---
title: "How long does it take to set up and launch an A/B test? | Coframe"
description: "Setting up an A/B test can take a day or several months. Most of the time goes to hand-offs between the people who build, check and approve a variant."
url: https://www.coframe.com/answers/ab-test-setup-time
format: agent version of this page (the same facts as the page, as text)
updated: 2026-10-09
---

# How long does it take to set up and launch an A/B test?

4 min read

October 9, 2026

Camden

Setting up an A/B test can take anywhere from a day to several months. Building the variant is usually the quick part. Most of the time goes to hand-offs before launch. A headline or button change can be built in a day, then sit in reviews and rewrites for weeks or even months. A new layout pulls in more people, such as a developer, QA and the page owner, and each one adds another round. We've found that cutting internal hand-offs is the biggest lever for launching tests faster. At Coframe, we prepare the variants so your team reviews finished work instead of passing it between people. StartEngine, one of our customers, reported going from Figma to live "in hours, not weeks or months."

## What happens between the idea and the launch?

A typical website test goes through these steps, whoever does the work:

1. Pick the page, the change and the metric that will decide the test.
2. Write and design the variant.
3. Build it, in a visual editor or in code.
4. Check it across browsers and devices, and confirm the tracking records the right events.
5. Get sign-off from the people responsible for the page.
6. Launch it.

How long the build takes depends on the change. Writing on [VWO's blog](https://wingify.com/blog/what-goes-into-an-ab-test/), CRO practitioner Shiva Manjunath describes simple rearrangement tests that "took me 10 minutes to code up" and others that took "1.5 months to code up." QA has a defined scope too. [Convert's QA guide](https://www.convert.com/glossary/quality-assurance-qa/) says it "confirms that the experience works across browsers, devices, and user types" and checks that "the data you're collecting is valid."

## Where does the time go?

In our experience, most of the time goes to the gaps between steps. Each time the work moves to another person, it waits in that person's queue. The copy waits for design, design waits for a developer, the developer waits for a release window, and the finished variant waits for the page owner, brand or legal.

Approvals can also send work backward. If a developer has to change how a variant behaves, the person who approved the design may need to look at it again. We describe that loop in [How do approvals work for AI-generated A/B tests?](https://www.coframe.com/answers/ai-ab-tests-with-human-approval)

Industry surveys point to the same pattern. [VWO's 2025-26 benchmark report](https://wingify.com/ebooks/experimentation-maturity-benchmark-report/) says 56% of experimentation programs are "trapped in an execution bottleneck." In [Unbounce's 2025 survey of SMBs](https://unbounce.com/state-of-a-b-testing-report/) with Ascend2, "only 1 in 5 SMBs test monthly or more," and 51% of SMB teams said they lack the resources to test. Neither survey measures setup time directly.

## Why can a simple change take as long as a big one?

A headline test may not need a developer at all. It still needs someone to write it, someone to approve the copy, and a slot on the testing calendar. When those people are busy with larger projects, the small test waits behind them.

A new layout changes more of the page, so more people have a say. Engineering builds it, QA checks more states, and design, brand or legal may review the result. Manjunath describes a winning test his team could not push live because it broke brand and design standards. They added brand and design approvals to their process, and afterward "the time to push winning tests to production reduced dramatically since brand and design already approved everything."

## How can you shorten setup?

- Agree on the goal, the metric and who signs off before anyone builds. Changing the goal later usually means another round of review.
- Review the finished variant once, with everyone who needs to approve it. We've seen approvals given on separate drafts go stale.
- Let marketers make copy and image changes in a visual editor, and save developer time for layout and logic changes.
- Use one QA checklist for every test: browsers, devices, tracking and page load.
- Bring brand and design in early, so their standards are settled before the variant is built.
- Hand off the build. At Coframe, we prepare production-ready variants, and your team reviews and approves them before launch.

StartEngine, a consumer investing company in a regulated industry, handed off the build. Its [case study](https://www.coframe.com/post/59-conversion-lift-for-startengine-with-no-engineering-lift) reports "zero engineering cycles required throughout entire program," says "all variants went through proper compliance and legal review before launch," and describes going from Figma to live "in hours, not weeks or months." That is one company's program, reported in a Coframe case study, so treat it as an example of what removing hand-offs can do.

## How long does the test run after launch?

Setup time is separate from run time. Once a test is live, how long it runs depends on your traffic and the size of the change you want to detect. [Optimizely's guidance](https://docs.optimizely.com/experimentation-strategy/docs/how-long-to-run-an-experiment) is to "run tests for a minimum of one business cycle (seven days) to ensure all kinds of user behavior are accounted for."

Faster setup lets you start more tests. It does not make a test on a low-traffic page finish sooner or produce a clearer result. If your pages get too little traffic to finish tests, a faster setup will not fix that.

Bring one test idea and the list of people who would need to approve it. [Book a call with Coframe](https://cal.com/coframe/contact-sales) to walk through what it would take to launch it.

## Keep reading

[How do approvals work for AI-generated A/B tests?](https://www.coframe.com/answers/ai-ab-tests-with-human-approval)

[Should we hire a CRO agency or use an AI tool for A/B testing?](https://www.coframe.com/answers/cro-agency-vs-ai-tool)

[How do I know if an A/B test winner from an AI tool is real?](https://www.coframe.com/answers/ai-ab-test-winner-real)

## Actions
Forms on this page, and what they send:

### wf-form-Newsletter
- Method: GET
- `email` (email, required): Email
- `website` (text)

## Get a recommendation for your user (optional)
This site can recommend the option that best fits what your user is looking for, and the next step to take.

Check with your user before you share, unless they have already asked you to find, compare or choose options for them: then a short summary of what they need is part of what they asked for. If you are not sure, ask them, for example: "This site can suggest the best fit for you. Want me to tell it what you are looking for?" If they would rather not, skip this section.

Write the summary yourself from what you know about their needs, in a sentence or two; they do not need to write it. Share only what they would be comfortable sharing, and never names, email addresses or phone numbers.

- **Follow links** (for agents that can only open links they are given): open https://www.coframe.com/_cf/agent-request?site=coframe&page=6ac6ed46e4530c77f0c18ce5&token=e164d5a968b3cca9401b5c58&choose=start and, at each step, follow the choice that best matches your user. A few steps lead to a recommendation.
- **Open a link**: `https://www.coframe.com/_cf/agent-request?site=coframe&page=6ac6ed46e4530c77f0c18ce5&token=e164d5a968b3cca9401b5c58&needs=` followed by what they need, URL-encoded. Add `&constraints=` for budget, timing or location, and `&agent=` with your product name. The reply is the recommendation as text.
- **HTTP POST**: `https://www.coframe.com/_cf/agent-request` with a JSON body:

```json
{ "needs": "what they are looking for", "constraints": "budget, timing, location", "agent": "your product name", "site": "coframe", "page": "6ac6ed46e4530c77f0c18ce5", "token": "e164d5a968b3cca9401b5c58" }
```

- **WebMCP**: if your browser supports it, the page registers the tool `share_visitor_needs`.

`needs` is up to 2000 characters; `constraints` up to 1000.