[Update: Addressing new level system questions and feedback]

3 hours ago, logoflow said:

the key metrics are also determine a score relative to other freelancers so based on that what I can understand is success score is a competition or can say it's clear comparison with the other freelancer.

I got the same response from CS as you attached and learning about the comparison part of the scoring makes me more worrisome because of my specific situation (but not unique on this platform).

As a higher-ticket freelancer with less frequent but longer orders, I'll be compared against sellers whose orders are "smaller" / shorter but much more often than mine. So I'll have much fewer opportunities to influence my score at a given period (if I only knew specifically how / what to improve).

[Important Updates: Level System (now live!) and Ratings & Reviews (testing changes)]

12 hours ago, voicedbyken said:

Just reread the Success Score page and from what I can gather [without it being explicitly stated] is that they are using a Grammarly-style ai to monitor and review communications, gig descriptions, etc... to determine the mood, context, and other factors in both the seller and buyers communications. This is why many of us who are borderline over-communicative are seeing strikes in our success rating for communication issues. The algorithm does not understand human speech in the same way actual humans do, so it makes a generic interpretation that does not align with what is actually going on.  

My thoughts exactly. 

I see more issues with the AI. 

Sometimes part of the communication goes outside Fiverr eg. because the buyer cannot bring their whole team here, so Slack / Google Meet / others are used. 

This means that the AI doesn't see the full picture of communication and if for example some topic was resolved during outside-Fiverr communication. 

 

The other thing is that non-native English speakers might use some expressions that come off as less polite than they think / would like. They might not know it their whole life because the other side understands the situation and the original intention, so there are no negative consequences and no feedback.

In practice, this isn't a problem, but AI doesn't get that.

[Important Updates: Level System (now live!) and Ratings & Reviews (testing changes)]

3 hours ago, Lyndsey_Fiverr said:

We apologize that you've noticed a stagnant state in regards to how to improve these indicators, or that the indicators have misjudged your interactions.

So you are admitting here, that the algorithm / AI can misinterpret some data (or human interactions) and the scores can be wrong / unreliable?

Does the Customer Support or other internal team have means / tools / rights to manually change the scores / metrics?

If so how manual re-scoring would be assessed? What factors would be considered?

[Update: Addressing new level system questions and feedback]

11 hours ago, Kesha said:

It’s important to emphasize that the displaying of these scores and metrics does not signify a sudden change in how your marketplace performance is evaluated. Instead, it represents a move towards enhanced transparency, providing you with deeper insights into the underlying evaluation processes.

How rating in private is transparent?

How I can learn from feedback that I cannot see?

Elusive rating scoring system with tips for better yet non-measurable in practice communication behavior is not a way for people to grow and provide better service.

[Important Updates: Level System (now live!) and Ratings & Reviews (testing changes)]

4 hours ago, Lyndsey_Fiverr said:

As for how to get back on track, for now, I would strongly encourage you to reach out to your Seller Plus Customer Success Manager. They can help you pinpoint any issues that the system may have flagged and give you next steps for improvement. 

So if someone doesn't have 'Seller Plus' and by extension no 'Success Manager', then they cannot get feedback about what actually (specific order, messages, etc.) affects the specific gig's score in the specific metric, and what specifically to improve?

Fiverr acts like: "You did something bad. You should know you did. We know you did. Our system knows what it was. But we won't tell you. Here is a long list of subjective, non-measurable communication behaviors that could cause your score." 

My gig has the 'FIVERR'S CHOICE' badge, I have 5/5 star reviews, and never had conflicts / disputes, and yet this is something I have "Room for growth". Customer Support forwards me copy-paste responses that gets me nowhere. 

How a month is enough to get people ready for a change of this caliber? 
How we can improve our "new statistics" in a completely new scoring system in a month when the algorithm takes into consideration a much bigger time scope?