Custom system or off-the-shelf SaaS: how to decide

The short answer

A custom system is software built for one company's specific process, with the code and the data under that company's control. Off-the-shelf SaaS is software rented by subscription, the same for every customer of the vendor. For a market-standard process (tax invoices, accounts payable, a sales pipeline), SaaS almost always wins on cost and on risk. Custom starts to pay off when the process is what sets the operation apart, when the monthly fee grows per user faster than revenue, or when the team spends hours working around what the tool does not do. The decision comes from a three-year bill with the assumptions written down, not from preference. If the case is already a system of your own, the custom web system development hub explains how the project is run.

1. What each option actually delivers

The three real options are not "ready-made" against "custom". They are three ways to pay for software, each with a different owner for the risk.

Off-the-shelf SaaS

You rent the use. The vendor takes care of the server, updates, security, backup and changes in tax rules. In exchange, the flow is theirs: you configure fields, permissions and automations inside what the product allows. The bill arrives every month and, on many products, it grows per user.

Custom system with a solo developer

One person maps the process, writes the code, ships it and maintains it. The flow is yours. The risk is continuity: if that person disappears, whoever inherits the code needs documentation, access to the repository and access to the infrastructure accounts. That is settled in the contract, not on trust.

Custom system with a software house

A team splits discovery, design, development, testing and operations. The flow is yours as well. It fits better in a project with several fronts at the same time or with a hard deadline. The coordination cost exists and it goes into the price.

The article on how much it costs to build a custom web system details what makes up the project price. The focus here is different: compare the three options over time and decide which one fits your case.

2. Where SaaS wins, and wins by a wide margin

Custom is not the automatic answer. There are processes where building from scratch means paying a lot to arrive at the same place, with more risk.

Tax and accounting. Issuing invoices, calculating tax and filing accessory obligations change by regulation, not by your decision. The consumption tax reform, now being rolled out, is the most visible example: a company on a market ERP receives the adjustment as a product update. A company with its own invoice issuer has to rewrite, test and get it approved on the same deadline, on its own.

Standard finance. Accounts payable and receivable, bank reconciliation and cash flow follow the same logic in almost every company. The gain from doing it differently is small.

A common sales pipeline. Lead, opportunity, stage, task and follow-up are the core of any market CRM. If your pipeline fits that mold, the CRM delivers reports, a mobile app and email integration that a custom project would take months to match.

Try before you pay. Bling offers 30 free initial days on any plan and Pipedrive offers a 14-day trial, according to the official pages read on 2026-10-07. No custom project offers that: you pay to find out whether the system fits.

Cost of entry. Bling's Cobalto plan costs R$ 60 per month with 5 users. No custom project delivers tax invoicing, stock and marketplace integration for that amount in the first month.

The practical rule: if the process is the same as your competitor's and the law changes the process often, stay on SaaS. Spending development there means paying to maintain what the vendor already maintains for thousands of customers.

3. The three-year bill, with explicit assumptions

Comparing a monthly fee with a project quote does not work: one is a flow, the other is an event. The fair comparison adds up everything that leaves the cash account over 36 months.

Assumptions behind the table

• A company with 10 people using the system and up to 500 orders per month.
• List prices read on the official pages on 2026-10-07, with no future price increase (which exists and should be estimated by you).
• Amounts in dollars stay in dollars: the exchange rate on the day you pay changes the result.
• No taxes, no paid onboarding and no extra integrations.

Cost table over 36 months

The last two rows stay as letters on purpose. There is no official price table for a project, and any number here would be invented. Your quote goes in place of D and M.

The formula that closes the comparison

SaaS over 3 years    = subscriptions x users x 36 + price adjustments + workaround hours
Custom over 3 years  = project (D) + maintenance (M) x 36 + infrastructure x 36
Compare only the same scope on both sides.

What the table already shows

The entry-level ERP is cheap for what it delivers. On the assumption of 10 users, Bling Titânio adds up to R$ 3.456 over three years and allows up to 50 users at the same price. No custom tax-invoicing project competes with that.

The CRM charged per user is the item that grows the most. The same 10 users on RD Station Basic add up to R$ 23.652 over three years. With 20 users, the bill doubles to R$ 47.304, with no new functionality. This is where the question "is a custom system worth it?" starts to have an answer other than no.

The infrastructure of a small custom system is the smallest part of the bill. A VPS and a managed database add up to low amounts next to what the project and the maintenance cost. The weight of custom work sits in D and M, not in the server.

To check current hosting and VPS plans, open Hostinger through my referral link. Pick the plan according to the resources your project needs.

If the three-year comparison still does not close because your process falls outside what SaaS covers, send the process, how many people use it and what needs to integrate. I will say what stays on SaaS and what calls for a custom system.

Talk about the project

4. What the table does not show

The subscription price is the visible part. Four costs stay off the spreadsheet and often decide the choice.

Workaround hours. When the SaaS does not accept a core rule of the business, someone keeps a parallel spreadsheet, copies data between systems or checks everything by hand at the end of the day. Multiply the hours per week by that person's hourly cost and add it to the SaaS row. In an operation with a daily workaround, that number can pass the monthly fee.

Tier changes. Plans have a limit. Bling's Titânio costs R$ 96 per month in the tier of up to 500 orders; going past that changes the tier. Omie sets the price by monthly gross revenue tier. Growing, in these models, raises the bill automatically, and the three-year projection needs to use the tier you will be in, not today's.

Minimum users and plan steps. RD Station CRM Pro requires a minimum of 4 users, according to the official page. Features you need may exist only on the plan above, and on a plan charged per user the jump applies to everyone, not only to the person who uses the feature.

Integration limit. Whoever connects the SaaS to another system hits the API limit. Bling's API accepts 3 requests per second and 120.000 per day, returns error 429 when that is passed and blocks the IP for 10 minutes after 600 requests in 10 seconds. A well-built integration respects those numbers with a queue and a retry. An improvised integration loses orders on a peak day.

There is also the cost of leaving. Before you sign, check whether the vendor exports all of your data in an open format and whether the API gives access to the full history. Data trapped in a tool is a switching cost that only shows up when you want to switch.

5. Where a custom system wins

Custom makes sense when at least one of these situations is true, and even more when several happen together.

The process is the differentiator. If the way you quote, produce, approve or deliver is the reason the customer buys from you, forcing that process into the generic mold of a SaaS means giving up the advantage. A typical example: a quote calculation with its own rules for material, lead time and margin that a generic CRM does not model.

Many people use little. Per-user billing weighs on operations with many people who only record or look things up (shop floor, delivery drivers, attendants). On a custom system, one more user does not change the license bill.

Three tools do the work of one. When the data is born in a form, passes through a spreadsheet and ends in the ERP, the real cost is the sum of the subscriptions plus the rework between them. A dashboard that concentrates the flow and talks to the ERP by API removes the middle of the path. The signal that this point has arrived is described in when to replace the spreadsheet with an internal dashboard.

A customer or a supplier needs to get in. An external portal with login, permission per company and order history rarely comes ready in a management SaaS, and when it does it is usually on a higher plan.

A rule that changes every week. If the operation adjusts a pricing, commission or approval rule often, depending on the improvement queue of a vendor that serves many customers is slow. On a custom system, the change goes into the next delivery.

If none of these situations describes your company, the honest answer is to stay on SaaS and configure better what you already have.

6. Solo developer or software house: criteria, not price

Once the path is custom, the second choice is who builds it. Both forms work. What changes is the type of project in which each one yields more.

Both models have the same critical point: ownership. In any contract, require these items in writing.

• The code repository stays in an account of your company, with the developer as a guest.
• Server, domain and database stay in accounts in the company's name, billed to your card.
• There is a deployment document: how to bring the system up from zero, where the environment variables live (without the values) and how to restore the backup.
• Maintenance has a scope: what is a fix, what is an improvement and how each one is charged.

With that, changing vendor is work, but it is possible. Without that, the custom system becomes a SaaS with a single customer, with the risk of SaaS and none of the advantages.

7. The model that shows up most in practice: SaaS for the standard, custom at the edge

The choice is rarely all or nothing. In most operations that grow, the design that holds up is hybrid.

1. The market ERP keeps ownership of tax, stock and finance. It receives the legal updates and you do not reinvent the tax invoice.
2. The custom system takes care of what is yours. A quote with an exclusive rule, a customer portal, a production dashboard, approval in stages.
3. The two talk by API, with a queue. The custom system reads and writes to the ERP while respecting the limits documented by the vendor, and it reprocesses what failed instead of losing it.
4. Each piece of data has a single owner. The price lives in one place, the stock in another, and neither is typed twice.

This design shrinks the custom project, because it does not build what the market already solves, and it shrinks the SaaS bill, because it takes off the people who only needed a simple screen. It is also where the three-year bill usually closes better, since D and M cover only the part that sets the operation apart.

8. Seven questions to decide

Answer in writing, with numbers from your operation. The answers point to a better path than any generic comparison.

1. Is the process the same as your competitor's? If yes, SaaS. If it is your differentiator, consider custom.
2. Does the law change this process often? Tax and labor call for a vendor that follows the regulation. Stay on SaaS for that part.
3. How many hours a week does the team spend working around the tool? Convert that into money and add it to the monthly fee.
4. How many users will you have three years from now? Redo the table in section 3 with that number, on the plan tier you will be in.
5. Does someone from outside need to enter the system? A customer or supplier portal weighs in favor of custom.
6. Can you export all the data from the current tool? If not, the cost of leaving grows every month.
7. Is there a budget for ongoing maintenance, and not only for the project? A custom system without maintenance ages. If M does not fit in the cash account, the answer is SaaS.

If most of the answers point to SaaS, invest in configuring and integrating what already exists. If they point to custom, start with the process that hurts the most, with a small scope, and leave the rest where it is.

Frequently asked questions

What is custom?

Custom is software made for one specific company's process, instead of a product that is the same for every customer. The code, the data and the rules stay under the company's control, and the company pays for the project and the maintenance. It does not have to replace everything: in most cases it covers only the process that sets the operation apart and talks to the ERP by API.

Custom system or SaaS: which to choose?

For a standard process (tax, finance, a common sales pipeline), SaaS. For the process that sets the business apart, for an operation with many lookup users, or when there is a portal for an external customer, custom. The decision comes from the three-year bill with the workaround hours included and from the answers to the seven questions in section 8.

Custom software vs SaaS: which produces more results?

What produces more results is whatever removes more manual work for the lower total cost over the period. SaaS yields more where the process is the same as the market's, because the vendor splits the maintenance cost across all customers. Custom software yields more where the ready-made tool forces a parallel spreadsheet, rework, or an expensive plan only to unlock one feature. Measuring the workaround hours before deciding avoids choosing by the sales pitch.

Is a custom system worth it for a small company?

It is worth it when the problem is narrow and expensive: one specific process that today consumes hours every week or blocks a sale. It is not worth it to replicate what an entry-level ERP already does. On the assumption in section 3, Bling Titânio adds up to R$ 3.456 over three years; competing with that by writing tax invoicing from scratch does not close the bill.

Can you start on SaaS and move to a custom system later?

Yes, and it is usually the safer path, as long as the data can leave. Use the ready-made tool to understand the process, write down what needs a workaround and check from the start whether the export and the API give access to the full history. When the custom system comes in, it can live alongside the SaaS instead of replacing it.

Conclusion

SaaS wins on what is standard and regulated; a custom system wins on what sets the operation apart, on use by many people and on access for people outside the company. The fair comparison is three years, with official prices, the future plan tier and workaround hours in the same bill, and the design that holds up most often is hybrid. If you want to run that bill with your company's numbers, send the list of tools the operation already pays for each month, with the user count of each one and the process that today depends on a spreadsheet or on rework. To see how a custom project is run, see the custom web system page.

Talk about the project