TL;DR
For a startup, build vs buy is not one decision. It is a decision per layer of the product, and the numbers make most of those decisions for you.
The layers you can buy, authentication, payments, hosting, email and analytics, start at no monthly cost and top out near $176 a month on published entry plans. The part you have to build, the thing only your product does, starts at a $15,000 median when an agency publishes a price for it.
So buy everything that is not your product, and spend the build budget only where there is nothing to buy.
Key Takeaways
- Buying the standard layers is close to free at the start. All 14 managed services named in the standard SaaS MVP stack can be started at no monthly cost.
- Even the paid tiers are small. The eight services with a flat entry plan charge $5 to $35 a month, a median of $22.50, and $176 a month for all eight together.
- Building is where the money goes. The median published entry price across 8 MVP development agencies is $15,000, from $3,999 to $45,000.
- The rule that follows: never build what you can rent for tens of dollars a month. Build the one workflow that is your product.
- The costliest mistake is neither. It is building before anyone has shown they want the thing, which buying first lets you test cheaply.
What build vs buy means for a startup
In an established company, build vs buy usually means one big choice: write the system in-house, or license a finished product that already does it. Thoughtworks frames it that way in its build vs buy guide: buying gains “proven capabilities quickly, at the cost of customization and control.”
A startup rarely faces that choice in one piece. A product is a stack of layers, and each layer has its own answer:
- Commodity layers every product needs and none is known for: sign-in, payments, hosting, transactional email, error tracking, product analytics.
- The differentiator: the workflow a customer would pay you for, and the reason your product exists at all.
Buying the commodity layers is almost always right. Building them is how early budgets disappear into login screens and billing logic that no customer will ever notice. The differentiator is the opposite: it is the one thing there is usually nothing to buy for, and the one thing worth paying to build.
The rest of this guide puts numbers on both sides, because the size of the gap between them is the argument.
What buying costs: the numbers
These figures come from reading each vendor’s own pricing page, not from a roundup.
Across the 14 managed services named as the standard SaaS MVP stack, MVP Development found all 14 can be started at no monthly cost. Read 24 September 2026.
Eleven of the 14 are subscriptions. Eight of those publish a flat price for their cheapest paid plan: $5, $20, $20, $20, $25, $25, $26 and $35 a month. The median is $22.50.
Paying all eight entry plans at once comes to $176 a month. That is a ceiling, not a forecast: it assumes you outgrow every free tier at the same moment, and most early products use four or five of these and cross one limit at a time.
The three payment processors in the stack charge no monthly fee at all. They take a share of each sale instead, so their cost is zero until there is revenue.
One caveat matters more than any of the numbers. The free tiers are not equivalent: their limits differ by orders of magnitude and are counted in different units, from monthly active users to events to emails. “Free to start” is true of all 14 and says nothing about how long each stays free for your product. The full breakdown, vendor by vendor, is in the SaaS MVP guide.
What building costs: the numbers
Across the 8 MVP development agencies that publish an entry price, MVP Development found a median of $15,000. The lowest published figure is $3,999 and the highest $45,000. Read 23 September 2026.
Those are entry prices, the lowest figure an agency will put in public, so they are a floor for a first version rather than a quote for yours. Most agencies publish no price at all: only 13 of 66 reachable MVP development agencies publish a price of any kind, read 23 September 2026.
The wider internet is less precise. Of 27 articles quoting what an MVP costs, the median claimed floor is $15,000 and the median claimed ceiling $150,000, and only 2 cite a source for their figures. The detail, and every article counted, is in the guide to what an MVP costs.
Put side by side
The layers you can buy cost tens of dollars a month, and nothing at all until you outgrow a free tier. The layer you have to build costs thousands of dollars before the first user signs in.
That gap is the whole decision for most startups. It means the commodity layers should never come out of the build budget, because there is no version of building your own sign-in that beats a service charging $20 to $35 a month for it, and no version of building your own billing that beats a processor charging nothing until you have revenue. And it means the build budget is precious, so it should go on exactly one thing.
The practical rule: buy everything that is not your product, and build only where there is nothing to buy.
When to buy
Buy a layer when:
- It is a solved problem. Authentication, payments, email delivery, hosting and analytics have mature vendors with free entry tiers.
- No customer would notice if you built it better. Nobody chooses a product for its password reset flow.
- The vendor’s limits sit well above your first year. Check the free tier’s unit, whether that is users, events or rows, against what your first year looks like.
- You are still testing whether anyone wants the product. Buying, or assembling the first version on a no-code MVP platform, gets a real test in front of people for a fraction of a build.
When to build
Build when:
- It is the reason your product exists. The workflow a customer would pay for is the part where owning the code is the point.
- Nothing on the market does it. If a tool already does your core workflow, that is a warning about the business, not a shortcut.
- A platform’s ceiling would force a rebuild soon. Some products hit the limits of a no-code tool within months; building those on a platform you will outgrow only moves the cost later. The trade-off is covered in the custom MVP guide, and when a rebuild becomes unavoidable in when to rebuild your MVP.
- You need to own and scale the code. Investors ask who owns the codebase, and a product whose core runs on someone else’s platform has a harder answer.
The mistake that costs more than either
The expensive failure is not choosing the wrong side of build vs buy. It is building the differentiator before anyone has shown they want it.
Buying the commodity layers and testing demand first is what makes the build budget safe to spend. The order that works for most founders: prove the demand cheaply, then build the part that proved out. The two-minute version of that decision is in how to build an MVP; this guide is the long version with the numbers.
A worked example
Take a scheduling product for physiotherapy clinics. Its differentiator is how it fills cancelled slots from a waiting list.
Buy: sign-in, card payments, hosting, the reminder emails and basic analytics. Every one of those has a vendor with a free entry tier, and paying for all of them at entry prices stays under the $176 a month ceiling above.
Build: the waiting-list matching, which is the feature a clinic would pay for and the one thing no vendor sells.
Test first: before building the matching engine, run it by hand for a few clinics behind a simple interface. If clinics use it, the build budget goes on automating something proven. If they do not, nothing was spent on code.
The example is illustrative. The figures in it are the published ones from the two studies above, not a quote.
What these figures are not
- Not enterprise figures. Both studies are measured at startup scale. A company choosing between building an ERP and licensing one faces numbers several orders of magnitude larger, and this guide does not cover that decision.
- Not a quote. Entry prices are the lowest figure an agency publishes. Your build depends on scope.
- Not equivalent free tiers. “Free to start” is true of all 14 services and hides very different limits.
- Ours, and in the sample. The entry-price study is of agencies like ours, and MVP Development is in it, measured the same way. Both datasets are published so every figure can be checked.
How to cite these figures
Free to quote with attribution.
> Across the 14 managed services named as the standard SaaS MVP stack, MVP Development found all 14 can be started at no monthly cost, with paid entry plans totalling $176 a month; across 8 MVP development agencies that publish an entry price, the median is $15,000 (read 23 and 24 September 2026). https://mvpdevelopment.company/blog/build-vs-buy-software
The datasets: SaaS MVP stack pricing and agency entry prices.
Related guides
- How to build an MVP, the full process this decision sits inside
- Custom MVP, when writing it from scratch is the right call
- No-code MVP, when assembling it on a platform is enough
- SaaS MVP, the standard stack and what each part costs
- What an MVP costs, the build side in full
- MVP tech stack, choosing the layers you buy
- Why MVPs fail, including the failure of building too early
Frequently Asked Questions
What does build vs buy mean?
Build vs buy is the choice between writing software yourself, in-house or through an agency, and paying for an existing product that already does the job. For a startup it is usually a decision per layer: buy the standard parts such as sign-in, payments and hosting, and build the workflow that is your product.
Is it cheaper to build or buy software for a startup?
For the standard layers, buying is far cheaper. Across 14 managed services in the standard SaaS MVP stack, all 14 start at no monthly cost and the paid entry plans total $176 a month. Building a first version starts at a $15,000 median published entry price across 8 agencies, so build money belongs on the part nothing on the market does.
When should a startup build instead of buy?
Build when the feature is the reason your product exists, when nothing on the market does it, when a platform’s limits would force a rebuild within months, or when you need to own and scale the code. Buy everything else.
What should a startup buy instead of build?
Authentication, payments, hosting, transactional email, error tracking and product analytics. All are solved problems with mature vendors and free entry tiers, and no customer chooses a product because its login screen was built in-house.
How much does it cost to build software for a startup?
The median published entry price across 8 MVP development agencies is $15,000, ranging from $3,999 to $45,000, read 23 September 2026. Entry prices are the lowest figure an agency publishes, so treat them as a floor rather than a quote for your scope.
Can I start by buying and switch to building later?
Yes, and for most founders that is the right order. Buy or assemble the first version to test demand, then build the part that proved out. The one risk is a platform whose limits you will hit quickly; if your core workflow is technical, start it on code you own.
Sources and references
- Thoughtworks, “Build vs. buy: a strategic framework for evaluating third-party solutions”, e-book, 2022
- MVP Development, SaaS MVP stack pricing study, read 24 September 2026: dataset
- MVP Development, agency entry-price study, read 23 September 2026: dataset
- MVP Development, MVP cost claims study, read 23 September 2026: dataset
