often larger than the licence. Ask for a reference customer of similar size and what their total came to.

Can we keep our custom system and still use a packaged product?

Yes, and that is the usual outcome. A product takes over the standard parts, such as finance or CRM, and the custom system keeps the process that is specific to the business, with an integration between the two. The integration is a cost in its own right and belongs in the comparison.

Who should do the comparison?

The finance or operations manager who pays for the system, with the IT manager, and someone independent of both the product vendor and the current maintainer for the state of the custom system. The vendor will price the product and the maintainer will defend the system. An independent audit of the custom system gives both sides the same facts.

Want the answer for your system?

Tell us what the software does and what you are trying to decide. We will reply with a straight answer and what we would need to know to be more precise.

Tell us about your system 0800 433 7990 Monday to Friday, 9am to 5pm. A first 20-minute call is free, and we reply to every enquiry within one working day. What happens after you get in touch