LaunchDarkly provides a feature management platform that enables DevOps and Product teams to use feature flags at scale. This allows for greater collaboration among team members, and increased usability testing before full-scale feature deployment.
N/A
Optimizely Web Experimentation
Score 8.7 out of 10
N/A
Optimizely Web Experimentation empowers teams to conduct experiments (without having to rely on developer resources) in order to test various user interactions, make website changes backed by data, and personalize customer experiences.
LaunchDarkly stood out to us because it put control of the application within the hands of our engineers. We didn't want to allow business users to manipulate the production site via a third-party tool. Instead, our focus was on delivering faster as an engineering team.
If a new feature should be added but unsure of how it will actually work or how users will accept the new enhancement or change, this tool allows you test and measure initial results. This saves so much time and energy knowing the results before it is deployed and might have low user adoption or acceptance.
This is our second testing platform, and is by far the best we have leveraged. Its intuitive in how you are able to build tests. Tracking is easy to set up within the visual editor or as a capability in URL tracking. The reporting tab does need more filtering capabilities for custom metrics.
A/B or Multi Variant Testing as a methodology to gather insight from customer usage. Experimentation as a feature within LaunchDarkly offers information around the success of one variant over another and whether the experiment has reached statistical significance.
Being able to decouple deployment of code from the release of a feature is hugely valuable.
Development teams are empowered to manage features within their production applications for reliability or testing purposes.
Hard to use extensions without some knowledge of HTML/CSS
Inconsistent user counts between GA & Optimizely when it comes to calculating MAUs, hard to be able to forecast budget and overages when our source of truth (GA) differs from our testing platform (Optimizely)
Impressions model doesn't support scaling personalization experiences - hard to run a serious data driven testing program when you have to cut tests short before 14 days (ideal length to get to any statistical read) in order to save on impressions
Because it's an incredible and essential tool for my line of work as a conversion optimization specialist. Really couldn't do my job nearly as effectively without it. It's paid for itself many times over and I feel like I'm only beginning to unlock the tools potential.
It's very easy to create new feature flags and set them properly. It is more difficult to get LaunchDarkly integrated within a distributed system so that flags can be used. Especially on stateless servers where gating features by user is not easy. Overall though, it is very easy to get started and I like how simple it is to use.
Usability is mostly great. I like the WYSIWYG functionality and adding in real code is simple as well. It's easy to target specific pages or audiences. I've knocked a couple of points off because of how difficult it is to set up URL redirect experiments, confusion around creating pages, and lack of data that can be further analyzed.
I would rate Optimizely Web Experimentation's availability as a 10 out of 10. The software is reliable and does not experience any application errors or unplanned outages. Additionally, the customer service and technical support teams are always available to help with any issues or questions.
From what I have seen, LaunchDarkly integrates well with your code and also services you might have in your tech ecosystem. We use Jenkins for automation and we were able to use it to build pipelines to automate the control of LaunchDarkly toggles in our code.
I would rate Optimizely Web Experimentation's performance as a 9 out of 10. Pages load quickly, reports are complete in a reasonable time frame, and the software does not slow down any other software or systems that it integrates with. Additionally, the customer service and technical support teams are always available to help with any issues or questions.
They always are quick to respond, and are so friendly and helpful. They always answer the phone right away. And [they are] always willing to not only help you with your problem, but if you need ideas they have suggestions as well.
The tool itself is not very difficult to use so training was not very useful in my opinion. It did not also account for success events more complex than a click (which my company being ecommerce is looking to examine more than a mere click).
In retrospect: - I think I should have stressed more demo's / workshopping with the Optimizely team at the start. I felt too confident during demo stages, and when came time to actually start, I was a bit lost. (The answer is likely I should have had them on-hand for our first install.. they offered but I thought I was OK.) - Really getting an understanding / asking them prior to install of how to make it really work for checkout pages / one that uses dynamic content or user interaction to determine what the UI does. Could have saved some time by addressing this at the beginning, as some things we needed to create on our site for Optimizely to "use" as a trigger for the variation test. - Having a number of planned/hoped-for tests already in-hand before working with Optimizely team. Sharing those thoughts with them would likely have started conversations on additional things we needed to do to make them work (rather than figuring that out during the actual builds). Since I had development time available, I could have added more things to the baseline installation since my developers were already "looking under the hood" of the site.
LaunchDarkly stood out to us because it put control of the application within the hands of our engineers. We didn't want to allow business users to manipulate the production site via a third-party tool. Instead, our focus was on delivering faster as an engineering team.
Overall, the tools we compared against were great, but we went with Optimizely because it has all the features we needed and has the market leadership that gave us trust we would be successful in our experimentation efforts.
While granted, I am not the one making the changes to deploy the software; it seems basically as simple as placing the optimized snippet on your site. If you place it in the header, it should reach across your entire site easily. I do think the test setup can be a bit temperamental at times, and having too many cooks in the kitchen can lead to errors. So, I would caution how many people have the access needed to make actual changes.
Improved developer experience with some teams moving to Trunk-based Development.
Increased deployment frequency due to smaller code releases.
Validation of the technical and business value of work is achieved more quickly through smaller pieces of work and through experimenting with a small group of users before a feature gets to 100% of customers.
Customer retention: We've reduced subscription service client churn by 20%+ using optimized unsubscribe flows.
Risk mitigation: Testing into full site redesigns has saved clients millions of dollars.
Feature prioritization: Identifying what painted door changes add value has allowed developers to focus on changes that add hundreds of thousands or even millions to the bottom line.
