Recommendation, personalization and optimization... made easy
June 22, 2017
Recommendation, personalization and optimization... made easy
Score 9 out of 10
Vetted Review
Verified User
Overall Satisfaction with Dynamic Yield
Dynamic Yield is the platform we choose for adding personalization and optimization to our e-commerce. Regarding optimization, we developed a set of new features that we wanted to test from a business perspective, to understand how much they would have impacted the user behavior. Basically, we used Dynamic Yield for A/B testing a few solutions.
Another feature we used a lot is the recommendation engine: we add the recommendation widget in a number of pages, adding more engaging content for the users, and testing different strategies, like similar products, most viewed, bought together, and so on. In our company, the solution hypothesis is prepared by marketing team and IT team, while the implementation and future analysis are run by IT team.
Another feature we used a lot is the recommendation engine: we add the recommendation widget in a number of pages, adding more engaging content for the users, and testing different strategies, like similar products, most viewed, bought together, and so on. In our company, the solution hypothesis is prepared by marketing team and IT team, while the implementation and future analysis are run by IT team.
- What we love the most about Dynamic Yield is how easy it is to create experiments that let us compare 2 or more variations. The feature of targeting the right users/pages/context for the experiment cover pretty much everything we wanted and the process of setting it up is very easy.
- The recommendation engine is another feature that we love. It's very easy to try different strategies to understand what performs best in each case, and using the templates you can save tons of work to set up new recommendation experiments (yes, every feature is an experiment from now on). Moreover, the direct revenue and assisted revenue metrics gives great insight on what is just holding the user on the site and actually shows what users wanted to buy (both are great results though).
- The possibilities to define goals based on custom javascript events made the platform very flexible and lets you create custom goals (different from the abused conversion rate or click through rate) that can be used by the engine to optimize the performance or to define the winner in an A/B test.
- One thing that we find limited is the rules you can set in the strategy for recommendation. Especialy the similar products one, has the possibility to define fields that must match and field that must not match: we'd like the configuration to be more flexible.
- Another thing we struggled with is the product feed. We have different price lists for different countries and we handle multiple languages. Representing all that information in one feed is pretty hard, and moreover, there's a strong limitation on the number of fields available for the feed. Basically, the product feed engine should be boosted.
- Talking about the reports, it would be amazing if the system would give the result of an experiment not only viewing it for all users, but even segmenting the audiences: in this case the system let you see the raw data, but you have no indication whether the test is performing good, whether you have enough data to take decisions, and so on
- If we were using just the recommendation widget, we would have had a ROI of 2.5x considering the whole investment on Dynaimc Yield.
- Testing messages on the cart, we saw a 7% uplift of conversion rate for that page. Moreover we discover that the message don't work at all in certain countries, and we are about to test other hypothesis for these countries
- Testing new ways to make the user subscribe to our newsletter (basically different UIs and positions) we incrementend the email subscription of 200%
- Monetate and Optimizely
First of all, the other tools we looked at was more focused on A/B testing and personalization but one of the that we loved immediately of Dynamic Yield was the recommendation engine: given that we were missing such a feature in our e-commerce, that really made the difference.
Another thing was about the people: Dynamic Yield staff looked more technical and results-oriented than their competitors, that in the first phase, was too focused on selling the product then proposing solutions.
Another thing was about the people: Dynamic Yield staff looked more technical and results-oriented than their competitors, that in the first phase, was too focused on selling the product then proposing solutions.