Is it cheaper to build software than to subscribe to it?

For a workflow specific to one company, frequently yes. Greenleaf, a 60-person Atlanta real-estate group, replaced Salesforce with a custom application built in 25 hours at a build cost of $150 to $200, saving over $100,000, and DocuSign alone had reached $25,000 a year. For commodity functions like email, payroll, and accounting, the rented product remains the better purchase.

A single-purpose application built for one workflow, small enough to be understood by one person and specific enough that no vendor built it. Owned MicroSaaS means the company holds the source code, the data, and the hosting.

What is the SaaS tax?

The recurring cost of renting software that does not fit: per-seat pricing that grows with headcount, features paid for and never used, lock-in that makes migration expensive, customer data split across systems that disagree, and nothing owned when the subscription ends.

No. Commodity SaaS is a good purchase and stays one. What is changing is the economics of the long tail: a workflow-specific application that cost six figures and two quarters to commission in 2020 can now be built in days, which moves the rent-or-own line a long way in favour of owning.

What should a company build and what should it rent?

Rent the commodity. Own the workflow that makes the business different from its competitors. The test is whether a vendor built the product for this company's process or for the average of every customer's process.

What do you get at the end of a MicroSaaS build?

The application, the repository under the company's own account, the data in the company's own hosting, and the operating system used to build it, so the next application is built in-house.

How long does a MicroSaaS build take?

Ten weeks for a full build-operate-transfer engagement at $22,000, including live operation alongside the incumbent tool and the migration at the end.

From SaaS You Rent to MicroSaaS You Own

The Rule

Rent the commodity, own the difference

Email, payroll, accounting, storage, and video calls are commodities. Building them is a waste of a quarter and the rented version is better than anything a 40-person company will produce.

The workflow that makes a business different from its competitors is the opposite case. It is the one no vendor built for, so the company bends its process around a product designed for the average of every customer, then pays per seat for the privilege, then pays again in the hours people spend working around the gap.

That workflow is the candidate. Everything else stays rented.

Evidence

Greenleaf: Salesforce replaced by a 25-hour build

Greenleaf is a 60-person vertically integrated real-estate investment group in Atlanta, operating across eight southeastern states, founded by Dave Codrea.

25 hrs

To replace the Salesforce CRM with a custom application

$150 to $200

Build cost. DocuSign alone had reached $25,000 per year

$100,000+

Saved. Five further applications in development at 10 to 25 hours each

The build is attributed to Replit in the primary source. A separate report, The Information on small companies using Claude to leave Salesforce, documents the same movement and is cited here as context rather than as the source of these figures.

Sources: Replit customer story: Greenleaf and The Information via TipRanks.

Two details in that case carry more weight than the savings figure. The build took 25 hours, which is inside one working week for one person. And five more applications followed at 10 to 25 hours each, which describes a capability the firm now holds rather than a one-off project it commissioned.

The Enemy

What the SaaS tax is made of

The enemy has five components, and each one is measurable inside any company that wants to check.

Component	How it shows up	How to measure it
Per-seat pricing	Cost grows with headcount rather than with usage	Annual spend divided by weekly active users
Unused features	80% of the product priced into every seat	Count the features used weekly against the feature list
Lock-in	Export produces a CSV that cannot rebuild the workflow	Ask what a migration would cost in hours
Fragmented data	Ten tools hold pieces of the truth and none agree	Count systems holding a customer record
Nothing owned	Cancel the subscription and the capability stops	Ask what remains after cancellation

The position is narrow on purpose: rent commodity software, own the workflows that make the business different. A blanket case against SaaS is easy to write and impossible to run.

Ownership

What “you own it” covers

Ownership in this model is specific, and each item is checkable at handover.

The repository. Source code in the company’s own version control, under the company’s own account.
The data. In the company’s database, in the company’s hosting account, exportable without a support ticket.
The hosting. Billed to the company, with the credentials held by the company.
The operating system that built it. The context files, the skills, and the scoring gates used during the build, so the next change is made in-house rather than commissioned.

That last item is the difference between owning an application and owning a capability. An application ages. A capability produces the next five applications, which is exactly the pattern the Greenleaf case shows.

What the build costs and how long it takes

A single-purpose application replacing one rented tool is a 10-week engagement at $22,000, delivered build-operate-transfer: built to the company’s workflow, run live alongside the incumbent until the numbers match, then transferred with the data migrated, the team switched, and the subscription cancelled.

A shorter first step is the MicroSaaS Opportunity Assessment at $2,500 over one week: an audit of current subscriptions, the worst-fit system named, and the replacement priced before anything is committed.

MENA

What changes about this arithmetic in the Gulf

Subscriptions are billed in USD against revenue earned in AED and SAR. A per-seat price that looks flat to a US buyer carries currency and card-processing friction for a Gulf company, and annual renewals arrive in a currency the business does not earn.
Compliance features arrive late or never. ZATCA e-invoicing in Saudi Arabia and the UAE VAT requirements are handled properly by local systems and handled by a workaround inside most international products. The workaround is a person, and that person is a recurring cost that never appears in the subscription line.
Arabic in the interface is usually an afterthought. A rented product with English-only screens on an Arabic-speaking operations team produces adoption at head office and paper in the branches. An owned application ships bilingual because the company decides the requirement.
Data residency is a live question for regulated and government-adjacent buyers. Where the database sits is a procurement question in Saudi Arabia, not a technical footnote, and an owned build lets the company answer it in one sentence.

Who should keep renting

Companies below roughly 20 people, where the workflow is still changing every month and the rented product is a cheaper way to find out what the workflow should be. Companies whose worst-fit tool costs less than $6,000 a year, where the arithmetic does not clear the build. Companies with no internal owner for the application after transfer, because an owned system with no owner becomes an unowned system within two quarters.

The Bill

What replacing a rented tool costs

Route	Price	Duration	What you leave with
MicroSaaS Opportunity Assessment (D2)	$2,500	1 week	The worst-fit system named, the replacement priced
MicroSaaS Builder (T4)	$3,500	5 weeks	A deployed application replacing a real subscription, built by your own team
AI-Native Revenue MicroSaaS (S2)	$22,000	10 weeks	The application, the repository, the data, and the OS that built it

D2 credits 100% against any engagement started within 60 days. The full list is published at pricing.

Questions

Seven questions about owning your software

Question	Answer
Is it cheaper to build software than to subscribe to it?	For a workflow specific to one company, frequently yes. Greenleaf, a 60-person Atlanta real-estate group, replaced Salesforce with a custom application built in 25 hours at a build cost of $150 to $200, saving over $100,000, and DocuSign alone had reached $25,000 a year. For commodity functions like email, payroll, and accounting, the rented product remains the better purchase.
What is MicroSaaS?	A single-purpose application built for one workflow, small enough to be understood by one person and specific enough that no vendor built it. Owned MicroSaaS means the company holds the source code, the data, and the hosting.
What is the SaaS tax?	The recurring cost of renting software that does not fit: per-seat pricing that grows with headcount, features paid for and never used, lock-in that makes migration expensive, customer data split across systems that disagree, and nothing owned when the subscription ends.
Is SaaS dying?	No. Commodity SaaS is a good purchase and stays one. What is changing is the economics of the long tail: a workflow-specific application that cost six figures and two quarters to commission in 2020 can now be built in days, which moves the rent-or-own line a long way in favour of owning.
What should a company build and what should it rent?	Rent the commodity. Own the workflow that makes the business different from its competitors. The test is whether a vendor built the product for this company's process or for the average of every customer's process.
What do you get at the end of a MicroSaaS build?	The application, the repository under the company's own account, the data in the company's own hosting, and the operating system used to build it, so the next application is built in-house.
How long does a MicroSaaS build take?	Ten weeks for a full build-operate-transfer engagement at $22,000, including live operation alongside the incumbent tool and the migration at the end.

From SaaS you rent to MicroSaaS you own