Build it in-house or buy it.
Seven questions that decide.
Written against our own interest. If someone in your building has already built a working version, you should probably fund them and stop reading vendor pages.
We lose deals to companies who build it themselves. Not occasionally. It's the single most reliable predictor of how a conversation with us ends.
So this guide is written against our own interest, and that's the point. If you have someone in the building who can already do this, you should probably let them, and you should stop reading vendor pages.
Seven questions. Answer them honestly and the decision usually makes itself. There's no scoring trick at the end that lands on buy.
The seven questions
Has anyone here already built a working version?
Not a plan, not an interest, not a good prompt. Something running that a second person has used.
If the answer is yes, you aren't choosing between build and buy. You're choosing whether to fund the person who already started. That's a much better position than most companies are in and almost nobody recognises it as one.
Yes points hard to build. This is the strongest single signal on the list. Everything below is a tiebreaker by comparison.
Is this a thing you build once, or a thing that runs?
A one-time build is a project. A running system is an obligation. They look identical in a scoping conversation and behave nothing alike six months later.
Runs continuously points to buy, but only if the vendor is still on the hook after handover. A vendor who builds and leaves has sold you a project and called it a system.
Who owns it on the Tuesday after it breaks?
Name the person. Then check whether that's already their job or an eighth hat.
No name points to buy. An unowned internal build is the most expensive outcome available, because you pay for it twice: once to make it and again to work around it.
What happens when the person who built it leaves?
Ask it about the specific person, not in the abstract. If the honest answer is that the work stops, you don't have a system, you have an arrangement.
Work stops points to buy, or to building it very differently than you were about to.
Is the hard part the knowing or the wiring?
The judgment is usually the easy half and it's the half that's public. Anyone can learn what to check. What costs weeks is connecting five systems that were never designed to talk, and keeping them connected when one of them changes.
Mostly wiring points to buy. Mostly judgment points to build, because you already have the judgment and it doesn't transfer well from outside anyway.
How many systems does it have to touch?
Count them. Your CRM, your email, your calendar, your billing, whatever holds the actual work. Each connection is where the hours go, and the count is close to linear.
One or two points to build. Four or more points to buy, and anyone quoting you a flat number without asking this question hasn't scoped it.
What does a month without it cost you?
A real number, even a rough one. Hours at a loaded rate, or deals lost, or a deadline you keep missing.
If you can't produce a number, do neither. Not yet. A build with no cost of delay attached will lose to every urgent thing for as long as it exists.
Reading your answers
Question 1 is yes. Build. Fund the person, give them the time formally instead of informally, and put a review schedule on the output. Nothing a vendor sells you beats someone who already made it work inside your business.
Question 1 is no and questions 5 and 6 lean wiring. Buy. The knowledge is free and the integration work isn't, and that ratio is the entire argument.
Question 7 has no number. Neither. Go get the number first. This is the most common real answer and the one nobody wants to hear.
Everything else. It's genuinely close, which means the deciding factor is capacity, not economics. Pick the option that doesn't require your attention every week.
When we're the wrong call
We turn down work that fits these, and it's cheaper for both of us to establish it early.
- Somebody internal has already built a working version. Covered above. Fund them.
- You want one workflow automated and nothing else. That's a project, and a freelancer will do it for a fraction of what an install costs.
- You can't name what a month without it costs. The engagement will get deprioritised in week three and we'll both have wasted the ramp.
- You want to be trained rather than installed. There are people who teach this well and we aren't one of them. We build and hand over.
Where to start
Give them the time officially
The usual failure isn't capability, it's that the build happens in gaps between real work. Make it the job for a defined stretch and put a check on the output.
Measure one week
Track hours spent on the thing for five working days. Multiply by a loaded rate. That's the number, and it's usually larger than people guess.
Get the integration list scoped
Ask any vendor to price per connected system rather than as one figure. It's the only way to compare quotes, and it exposes who has done it before.
What we actually do
MarqueeOS installs the connected system, hands it over, and optionally runs it with you. The judgment in these guides is free and always will be. The wiring is the engagement.
If the seven questions point you at building it yourself, that's a good outcome and we'd rather you get there from our page than from someone else's.
Keep reading
The 6-Point Check For AI Work Before It Leaves Your Company
Most AI failures don't look like errors. Here's how to grade output you didn't build, and turn it into an error rate you can actually act on.
Read the guide → Operations · 6 min readThe 30-Minute AI Inventory: What Is Running In Your Business And Who Owns It
Four questions, four groups, one specific order. Surfaces the undocumented setup carrying real volume before the person running it resigns.
Read the guide →Run the seven questions
with someone who loses to answer one.
Thirty minutes, no deck. If the honest read is build it yourself, we'll tell you on the call and you'll have saved a procurement cycle.
Book a 30-minute call →