---
title: "How Coframe Finds Winners 90% of the Time | Coframe Blog"
description: "Understand how to optimize your site using Coframe."
url: https://www.coframe.com/post/how-coframe-finds-winners-90-percent-of-the-time
format: agent version of this page (the same facts as the page, as text)
updated: 2026-10-06
---

### Traditional experimentation finds winners ~15% of the time. We, on the other hand, find winners 90%+ of the time. Here's how.

UX research has traditionally been dominated by A/B testing: incoming users are randomly shown one of two versions of a website or product for some period of time. This approach allows marketers to compare the performance of the two options and determine which performs better according to some metric, like conversion rate. However, studies from experimentation platforms like [Optimizely](https://www.optimizely.com/company/press/the-experimentation-experience-report/), [VWO](https://vwo.com/blog/cro-industry-insights/), and [Convert.com](https://www.convert.com/blog/a-b-testing/ab-testing-stats/) have shown consistently that winners, or variations that out-perform the original, are only found ~15% of the time.

There's another issue. While A/B testing effectively measures uncertainty and produces statistical data on which option is superior, it’s limited to comparing options over a fixed, often lengthy, testing period. During this time period, you're sending a lot of traffic to the losing variation(s), missing out on conversions. Fortunately - there's another popular optimization method that solves for this: multi-armed bandits (MABs). MABs continuously test several “arms”—like different variants of a signup button or several versions of copy text—while tracking their performance. As users visit the site, a higher proportion are served variants with higher success rates, while others may still be shown new or slightly lower-performing variants to gather data on their performance. While this provides a noticeable improvement over traditional A/B testing, it still requires a human to generate variants to test and may miss potentially better options.

We take this a step further. Here’s where Coframe’s approach differs.

### Rather than manually-created, static variants, Coframe combines the power of generative AI, business context, and our unique *Optimizer* algorithm to constantly optimize your website.

Coframe continually tests a broad range of options using our [Optimizer algorithm](https://docs.coframe.com/optimization/optimizer-methodology); dropping low-performing arms and using generative AI to recommend new, potentially superior variants, increasing the likelihood of finding the optimal option. Note that **generative AI is not acting without permission or supervision on your website**; you have full control to approve variations, but AI helps to accelerate the variation creation process significantly. You or your team can also add variants manually; we simply use AI to accelerate the process of creating variants. This method still quantifies uncertainty—so it answers the same questions as traditional A/B testing—but *optimizes* as it performs its experimentation. What’s more, in an era where personalization is king—think of your tailored Amazon or Netflix homepage—this approach can also take in user-specific contexts, like their demographics or prior interactions, to not only find the best variants but also tailor them to individual users.

### How can you leverage this? It’s simple.

Start by choosing a portion of your website you’d like to optimize—perhaps the copy text on your landing page or the UI on your demo page. Next, define a metric for Coframe to optimize, like clickthrough rate. Then, sit back and let Coframe work its magic. It’ll generate new arms for you and continually test them on your incoming traffic. As users interact with these variants, Coframe iterates the process, dropping and adding new arms based on what works and what doesn’t.

In addition, it might perform *segmentation*, or partitioning traffic based on data such as demographics or geographic location. Then, we can look at what works best for particular user segments and also track the outcomes of the overall *frame* (the experiments on all segments). To evaluate how well this works, we might look at both *experiment-level holdouts*, which measure how much improvement specific experiments provide, and *global holdouts*, which look at how much improvement the aggregate of all experiments have provided since the beginning.

But Coframe’s optimization approach extends far beyond just tweaking a piece of copy on your website. It can enhance many aspects of your online presence, from crafting engaging emails and messages to optimizing and fine-tuning your entire website, adjusting ad campaigns over time, and aligning SEO keywords. Managing all these elements manually can be overwhelming, especially for startups and small companies. There are so many moving parts involved in a company’s online presence that it’s simply impractical to rely solely on human effort to monitor, adjust, and modify them.

Generative AI has delivered on its promise to forever alter how we work—why not let it work for you? With Coframe, you can rely on AI to handle the heavy lifting, leaving you free to focus on more critical strategic aspects of your business.

Blog

## Related articles

[View all](https://www.coframe.com/blog)

[Customer StoriesAIGrowth Driving 410% Conversion Lift for Replit Coframe delivers incredible conversion lift for Replit "in an insane amount of time", running experiments and personalization across its home page, pricing page, enterprise page, and paid landing pages. Wins include 410% conversion lift for Replit's enterprise funnel with high-velocity experimentation.](https://www.coframe.com/post/driving-410-conversion-lift-for-replit)

[Customer StoriesGrowth +59% Conversion Lift for StartEngine With No Engineering Lift How we drove a whopping 59% conversion lift for StartEngine with no engineering cycles needed](https://www.coframe.com/post/59-conversion-lift-for-startengine-with-no-engineering-lift)

[AIPlatformCompany Coframe Develops UI Code Generation Model and Benchmark with OpenAI Read how you can generate visually-aligned website sections with a single prompt.](https://www.coframe.com/post/coframe-openai-ui-code-generation-model)

## 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=6ab8bd28961530e0aae1d67a&token=881ec72c88865eeadf9ed332&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=6ab8bd28961530e0aae1d67a&token=881ec72c88865eeadf9ed332&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": "6ab8bd28961530e0aae1d67a", "token": "881ec72c88865eeadf9ed332" }
```

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

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