RFC-44: Marketplace Store Revamp

If they were trustworthy then yes.

However, getting reviews (the good once) is hard, users tend to disregard review requests unless the response of support was not quick enough for them, or they use the review section to reports “bug” then they so seem to know where to leave reviews.
It is also hard to trust the source, even with the new requirement that you must have the app installed.

If it was easier for partners to get reviews from actual users that can be checked by not only the partner but also any other user browsing the marketplace, then I would see value in keeping the review section.

As for install count, to me its not clear what this means. Is this an active counter that adds installation events and subtracts uninstallation events? What about instances that have the app installed but don’t use it? With paid apps not likely, but with free apps there is no financial trigger to remove the unneeded app thus bloating there install count.

Why not use actual Jira Service Management data? Partners get a free instance to support there customers, so use those portals. Get hooks into the service portals to show response times, quality of responses and resolution times. I would argue that this provides more value to browsing users then a review left x days/week/months ago no matter the number of stars.

And don’t use a simple installation counter, but use an active users counter. Its not the number of installations but the total number of users using the app that gives meaning. I would argue that knowing how many users use the app daily gives more context and value to browsing users then a install counter. And the data is already there since, Altassian knows where your app is installed and also knows the user tier for each of the installations.

Both these alternatives don’t need to be realtime data, there can be a gap of a couple of hours to a day or 2 and this would still bring more value in my opinion.

Thanks for bringing this up and looking for ways to evolve the buying experience for our mutual customers. The timeline is a bit tight on this RfC but many important things have been mentioned already.

The one thing I want to call at is that Atlassian needs to publicly document the criteria for search results ordering as well as personalization an recommendations. If not executed well this bears a high risk of individuals gaming the system (some of that is present on MPAC already for some time) in a way that hurts customer experience and app adoption. It’s not been stated but it feels like Atlassian is looking into a “security by obscurity” design which would be fatal. We need something like that in future as well: https://developer.atlassian.com/platform/marketplace/marketplace-search-results-and-rankings/

2 Likes

Hi @anton2 ,

  1. We are implementing a clickable banner at the top of the homepage to facilitate content promotion in the future.
  2. In the future, we will explore the possibility of adding collections categorized by industries.
  3. The collection tile has been enhanced to provide more comprehensive information. You can view the complete tile by clicking on expand view of new homepage design.
  4. This block is specific to certain types of collections only. Your feedback will be taken into account for future iterations of collection pages.

Hi @nathanwaters
We recognize the importance of showcasing new apps on Marketplace. To tackle this issue, we have already made adjustments to prevent duplicate appearances of the same app on the homepage. Furthermore, in Phase 2 of Project Aether, we aim to introduce personalized recommendations on the Homepage. This will help users to discover different apps customized based on their industry, previous installations, and search activity.

Note: Regarding the discussion about removing install counts and reviews, it’s worth noting that our current research indicates these elements significantly contribute to building trust among customers and serve as crucial indicators they consider before downloading apps. We will consider exploring different formats for these trust indicators in the future, wherein we will validate these modifications by running an experiment in FY25.

1 Like

Hi @dmitry.astapkovich
The product filter will be prominently featured on the search page and categories page, as illustrated below.

Even though we are showcasing these changes for now, we are also working on enhancing our search. This will solve some of the problems that you have mentioned.

Hi @MattiaLomazzo ,
Users would be able to select filters on the search results page.

While we are improving the search component along with providing search suggestions (auto-complete, previous searches, etc), currently the search algorithm is not changing. The results are ranked based on the relevance of the search term. Relevance is based on how well the search term matches the following app fields, in order of importance:

  1. App name
  2. Partner name
  3. Tagline
  4. Category
  5. Highlights text (latest version)
  6. Compatible Atlassian product(s) (e.g. Jira Service Management, Jira Software, Confluence, Bitbucket)
  7. App key

We are currently developing an NLP-based search algorithm, which will be launched as an experiment. Stay tuned for more details about the new algorithm’s release and when it will become GA.

Thanks for providing your detailed feedback. We will continue to refine the designs in order to further optimize them, and we will certainly take your suggestions into consideration.

The Featured collection will be rotated and utilized by the Marketing team to highlight specific use cases or events.

Regarding app tiles, our current tech stack has limited our ability to conduct experiments. However, with the new stack, while maintaining existing elements such as vendor name, ratings, and installs in app tiles, we are planning to initiate experiments and constantly iterate to enhance the user experience and make more efficient use of our touch points.

The Keywords will initially serve as use-case filters on both the Search and Categories pages within the new platform. Furthermore, they will also be integrated into search functionality by the end of CY24.

As mentioned in another comment, Company’s stats are not partner stats but overall Marketplace stats to build customer trust in the Marketplace.

The new categories and keywords would be part of the revamped store we were talking about in this RFC. We will do a controlled rollout of these changes and a few of our customers will see the new categories and keywords before they are available for GA.

Hi @UlrichKuhnhardtIzym1
We are unsure what you are asking.

How can vendors ensure that apps are listed correctly?

If you are referring to impacting the ranking in categories and collections, the search ranking algorithm will remain unchanged for now. However, it will be modified with the introduction of NLP-based search. At that time, we will communicate the new parameters for search rank to partners.

How is Atlassian MP implementing transparency w.r.t to listings and what is the feedback loop for vendors to influence the mappings?

If you are asking how we ensure that apps are listed correctly, we are currently cross-checking the apps that are in the collections. We will work on informing Partners about the mapping of their apps to collections in the future so that partners are made aware of any apps of theirs that will be featured in these collections. We will also work on establishing a feedback loop as we introduce Product-specific collections and other collections in the homepage, the details of which will be shared in the future.

Hi @rlander ,
Your assumption that these changes relate to marketplace website is correct.
We will inform the DC team of partner concerns regarding UPM and will try to get some of the changes done by FY25 :slightly_smiling_face:

1 Like

Thank you all for giving your feedback on the new Homepage and the insightful discussion that happened in this RFC.
We will maintain the current designs for the controlled rollout in May. Enhancing app discoverability will be an ongoing process, and we will experiment with some of the suggestions to optimize the user experience in our future iterations.

Two major concerns were raised in this RFC:

  1. Promoting the same old incumbents on the homepage: To address this issue, we have devised the following strategies:
    Personalization: This feature will present different apps to users based on their industry, company size, past installs, and searches.
    Search Rank: We will adjust the ranking mechanism of search to consider recent installs and churn numbers more prominently.
  2. Transparency related to search: While we are currently making changes to the search component, it’s important to note that the underlying algorithm for search remains unchanged at this time. We plan to introduce NLP-based search in the future as an experiment and fully implement it when it outperforms our current search algorithm. Additionally, we aim to provide high-level details (signals/parameters) about search ranking without divulging specifics so that you can modify your listing accordingly.