Fiverr 3.0: Revisited (a.k.a. Fiverr 3.1)

[vickiespencer]

2 hours ago, miiila said:

I even moved a regular Monday "duty" that I had, to Tuesdays

I used to take every Monday off due to caring for my young granddaughter on that day. I hoped Tuesday would become my big day, but nope! Her mother could not change the granddaughter's daycare day, so now I depend on My husband to take over when I am busy with buyers.

This morning right on cue, the buyer messages started at 12:45 am even though I was asleep. Those are the messages that come from Europe.

[fiveroptic1]

Thanks @frank_d. So much help and as always, very interesting insights.  

[zeus777]

I can't believe I missed this post...I'm sorry Frank, I know I am late but happy birthday!!
I hope you had food, cake, drinks, singing and dancing, laughing, hugging and high-fiving, all the good stuff! 😄

[bilalhaider23]

Happy birthday frank....

A good post to dive deeply in fiverr working  and understanding fiverr system. 

[frank_d]

5 hours ago, zeus777 said:

I can't believe I missed this post...I'm sorry Frank, I know I am late but happy birthday!!
I hope you had food, cake, drinks, singing and dancing, laughing, hugging and high-fiving, all the good stuff! 😄

Hey @zeus777! Thank you!

Don’t worry about missing my post. I hope you find it helpful.

I had a lot of cake. It was a ridiculous amount. And now I am happy for 3 days straight. Maybe I am on to something…

[rabihumakhan]

Happy belated birthday, @frank_d. I missed such a great post.

I hope you had lots of cake and love with your loved ones. 

The ASP and briefs parts are interesting and I didn't think on those sides. 

However, after deleting my 'philosophy gig' (because I was tortured by students persuading, begging, and harassing me to do their works), sales seem to be rare for me. My average selling price is around 60$ and I dunno if Asp is one of the reasons of why I don't sell like I used to.  

[gajuseidi]

@frank_d I saw your post a bit earlier, but left it as a treat for later 😄

I really appreciate your posts as they not only contain invaluable insights for sellers of all sizes, but they also inspire us to push further that much more, no matter where we are at in our journey.

Without getting too much into the metrics of seller's performance, it's reassuring to know that if we deliver top-quality service as well as make sure our clients have a great experience, we are good to go.

I hope you had a fantastic birthday and looking forward to your new posts!🥳🎂

[frank_d]

On 3/24/2022 at 4:30 PM, gajuseidi said:

@frank_d I saw your post a bit earlier, but left it as a treat for later 😄

I really appreciate your posts as they not only contain invaluable insights for sellers of all sizes, but they also inspire us to push further that much more, no matter where we are at in our journey.

Without getting too much into the metrics of seller's performance, it's reassuring to know that if we deliver top-quality service as well as make sure our clients have a great experience, we are good to go.

I hope you had a fantastic birthday and looking forward to your new posts!🥳🎂

Thank you!

I did have a fantastic birthday celebration.

Working on a new post right now, this will be short and sweet, and will offer 100% actionable advice.

[kerrimax]

"Without getting too much into the metrics of seller's performance, it's reassuring to know that if we deliver top-quality service as well as make sure our clients have a great experience, we are good to go."

I believe the simple and basic things remain the same and is the key no matter how much Fiverr changes their system. They might end up with a system that stops functioning if they over do and tinker too much. The older system might have been much simpler to weed out the bad sellers/buyers. Just my thoughts here.

[frank_d]

19 minutes ago, kerrimax said:

The older system might have been much simpler to weed out the bad sellers/buyers. Just my thoughts here.

Actually, the old system was a playground for scammers, grifters and poor facilitators.

The new Fiverr version makes it easier to spot any seller who is offering a poor experience, not focused on the client journey or who is attempting to game the system.

Just take a look at how many moderately successful sellers post about their gigs going dead on a daily basis.

[proconsultn2035]

"Seems like the profile performance metrics are refreshed on a 60-day window, whereas the gig's performance metrics are calculated on a 90-day window."

Thanks for these tips. Great write up. Will try to follow them up and see their outcome. 

Fiverr 3.1, here we go!!!

[simiseo]

De-ranking is simple giving chance and space to new users nothing more. I think There is 5% metrics.

[frank_d]

29 minutes ago, simiseo said:

De-ranking is simple giving chance and space to new users nothing more. I think There is 5% metrics.

I don’t think that’s true. Fiverr doesn’t stop serving your gigs just so other untested sellers can have a go at it.

”5% metrics” is equal parts oddly specific and arbitrary.

I saw your thread about your gigs No longer being served and if I were you I would look into your past performance for the answers.

Also, apropos of nothing, Fiverr knows when certain transactions are iffy in terms of who places them and when.

[khafff]

That was a great post, thank you!

On 3/21/2022 at 1:23 AM, frank_d said:

Relevancy is calculated using a more complicated formula now. ASP (average selling price) on your gigs, is also now being calculated when determining where your gig is placed on search results, past performance matters more and is actually used to project future outcomes when Fiverr matches you with buyers, and this new Fiverr briefs feature is also trying to guide buyer's behavior and eliminate the need to search altogether. 

I know it may be a bit complicated to analyze, but if I send everyday cheap custom offers to one of my buyers (basically he needed tiny custom orders after the first default order), am I ruining my performance? Because therefore the average selling price is far lower than my basic package. But the buyer is returning every day, so that should be great for metrics... A bit confusing situation.

[frank_d]

2 minutes ago, khafff said:

That was a great post, thank you!

I know it may be a bit complicated to analyze, but if I send everyday cheap custom offers to one of my buyers (basically he needed tiny custom orders after the first default order), am I ruining my performance? Because therefore the average selling price is far lower than my basic package. But the buyer is returning every day, so that should be great for metrics... A bit confusing situation.

It is complicated. 🙂

ASP counts when a competitor is also performing roughly the same as you. In which case whoever has the highest ASP outranks the other.

On the other hand, repeat buyers are also a big part of performance metrics currently.

I can’t say for sure that this is hurting or helping. It just is what it is. 🙂 

[timir69]

On 3/21/2022 at 2:23 AM, frank_d said:

) The promoted gigs feature is not as simple as I initially suspected it to be. It's actually part of this new complex system that Fiverr has set up, as it is being fed with your gig's performance numbers constantly, in order to only promote the best gigs out there. It is also intertwined with the actual system that understands if you are relevant to what people are searching for.

So when you lose eligibility and can no longer promote a gig you used to be able to promote, this is an indicator of a "buyer satisfaction rate" slip-up, most likely on a gig performance level. By then of course it's too late to act upon it, but it is an indicator nonetheless.

Does this mean it can get the feature again? coz I think this pushed the whole profile.

[frank_d]

5 minutes ago, timir69 said:

Does this mean it can get the feature again? coz I think this pushed the whole profile.

Yes you can become eligible again once your gigs are performing well again.

[timir69]

2 minutes ago, frank_d said:

Yes, you can become eligible again once your gigs are performing well again.

I m facing for 30days, working with repeat buyers and not getting any new responses or any orders. The last time I get it back within 1 week and I found a few sellers in a forum saying they are not getting it back even in years which makes me worried.

[timir69]

59 minutes ago, frank_d said:

Yes you can become eligible again once your gigs are performing well again.

this is great hope 

[williambryan392]

I noticed the 'recommended' element got added to promoted gigs recently. The obvious insight is that you are more likely to be successful by promoting these gigs. Currently 4 of my gigs have recommended and 2 don't. So what I'm also taking from it is that I need to do better on the 2 that don't.

[frank_d]

4 minutes ago, williambryan392 said:

I noticed the 'recommended' element got added to promoted gigs recently. The obvious insight is that you are more likely to be successful by promoting these gigs. Currently 4 of my gigs have recommended and 2 don't. So what I'm also taking from it is that I need to do better on the 2 that don't.

Remember how I always talk about indicators?

This is a great example. Fiverr knows which gigs are performing well so they recommend you promote them to get the best out of your promotion.

[uk1000]

On 4/25/2022 at 8:05 AM, frank_d said:

I don’t think that’s true. Fiverr doesn’t stop serving your gigs just so other untested sellers can have a go at it.

Though customer support said:

Quote

...Also, the gigs are being repositioned on a daily basis, so that more users can be noticed...

Though they're probably more likely to move a lower performing gig further down in the results (eg. temporarily) to give newer gigs/users a chance than a higher performing gig I assume. So for a high performing gig I assume it's correct that "de-ranking" (ie. moving it to the lowest/very low page) won't happen/is unlikely to.

But since they need to show new gigs somewhere to give them a chance, how it works might be something to do with (or like) Thompson Sampling. See Fiverr video: https://youtu.be/_Or7r50wENQ?t=2281

Their Fiverr Data Drive Meet-up video has main section on Thompson Sampling.
They show a slide that says:

Quote

 

The Problem

* Crowded marketplace controlled by Top Rated Sellers

* New sellers hardly will appear on listing

* Thousands of new services each day

 

Then they show the benefits of using Thompson Sampling to help with that.
I assume they're saying - for newer gigs they don't have enough info about them to know whether the gigs are good enough (eg. get high enough conversion rate (clicks->orders)) - so don't know where to normally display them in the results.

So maybe they use Thompson Sampling (or something like it) to help decide the gigs to show where and how often so they can get enough data on them to see how good they are (click through rate, conversion rate etc.), and using that sampling should be better (according to their slide I think) than using a random element to help decide which new gigs to show in which positions (eg. until they get enough data on them).

eg. their slide seems to say they did 4 weeks of A/B tests of random vs Thompson Sampling (checking CTR, conversion etc.) and seem to be saying that the latter was better. I'm sure they'd still use all the other metrics/variables but this might be one other thing they use.

Though in their slides they don't mention gig/order price, but I'd assume they'd take that into account in the actual algorithms(s).

Edited April 28, 2022 by uk1000

[frank_d]

15 minutes ago, uk1000 said:

Though customer support said:

Though they're probably more likely to move a lower performing gig further down in the results (eg. temporarily) to give newer gigs/users a chance than a higher performing gig I assume. So for a high performing gig I assume it's correct that "de-ranking" (ie. moving it to the lowest/very low page) won't happen/is unlikely to.

But since they need to show new gigs somewhere to give them a chance, how it works might be something to do with (or like) Thompson Sampling. See Fiverr video: https://youtu.be/_Or7r50wENQ?t=2281

Their Fiverr Data Drive Meet-up video has main section on Thompson Sampling.
They show a slide that says:

Then they show the benefits of using Thompson Sampling to help with that.
I assume they're saying - for newer gigs they don't have enough info about them to know whether the gigs are good enough (eg. get high enough conversion rate (clicks->orders)) - so don't know where to normally display them in the results.

So maybe they use Thompson Sampling (or something like it) to help decide the gigs to show where and how often so they can get enough data on them to see how good they are (click through rate, conversion rate etc.), and using that sampling should be better (according to their slide I think) than using a random element to help decide which new gigs to show in which positions (eg. until they get enough data on them).

eg. their slide seems to say they did 4 weeks of A/B tests of random vs Thompson Sampling (checking CTR, conversion etc.) and seem to be saying that the latter was better. I'm sure they'd still use all the other metrics/variables but this might be one other thing they use.

Though in their slides they don't mention gig/order price, but I'd assume they'd take that into account in the actual algorithms(s).

I had an hour long conversation with someone from the actual search/algo team.

Things are “interesting” to say the least. 🙂 

I promise to post something extensive soon.

[myb2bservices]

On 3/21/2022 at 1:53 AM, frank_d said:

Hey gang!

A disclaimer: The following post/article is not an official Fiverr statement. It’s a summary of my personal observations over how Fiverr works and I am sharing because I noticed that more and more sellers come here, stating that they “lost their ranking”, or claiming that their gigs are "all 5 stars" but they can't get any messages/sales.

This is an effort to provide them with some answers and some food for thought.

Hold up. Fiverr 3.1?

I know you were probably expecting something a bit more dramatic, but this is not a brand new thesis. 

This is me revisiting my previous one, and updating it after seeing a few new things popping up and some previous assumptions being confirmed as either off-base or not as important as I previously thought.

"Ha! I knew it, you were wrong!"

Well actually no. Let me preface this by saying that I am revisiting my original thesis, not because it was proven wrong.

In fact the core premise of my past article, still stands true.

Fiverr favors performance, and is still trying to get a perfect match between buyers and sellers.

So it's still about relevancy, performance and quality.

The reason why I am revisiting the thesis is because I have stumbled upon new evidence that tell me there have been a few updates during the past two years. Not an overhaul mind you, but several "Quality of Life" updates and additions.

Let's talk about indicators

OK so we are all familiar with our profile's metrics.

These are the actual metrics that Fiverr chose to reveal to us sellers in an attempt to provide us with some crucial performance indicators.

This is actually the original system Fiverr put into place back in the "2.0" era, when it needed to be able to qualify sellers and update SERP dynamically. Fiverr 1.0 only had the thumbs up/down rating system and other than the actual level sellers and the occasional reporting/flagging by buyers for inappropriate behavior, Fiverr had very little to go on when it came to understanding who was a good freelancer and who wasn't.

So it would make sense for a seller at any given time, to just look at them and go: "I must say, I am doing a jolly good job! Let me pat myself on the back before I deliver this order and get my 5.0 star review!"

Right?

Wrong.

Two systems colliding

As you can probably guess, Fiverr grew even more. And what usually happens when a company grows this much and at a certain rate, its initial systems are no longer able to keep up with growth pains that come up.

So a new system was put into place. It didn't replace the old system, but it is working alongside it, in an effort to help Fiverr understand performance.

This new system is essentially there to give more data points to Fiverr and help qualify certain actions.

That new system started off with the "buyer satisfaction rate".

That's not an unfamiliar term nowadays, as Fiverr employees started actually talking about this openly for the past year or so.

I talked about this on my previous article, but now it's not just a rumor or the ramblings of a mad man, Fiverr is actually talking about this, thus confirming part of my thesis.

There is a formula in place that makes sure the system get valuable information on whether a seller is performing well or not.

So what's new?

Well, that buyer satisfaction formula is apparently more extensive now, and a lot more complicated.

Furthermore, there are now two different systems in place, working towards the same goal.

One system is there to determine whether or not your profile is in good standing and then there's a different system in place that actually grades your gigs individually.

(the average Fiverr seller's reaction when they hear of a new performance rating system)

Now before you start grabbing your pitchforks, let me explain why that is potentially a good thing.

First of all, I have been rather vocal about how this new system -no matter how imperfect it may be- is at least allowing for the possibility of recovery. So when a seller's performance slips, it just takes some time after which you are back in the game.

It's a more forgiving system altogether, or it seems as it was designed to allow for more margins of error.

But let's get back on track.

The two systems are not fully aligned.

Seems like the profile performance metrics are refreshed on a 60-day window, whereas the gig's performance metrics are calculated on a 90-day window.

I don't know why that is, but a good guess would be that they are just stacking different systems on top of the old build so they can't just nuke previous choices that used to make sense 4 years ago.

I am taking a deep dive into the macro-economy of the system, in an attempt to help de-mystify the issues I see popping up on the forum on a daily basis.

People come here and complain that things are A-OK with their clients and reviews, but they either lost promoted gigs eligibility or their gigs are not "ranking" the same anymore. 

(notice the overdramatic air quotes when I mention ranking)

 

Fiverr's systems are opaque. That's by design. So I understand that it's frustrating.

But on the other hand, I have been here since 2013, where things were less complicated and you could simply piece things together by landing the right CS agent on your support ticket.

And let me tell you: people used to game the living crap out of the old system.

(Me after realizing public reviews are now worthless)

Which is basically why Fiverr constantly makes changes.

Reviews were exploited -people still to this day pressure their buyers for positive reviews-, the number of sellers exploded since 2019, and Fiverr constantly expanding across more verticals.

Fiverr is trying to come up with a legitimate way to understand two main things:

• Buyer behavior

• Seller performance

 

But how can one improve if the metrics are not opaque?
 

This is where this whole "indicators" business comes in.

We need to look at what the system is telling us.

Stop focusing on what the system is not telling you.

That was a big part of my original thesis that people focused on and tried to debunk/discredit it.

I talked about speed and gave some specific examples of indicators that Fiverr has subtly left there for us to see, and people were very keen on rejecting them outright.

On this update I chose not to focus on the indicators as much, as a lot of people were focusing on whether or not they existed and/or utilized by the system. Instead I want to focus on the bigger picture, and some good practices for sellers on Fiverr in 2022.

This may be a good place however to brag about how I was right when I said that Fiverr likes speed and favors sellers who respond quickly. As since I made that claim in my previous article, Fiverr actually came forward with the "order response rate" which actually shows you if you responded fast enough when a new order was submitted.

(me after Fiverr updated the seller's dashboard proving half of my assumptions to be RIGHT)

What does this all mean for us sellers?

That's a legitimate question.

1) Well first of all you need to understand that looking at your dashboard metrics and getting 5.0 star reviews from your buyers, is not giving you the whole picture anymore. (It's still important that you keep those metrics as close to perfect as possible, and not getting negative reviews should be the absolute baseline.)

2) It is also important to understand that the "Buyer satisfaction rate" dates back 90 days, so it's a delayed view on your performance.

3) It is also my understanding if multiple orders don't go that great during the same 90-day window, they stack up, meaning things will not improve for your gig's standing until at least one of them is removed from that rolling window.

4) Fiverr no longer looks for just amazing web developers, cool logo designers and Pulitzer-worthy writers when evaluating performance. In fact, Fiverr is not equipped to understand if you are delivering a high quality design or a photo-copy of mashed potatoes.

Fiverr is looking for great facilitators. As a seller you should be focusing on your buyer's experience overall as much as you do on your actual craft.

5) The promoted gigs feature is not as simple as I initially suspected it to be. It's actually part of this new complex system that Fiverr has set up, as it is being fed with your gig's performance numbers constantly, in order to only promote the best gigs out there. It is also intertwined with the actual system that understands if you are relevant to what people are searching for.

So when you lose eligibility and can no longer promote a gig you used to be able to promote, this is an indicator of a "buyer satisfaction rate" slip-up, most likely on a gig performance level. By then of course it's too late to act upon it, but it is an indicator nonetheless.

6) Relevancy is calculated using a more complicated formula now. ASP (average selling price) on your gigs, is also now being calculated when determining where your gig is placed on search results, past performance matters more and is actually used to project future outcomes when Fiverr matches you with buyers, and this new Fiverr briefs feature is also trying to guide buyer's behavior and eliminate the need to search altogether. 

7) I noticed that the "returning buyer" metric is not just a nifty little animated badge you get on your profile to serve as a trust signal toward your prospects. It is actually a brand new metric that Fiverr uses to calculate performance. I was initially super-annoyed by it, as I am a Pro seller who has a higher barrier of entry due to pricing alone, so when I realized that Fiverr was valuing return buyers, I'm not going to lie, it felt like it was game-over for me.

But I did try to assess the situation and when it became apparent to me that it's a brand new metric that Fiverr will use to evaluate me, I pivoted and adjusted my gig strategy to compensate and stay relevant.

(me going back to the drawing board every time there's an algorithm update)

I have a birthday cake to go to, turning 41 today, so I will try to sum things up and give you some things to think about.

Fiverr is not a perfect system, nor does it try to be a fair system. (more on that on a future post)

It is doing its best to evaluate millions of gigs and hundreds of thousands of active sellers, to make sure it creates a perfect match to buyers that have their credit card at hand, ready to spend money.

It is constantly adding metrics, tools and data points, in an effort to perfect it formula and adapt to ever-changing buyer/seller behavior.

Whether we like it or not, by being a part of this platform, we have to realize that there is a system in place and that the only way to grow as a business and be favored by this system, is to understand its rules and what the system is trying to achieve.

Fiverr is making huge strides in understanding and mapping out buyer behavior, which is a good thing for us sellers.

But the most complicated part of Fiverr right now, is the system trying to evaluate seller performance on so many levels.  

OK so I kept talking about a really complex system, without being able to dive into as much details as I would like.

I also mentioned that indicators are important, but I also didn't want to have to reply to dozen of messages telling me that "there's no way the system would measure X Frank, that's totally not cool!"

But I think I can share this one theory that I have, that is also a really strong indicator to anyone who is willing to research their gigs and past performance.

Unfortunately, this is only for Seller Plus members, as I will mention a tool that is only available to them.

So here goes. 

Seller plus members are able to create special discount coupons, and then offer them to their past clients, in order to create more returning buyers and a stronger relationship between vendor/client.

Now here comes the important part:

You can't offer coupons willy-nilly. There are apparently some metrics that won't allow you to even see a past buyer in your list when trying to send coupons.

And here's my theory:

Since the list of past buyers available to you is based on the past 90 days, and since there is a visual indicator right there on the UI telling you that the system picked which buyers you get to see/send the coupons to, what if this is a good way to look into past buyers and figuring out what the system won't tell you:

A) either which type of buyer was a good match for your selling profile

or

B) which past orders affected your buyer satisfaction rate.

I am not in any way suggesting that you should rush on your coupons dashboard and try to look at that list to find out which buyers may have left a less than positive exit survey.  That's not the case.

I am simply saying that if Fiverr is trying to create great matches, and then it sort of/kind of tells you who wasn't a great match, maybe you can put the missing pieces together and see how you can improve moving forward.

This has not been proven, nor have I discussed this with anyone on Fiverr. (I don't think I want to do that)

I am bringing this up just to give you an example of what an existing indicator may be. 

Fiverr won't give you the formula, or a straight answer of how performance is calculated, but there are strong indicators peppers in throughout the UI telling you what Fiverr is looking for.

Long post over.

Welcome to Fiverr 3.1.

As always, I am here to answer any questions and feel free to share your experiences!

 

Thank you! 🎂

 

 

Happy Birthday!

All these changes are affected heavily on their stock value. Companies should scale, but should not change core values. 

 

 

[williambryan392]

6 minutes ago, myb2bservices said:

All these changes are affected heavily on their stock value. Companies should scale, but should not change core values. 

Huh?