Friday, 7:40 pm. The release is out.
A backend developer opens Postman, hits Send on his seven favourite requests, sees seven 200s and types “API is fine” into the chat. On Saturday morning checkout fails for every third customer, and “API is fine” becomes a screenshot your manager forwards with the caption “how?”.
Sound familiar? Then sit down. This will not be a pleasant conversation.
You did not lose Postman. You never had it — not in the sense you thought. You had an account, three collections built by an engineer who quit last summer, and a ritual: poke the requests by hand before each release and call it “API tested”. That is not testing. That is fortune-telling with status codes.
Meanwhile Postman now forces a cloud account, charges per seat and keeps raising prices. And here is the thing you will resent me for saying: this is the best thing that ever happened to your testing. Because now you have a reason to replace the approach, not just the tool.
What to replace Postman with
If you need a one-to-one replacement — a client for sending requests by hand — any open-source Postman alternative will do. If you need a replacement after which API checks actually run in every regression instead of living in someone's head, use TestManager: you import an OpenAPI/Swagger spec or add methods just like in Postman, but you write checks in plain words. The system does not need your JSONPath — it maps the fields itself.
Why “we have Postman” is the most expensive phrase in your team
Let's count, since you are reading anyway.
Say your API has 80 methods. How many actually get checked before a release? Be honest. Five. Seven at best — the ones everyone remembers. The other 73 methods ship to production on the “a customer will let us know” principle.
The customer lets you know. The customer always lets you know. The bill just arrives later — as churn, refunds, a midnight hotfix and a developer who spends Sunday hating you, the company and the profession. One missed bug in a payment API costs more than a year of any testing tool. You know this. You just prefer not to think about it on Fridays.
And here is the important part: Postman is not to blame. It is an honest tool for sending requests. A hammer. A good hammer. But a hammer does not build the house — and a request collection does not test an API. Testing is a check that:
If a check has to be run by hand, you do not have a check. You have hope with a user interface.
Three excuses I have heard a hundred times
“We have collections, there are even tests in them.” Open the Tests tab. There is JavaScript in there written by one person. He quit. Now those scripts are archaeology: too scary to touch, too dear to delete, nobody remembers what they check. A collection people are afraid to open is not a test suite. It is a sarcophagus.
“We will set up Newman and CI next sprint.” That “next sprint” is a year and a half old. I have seen teams where the “wire up Newman” ticket survived three re-orgs and two team leads. It is immortal. It will outlive you too.
“Manual checks are more reliable, a human sees more.” A human sees the 200 and closes the tab. He did not read the response body — it could have been "items": [] instead of data, and nobody would know until Monday. Humans get tired, humans rush, humans have Fridays. An automated check does not have Fridays.
What API checks in plain words look like
Down to business. Here is how it works in TestManager — and note that I am not promising magic:
Catch the real difference. It is not about buttons or “nicer UI”. It is about who can write checks. In the Postman setup it is the one person who knows how $.data[0].id differs from $..id. In TestManager it is any manual tester from day one. You just multiplied the number of people able to cover your API by your whole QA team. For free.
Postman vs TestManager: a table without marketing
| Capability | Postman | TestManager |
|---|---|---|
| Send a request by hand | Yes, best in class | Yes |
| OpenAPI/Swagger import | Yes | Yes |
| Checks | JSONPath / JS scripts | Plain words |
| Who writes checks | Whoever can code | Any QA |
| Running in regression | Newman + CI, a separate project | Out of the box, next to UI tests |
| Team reports | Build your own | Out of the box |
| Pricing | Per seat, in USD, cloud account required | Flexible, pay for what you use |
When Postman is still the better choice
No fairy tales — they cost everyone too much. If the tool is for a developer — debugging webhooks, poking GraphQL, playing with gRPC, testing a quick hypothesis — take Postman or an open-source alternative and live happily; I even envy you. TestManager solves a different problem: making API checks exist in the regression, not in a backend developer's head or a collection sarcophagus — and run even when the whole team is on vacation.
The developer gets a hammer. The team gets a conveyor. Mixing those two up is the mistake that costs you the most.
FAQ
Why are teams leaving Postman?
Per-seat pricing that grows with every hire, a forced cloud account for most features, and free-tier limits that stop fitting a team. And the quiet realisation that collections still do not run by themselves.
What replaces Newman for running collections in CI?
Nothing — in the sense that you no longer need it. In TestManager API checks run from CI/CD together with the whole regression and land in one report. The runner-config-parse-the-output side project stays in your past life.
Do I need to know how to code?
No. You phrase the expectation in words — “expect 200 and a non-empty list of orders” — and the system maps the fields and builds the check itself.
What about my existing Postman collections?
Got an OpenAPI spec? Import it and rebuild the checks in plain words. Takes an evening. As a bonus you will discover that half your collections were checking things your API no longer has.
What if we have no OpenAPI docs?
You can add methods by hand, like in Postman. But if there are no docs at all, you have a bigger problem than tool choice. The good news: now you know about it.
What to do right now
Do one thing. Not “adopt a platform”, not “migrate the process” — one thing. Take the method that keeps you up at night: auth, payment, cart. Import the docs. Write three checks in plain words. Put them into the regression. Twenty minutes on the clock. If after the first automated run you want to go back to hitting Send by hand — go back, Postman is not going anywhere. So far nobody has. Try for free →
P.P.S. Six months from now you will either be reading a regression report over your morning coffee, or still hitting Send before releases and hoping. The tools already exist, both scenarios cost about the same money. The only difference is who finds out about the bugs first — you or your customer. The choice, unfortunately, is yours.