How do you decide if a software tool is worth buying?

Four questions. Does it move revenue, margin, or time. What do the unit economics look like at your real volume rather than the demo volume. Could you build it yourself in an evening. And who owns it if you stop paying. A tool that fails any one of those is usually a no, regardless of price.

When should you build software instead of buying it?

Build when the tool is small, when the subscription costs more per month than your maintenance time would, and when nothing regulated is involved. Buy when it touches payments, customer data, authentication, or compliance. The liability is worth more than the subscription you would save.

What is the true cost of a software subscription?

The sticker price plus usage fees at your actual volume plus the hours you spend setting it up plus the hours you spend every month keeping it running. Most tools that look cheap are cheap only in the first of those four.

Why do AI tools that work in a demo fail at scale?

Because they price per unit of usage while the value they return per unit is fixed or smaller. When the usage fee per customer is higher than the return per customer, growth makes the problem worse rather than better. It is not a pricing problem you can grow out of.

Buy, build, or skip?

I spent about thirty thousand dollars on software, courses, and communities in a year. A good chunk of it went to tools I opened twice. The expensive purchases that survived had nothing in common except that they passed four questions, and I only worked out the questions after the money was gone.

This runs any tool you are about to buy through those four questions and tells you which of the three to do: buy it, build it yourself, or leave it alone.

The purchase

Step 1

Sticker price is the small part. Usage fees and your own setup time are where the real number hides.

What is it called?

Only used on your receipt.

Billing

Price

Per month, all seats.

Their usage fee, per unit

Per contact, per credit, per run. Zero if it is flat rate.

Units per month

Real volume, not the volume in the demo.

Your setup time

hrs

Onboarding, imports, wiring it up. One time.

Your time each month

hrs

Babysitting, fixing, re-running. Almost never zero.

What it moves

Step 2

Pick one. Revenue, margin, or time. A tool that claims all three usually moves none of them.

The lever

Hours saved per month

hrs

Hours, not dollars. The rate below turns them into money.

Your hourly rate

What an hour of your time is worth in billable terms.

How sure are you?

Counted at 50% of what you claimed.

Do those saved hours become paid work?

Counted at 50% of the hourly rate.

Building it yourself

Step 3

The question that kills about half of these purchases. Start with whether you are allowed to.

This touches payments, customer data, logins, or complianceCard numbers, personal data, authentication, anything with a legal surface.

Hours to build v1

hrs

Working version, not polished.

Your own cost per unit

Raw API cost if you run it. Zero if there is none.

Upkeep each month

hrs

Things break. Budget for it.

Who owns it

Step 4

The question worth asking before the money leaves. What is left in your hands the day you stop paying?

I own the code and the dataSelf hosted or built. Nobody can switch it off.I own the data, they own the softwareThere is a real export and I have tested it.They own all of itNo meaningful export. Leaving means starting over.

Shelf risk

Step 5

Three honest questions about your own history with this. The answers do not change the maths, they change the odds.

Tools you own that already do part of this

Count the ones you forgot you were paying for.

Tools bought and abandoned in 90 days

Bought with intent, opened twice, never again.

There is a deadline on the priceDeal ends soon, codes running out, price goes up tomorrow.

The call on this tool

Build it

You are ahead on building it by month 1, and the build pays for itself in 5.1 months. You keep the code, the data, and the ability to change it.

$5,958

True cost, year one. Price plus usage plus your hours.

Never

Payback if you buy it.

5.1 months

Payback if you build it.

Rented

What you keep if you stop paying.

What happens as you grow

Monthly cost against monthly return, as volume rises. This is the chart that catches tools which look great in a demo.

Monthly costMonthly return

You break even at 3.4k units a month. You said you are at 2.0k, so you are short of it.

Buying against building, 24 months

Cumulative cost of each path. Building starts high because your hours are real money, then flattens out.

BuyingBuildingWhat it returns

Building costs more up front and less every month after. You are ahead by month 1.

Shelf risk: worth watching

Some overlap with what you already own. Worth opening those first and checking they cannot do this.

You can get your data out but not the system. Keep an export on a schedule so leaving stays cheap.

The discounts on your estimate and the shelf risk bands are my rules of thumb from spending about thirty thousand dollars badly. They are not benchmarks and they are not research.

Your decision receipt

Every calculator gives you a number and you forget it by Friday. This one writes down what you claimed, so you can be checked against it later.

PURCHASE DECISION RECEIPT
Recorded today using jakebauman.io/tools/buy-build-skip

TOOL:      This tool
PRICE:     $99 monthly
VERDICT:   BUILD IT

WHAT I CLAIMED IT WOULD DO
  Save 14 hours a month, and I said those hours would partly become paid work.
  Confidence at the time: guess (counted at 50%)

THE NUMBERS I SIGNED OFF ON
  True cost, year one:     $5,958
  Counted return, monthly: $333
  Payback if I buy:        never at these numbers
  Payback if I build:      5.1 months
  Ownership:               I own the data, the vendor owns the software

THE CHECK
  In 60 days, open this file and answer one question:
  did the number above actually move?

  If it did, the purchase was right and you now know your rate of return.
  If it did not, cancel it today rather than at renewal.

Or leave your email and I will send you the thinking behind this filter, then ask you in sixty days whether the number actually moved. That is the whole point of writing it down.

The four questions underneath it

Does this move revenue, margin, or time? Not is it interesting. Not did you see it on someone's feed. One of those three, for your business, this quarter. If a tool claims all three it usually moves none of them, which is why the calculator only lets you pick one.

What do the unit economics look like at real volume? The demo runs on ten records. Your business runs on ten thousand. Tools priced per unit can return less per unit than they charge, and that gap widens as you grow. The first chart is there to catch exactly that.

Could you build it yourself in an evening? This question is new and it kills about half the purchases I would have made a year ago. Sitting down with Claude and describing the thing you were about to pay for is now a real alternative for small tools. It is not an alternative for anything touching payments, customer data, logins, or compliance. Buy those. Let someone else carry the liability.

Who owns this if you stop paying? If the answer is that the vendor owns everything and you own nothing, that is rent. Rent can still be worth paying. It just should not be a surprise when you find out.

What this tool is not

It is not a benchmark. The discounts it applies to your estimate, and the bands on the shelf risk score, are my rules of thumb from spending money badly and paying attention afterwards. I have labelled them on the page rather than burying them, because a calculator that hides its assumptions is just your own guess handed back to you with a green tick.

It also will not tell you the answer you want. If you say the hours it saves would not turn into paid work, it counts those hours at zero, because they are worth having and they are not money. That is the part most spreadsheets quietly skip.

If you are on the other side of this and trying to work out whether something you are building will make money, that is a different question with different maths. The AI Profit Test covers it, and there are more free tools here.