Amplitude Web Experimentation Reviews & Product Details

The biggest strength is how seamlessly Web Experimentation fits with Amplitude Analytics. Having experiments and analytics in one place makes it much easier to track conversion and revenue by experiment cohort, without having to stitch data together across multiple tools. Setup and launching experiments feel straightforward and low-friction. The WYSIWYG editor is a standout: it’s intuitive and powerful enough that I can run quick experiments myself without pulling in developers, which speeds up iteration and keeps engineering focused on other work. Review collected by and hosted on G2.com.

The WYSIWYG editor can feel clunky or get out of sync at times. When that happens, you often have to dig into the edit history and remove or clean up steps before the experiment behaves as expected. It’s usually fixable, but it adds unnecessary friction.

A bigger limitation is the lack of support for holdout groups. Being able to run web experiments easily is a big plus, but without proper holdout capability, the rigor and long-term value of experimentation are limited. That’s a real drawback for teams that care about clean measurement and true incrementality. Review collected by and hosted on G2.com.

I like the Visual Editor and Customizable Metrics in Amplitude Web Experimentation. It's great that our Project Manager can set it up easily without needing help from developers. Review collected by and hosted on G2.com.

I have set up Amplitude Analytics and Experiments with the proper deployment key, but the network response from Amplitude was an empty object {}. The concept is misleading with the Experiment Feature Flag. I was confused about what I should use for A/B testing. I had to use a feature flag with adding an event in the end for A/B testing. Better to rename the menu or give a clear guideline according to the user's purpose. The initial setup was not easy because of the confusion I mentioned. Review collected by and hosted on G2.com.

What I like most about Amplitude Web Experimentation is the ease of making changes to page content with the visual change tool. It's quite easy to use, although I also feel that it is somewhat limited. Another thing I find very interesting is the ability to customize who to show the experiments to, which I consider to be a standout feature. I also find the content editing part directly within Amplitude interesting, although it is true that it has its limitations. Review collected by and hosted on G2.com.

The first thing that comes to mind is the support, it hasn't been entirely good. With the plan I have, I don't have live support, which makes it difficult to solve problems or understand why things fail. Additionally, there is a 'variant jumping' issue, it affects the data and they don't give us a clear solution in support. Also, when I make visual edits to the content, if I include new content, it is reflected in the "control" which doesn't make sense. Finally, when doing more complex tests and moving things around in the structure, everything starts to break and it works quite poorly. Review collected by and hosted on G2.com.

The integration with Amplitude analytics lets us use metrics that are already available and widely understood across the business, while also giving us the flexibility to create new metrics as needed. Review collected by and hosted on G2.com.

The experimentation features aren’t always intuitive. For example, to set up an A/A test we currently need to add a console log to the variant so that Amplitude will serve traffic to both variants. Otherwise, it only recognises the control and won’t split traffic equally, even when the test is set up that way. It would be great if there were an A/A test toggle to bypass this.

There also isn’t an option to set up a heatmap for the experiment, which would be really useful to view alongside the experiment results.

Finally, despite obtaining the Amplitude Specialist – Experiment certificate, I don’t feel I was prepared to use the platform. There are a lot of scenarios and edge cases that would have been helpful to cover in the certificate, but they sadly weren’t included. Review collected by and hosted on G2.com.

I like the ease of setting up the dashboards for the results, they are quite easy and any product manager, even non-technical profiles, can do it. Also, the initial setup was simple and the Amplitude team helped us. Review collected by and hosted on G2.com.

The methodology used by Amplitude Web Experimentation to assess the statistical significance of a test is a bit slow. For businesses that need to conduct very quick tests, this is an inconvenience, as sometimes the data volume is not sufficient. It would be interesting to have a more dynamic mode, even though it might be riskier due to less volume and possibly less reliable results, but that allows obtaining results before waiting a long time. Review collected by and hosted on G2.com.

I really like how easy it is to set up and use. It walks you through a really structured setup of the experiment to make sure you've considered every variable necessary to make it a success. We also use Amplitude for product experiments, so it allows our product and marketing teams to see what each other is working on at any given time. Review collected by and hosted on G2.com.

I wish you could more easily move web components around. We decided to go with Amplitude for this because the A/B test add-on for our website platform was too expensive, but this really doesn't go as far as we need it to in order to really make it the best possible tool for us. It's pretty limited to edits that you can make within the existing structure. Review collected by and hosted on G2.com.

The idea that the testing platform is connected directly to Amplitude analytics Review collected by and hosted on G2.com.

The fact that the testing platform isn't allowing deep dive analysis that leverage the existing reporting feature of amplitude analytics and we have experienced a lot of data issues misleading us into the wrong outcome - the customer support that hasn't been very reactive and our account manager being very dismissive about it ("it is free what do you expect"?). The testing tool despite being integrated in Amplitude very much feels like it is a 3rd party tool that was grafted onto the analytics platform but where the integration wasn't full completed. The tool is not mature enough to compete against the big players on the market and the Amplitude team have very much relied on their customers to be the guinea pigs. I believe it has strong potential however it will require some serious effort to make it a reliable and strong contender Review collected by and hosted on G2.com.

The flexibility it offers along with the ease of creating experiments Review collected by and hosted on G2.com.

Limitations of adding more custom events and metrics Review collected by and hosted on G2.com.

Convenience and intuitiveness as the platform is connected to analytics and all other blades within the same UI.

Easy to get up and running with WYSWYG to setup web experiments with low lift from engineering side.

The agentic capabilities allow clients to setup more tests running at the same time and at a higher frequency.

Seamless integration with Cohorts makes it easy to have data continuing between capabilities within Amplitude. Review collected by and hosted on G2.com.

Its new(er) to amplitude product engineering team so product support takes longer than analytics support. Flicker issues still remain on clients and that puts it technologically weaker compared to competing experiment platforms. Review collected by and hosted on G2.com.

It takes away the development time, and opens up and all range of possibilities potentially that are in the hands of the team. It's potentially so easy to set up and launch a test. I find one of the strongest use cases with copywriting team. We always have endless discussions about whihc copy to use, low-code experimentation removes this and allow us to simplytest our hypothesis. Review collected by and hosted on G2.com.

* the tool is affecting our performances on mobile especially, this forced us to limit the use of Web experiments for mobile users, where we have most of the traffic

* we have languages subfolders, so far I haven't found a way to address this issue: basically I need to set up a new experiment for each languages (therefore URL), this reduces A LOT the chances of having enough traffic

* the interface is not yet perfectly functioning, it's not immediate to select items and understand if the changes was actually applied Review collected by and hosted on G2.com.