Three routes, and what they actually cost
The term "integration" actually refers to three different types of purchases, and most pricing confusion comes from comparing these options.
- A purpose-built connector app costs about $25 to $750 a month. These apps are made to connect two products you already use. For example, Shopify to QuickBooks connector apps run from $24 a month for up to 100 orders to $749 a month for 5,000 orders, depending on the app and your order volume.
- A general-purpose automation platform can be free or cost several thousand dollars a month. Examples include Zapier and Make. With these, you build your own workflows using prebuilt blocks.
- Custom development is handled as a project, not a subscription. This option makes sense when there is no existing connector for your needs, or when your data needs to be transformed in ways the platforms cannot handle.
Begin by looking for a connector app first. Many integration projects can be solved with a $29 a month connector that people simply did not know was available.
The bill escalates in a way the pricing page does not show
This often surprises people, but it is predictable once you understand how the pricing works.
Zapier's Professional plan costs $19.99 a month for 750 tasks and $3,389 a month for two million tasks, both billed annually. The plan and features are the same, but the price can be 170 times higher. Although the per-task rate drops as you use more, your total bill still increases as your business grows.
What often surprises people is what counts as a unit. On Make, an operation is one module run, so if a trigger returns ten records, every step after that runs ten times. On Workato, each connector action is a task, so syncing an order across six systems counts as six tasks. Your invoice is based on machine steps, not business events, and there are usually more steps than you expect.
There are two more details that are easy to overlook. Connector apps often charge extra for going over your volume limit, such as $15 for every 500 orders above your plan. Also, 'premium' connectors often require a separate license. For example, Microsoft's Power Automate included with Microsoft 365 only covers standard connectors. To use a premium, custom, or on-premises connector, you need Power Automate Premium at $15 per user per month. This is a clear example of a hidden surcharge that many people do not notice until they are halfway through building.
Most of the serious vendors will not tell you the price
Workato, Tray and Celigo do not list any prices on their product pages, and MuleSoft lists only a starting floor. Boomi only publishes one tier out of about fifteen. If you are trying to budget using their websites, you will not find the information you need.
There is a way around it that most buyers do not know: these vendors list on the AWS Marketplace, where contract prices have to be published. On 26 September 2026 that showed Celigo Standard at $36,000 for a twelve-month contract and a Workato enterprise workspace at $143,750. MuleSoft does publish a floor on its own pricing page, at $2,000 a month billed annually, with a $50,000 minimum annual commitment on one of its buying models.
These are list prices for enterprise tiers, not what a small business would pay. It is important to know these numbers for comparison. If you have a small company and someone suggests one of these tools, expect a high quote and ask for the price early.
| Vendor | What is priced | Published price | Published on |
|---|---|---|---|
| Shopify to QuickBooks connector | Monthly plan, by order volume | $24 to $749 a month | Vendors' own sites |
| Zapier | Professional plan, 750 tasks | $19.99 a month, billed annually | Vendor's own site |
| Zapier | Professional plan, two million tasks | $3,389 a month, billed annually | Vendor's own site |
| Microsoft Power Automate Premium | Premium, custom and on-premises connectors | $15 per user per month | Vendor's own site |
| MuleSoft | Starting floor | $2,000 a month, billed annually | Its own pricing page |
| MuleSoft | Minimum annual commitment, on one buying model | $50,000 | Its own pricing page |
| Celigo | Standard, twelve-month contract | $36,000 | AWS Marketplace; not published on its own site |
| Workato | Enterprise workspace | $143,750 | AWS Marketplace; not published on its own site |
| Tray | No plan priced | Not published on its own site | No price quoted on this page |
| Boomi | One tier of about fifteen | Only one tier published | Vendor's own site |
The cost nobody quotes: the other end keeps moving
Integration is not just something you build once. It is an ongoing relationship with another company's software, and they can make changes without telling you.
Every platform you might connect to states how long it will support each version of its API. These support periods are usually shorter than most people expect, and two of the most common platforms for small businesses do not make any commitment.
| Platform | Minimum support |
|---|---|
| Salesforce | 36 months |
| Microsoft Graph | 24 months notice |
| HubSpot | 18 months |
| Shopify | 12 months per stable version |
| QuickBooks Online | No published commitment |
| Xero | No published commitment, and no notice period |
Shopify takes it a step further and warns that any app using an unsupported version could be removed from its App Store. This means maintenance is not optional like a regular software update.
Often, you do not get any warning. A study of 1,368 real APIs found that out of 251 versions with breaking changes, 87.3 percent did not deprecate anything in the previous version, and only about five percent said when a feature would be removed. This is what a maintenance retainer really pays for: catching problems before your customers notice.
What doing nothing costs, without the invented statistics
Almost every statistic you see quoted here does not hold up to a source check. The well-known claim about bad data costing the US economy trillions comes from a marketing infographic with no real data behind it. The rule that says an error costs a dollar to prevent and a hundred to fix cannot be verified in the book people cite. I will not use either of these.
What is more reliable is data that has been measured, not estimated. Professional data entry operators make mistakes in about 0.34 percent of fields. Regular staff re-entering data from one system to another, which is likely your situation, made errors in 3.7 percent of entries. A 2026 study found that manual transcription had errors in 8.23 percent of fields, while direct system-to-system transfer had none.
This difference is the real point. Instead of relying on made-up dollar amounts, consider that copying between screens is about ten times more error-prone than typing carefully once. Integration removes this problem completely.
To estimate the cost for your own business, add up the hours spent, multiply by the total labor cost (not just wages), and include the cost of mistakes that reach your customers. This last part is usually the biggest cost, but it is rarely calculated.
When not to integrate
There are four main cases, and the first two account for most situations.
The volume does not justify integration. If someone is re-entering eight orders a week, that only takes about fifteen minutes. A $79 a month connector plus setup costs will not pay off, and there are better areas to improve.
If one of your systems will be replaced soon, do not integrate it. Building an integration for something you plan to replace within a year means you will have to do the work twice. Replace the system first.
If your process is still changing, wait before automating. Automating a workflow that is not finalized locks in the current version. Let the process settle, then connect it.
Sometimes, what you really need is a report, not a full integration. Many requests for systems to 'talk' actually mean wanting a single view of both. Scheduling an export to a spreadsheet is free and does not risk breaking anything.
What I charge
A Diagnostic Week costs $2,500, fixed. During this time, I focus on what moves between which systems, how often, by whom, and which of the three options is the cheapest that works. You receive a written scope, a plan, a fixed quote, and a clear summary of any remaining issues. You keep all of this, no matter what you decide to do next.
Builds usually cost between $15,000 and $50,000, based on the findings from the Diagnostic Week. However, I often find that a connector app you can buy right away solves the problem, and I have recommended this before. Fixing system connections is often the most cost-effective big improvement, especially since the alternative is usually a full replacement.