Mutree SolutionsTalk to sales

Custom Software vs Off-the-Shelf: How to Decide

Ozzy NsakaBuilding agents7 minute read

Most companies asking whether to build custom software already own the answer. It is the off-the-shelf tool they pay for today, set up properly, with one or two pieces built around the edges where it falls short. Our starting position is that buying is the default and building has to earn its place.

That is not modesty. A tool that many companies use has had its rough edges worn down by all of them. Something built for you starts life with none of that wear. The question is never "which is better". It is "where does the standard tool stop fitting the way this business actually works, and is that gap worth owning?"

What off-the-shelf gets right

A mature product arrives with things that are slow and dull to build well: sign-in, permissions, backups, an audit log, a mobile app, an export button, a support desk. Someone else fixes it when the operating system changes underneath it. New staff may have used it at their last job.

It also arrives with opinions. The vendor has decided how a sale moves through stages, how a job is scheduled, how an invoice is approved. For most companies those opinions are fine, and adopting them is cheaper than arguing with them. A team that changes its process to match a good tool often ends up with a better process.

Pick off-the-shelf when the work is common to your whole industry, when nothing about how you do it sets you apart, and when the people who will run it day to day are not technical.

Where it stops fitting

The signs show up in how people work around the tool rather than in the tool itself.

  1. The spreadsheet beside the system. Data is exported, fixed by hand and pasted back, every week, by the one person who knows how.
  2. The same thing typed three times. A job is entered in the sales tool, again in the scheduling tool and again in the accounts package, because none of them talk to each other.
  3. The process bent to fit the software. Staff skip a step the business needs because the tool has no place for it.
  4. The rule that lives in one head. The tool cannot express how you decide, so one person makes every decision of that kind and becomes the bottleneck.
  5. The data you cannot reach. The answer exists in the system, but getting it out takes a support ticket or a report nobody can change.

One of our clients put it plainly: its operations were not efficient. People were duplicating work, redoing things a colleague had already done, and writing the same record in several places.

When we traced one job from the first call to the paid invoice, we counted at least eight manual handoffs. The job ticket was a paper form copied three times. Hours were written in pencil, typed into a spreadsheet by a second person and corrected by a third before payroll. Crews and trucks were assigned from a roster, memory and phone calls, so nobody could look at the bookings and see where more people or equipment would be needed.

One of these on its own is rarely a reason to build. Several together, on work that matters to how you win customers, usually are. Our operations problems and what we would put in place page lists the ones buyers describe to us.

What custom actually buys you

Custom software buys fit. It does what your business does, in the order it does it, with your words on the screen. Nobody learns a workaround, because there is nothing to work around.

It buys control. When we build custom software for you, you own the code and the intelligence in it: the rules, the prompts and the data that make it yours. You decide what changes and when. A vendor cannot retire a feature you depend on, change how it works overnight, or decide your industry is no longer a market they serve.

It can buy an advantage, but only when the thing being built is part of how you compete. A custom tool for something every company does the same way is effort spent on a problem that was already solved.

We built software we run ourselves, Scout360 and Orbit Fleets, rather than assembling them from existing tools. Each started as a problem one of our clients could not solve with the tools it already owned.

What custom asks of you

This is the part that is easiest to leave out of a sales conversation.

Custom software has an owner for life. Someone has to fix it when it breaks, update it when the systems around it change, and answer when a user says it did something odd. If that owner is an outside firm, you depend on that firm. If it is your own team, they need the skills and the time.

It takes longer to reach the first working version than switching on a subscription does. It needs your people in the room while it is designed, because the knowledge it encodes is theirs.

And it carries risk that a mature product has already absorbed: the first version will be wrong in ways nobody predicted. A good build plans for that with a short first release, real users early and a way to change course.

Off-the-shelfCustom
Fits whenThe work is common and does not set you apartThe work is how you compete, or no tool fits it
First working versionSooner: setup and configurationLater: design, building and testing
Who keeps it runningThe vendorYou, or a firm you depend on
Main riskBending the business to the toolOwning every part, including the parts that break

The middle route most companies take

The choice is rarely all or nothing. The pattern we usually start from keeps the core systems off-the-shelf and builds only the connecting piece: an agent that reads from the CRM and writes to the schedule, a small service that removes the weekly spreadsheet, a screen that shows the one view no vendor offers.

This keeps the parts every company shares on software someone else maintains, and spends building effort only where the business is different. It is also how agents usually fit in. An agent is rarely a whole new system. It is a worker that sits between the systems you already have. We wrote about how we choose where an agent lives between Copilot Studio, n8n and code.

Five questions to ask before you decide

  1. Is this work how we win, or how everyone works? Build for the first. Buy for the second.
  2. Can we name the gap in one sentence? If not, the problem is not understood well enough to build for yet.
  3. Have we tried configuring what we own? Many gaps close with settings, an integration or a change of process.
  4. Who owns it after launch? Name the person or team. If nobody can be named, do not build.
  5. What happens when it is wrong? Every system needs a person it hands to and a record of what it did.

When not to build custom

Do not build to avoid a subscription. Do not build because a vendor annoyed you once. Do not build for a process you are about to change, or a team that will be reorganised next quarter. Do not build when nobody inside the company can say what good looks like.

We have given that advice ourselves. For that client, our diagnosis was that it did not lack forms or policies. It lacked a single record of each job and handoffs that were enforced. We told it to fix its data and decide who owns each step first, and to pilot new software on one line of work after its busy season, not across the whole company at once.

Where this leaves you

If the standard tool fits, use it, and put the effort into setting it up well and training people on it. If several of the signs above sound familiar, the next step is not a build. It is a diagnosis: what the work is, where the tools stop fitting and what the smallest useful fix would be. That is how an engagement with us runs, and it often ends with a recommendation that involves very little building.

Talk to sales. Tell us what the workaround looks like and we will tell you whether it is worth building.

Read next

Get new field notes by email

An email when we publish something new. You confirm by email first, and every email has a one-click unsubscribe.

We email you a link to confirm. Every email has a one-click unsubscribe. Privacy