Jump to content

What if User Behavior is part of the Algorithm Calculations?


eoinfinnegan

Recommended Posts

Note: I have zero evidence that User Behavior is a factor, I just like exploring rabbit holes like this and seeing if there seems to be any similarity to reality.

Intro: User Behavior is used in many algorithms and is now (imo) the biggest factor Google uses for ranking websites. User behavior in Google’s case includes things like:

  1. Time spent on a website
  2. What and how many pages the user visits
  3. Did the user continue to search for a different site or was their query solved by the site (did the user go back to Google)
  4. Did a user click on a specific result or not (if yes then it’s positive for that site and negative for the others)
  5. Conversion rates

That’s a basic overview of User Behavior in Google’s Algorithm which I won’t go into any further as it’s not the purpose of the topic. When looking at some website issues for a client recently and then coming to the forum to see an Algo debate, I wondered about the idea of Fiverr using a similar system to Google or even just a very basic version of it. It made me think and so I am going to mention a few things that Fiverr sellers talk about doing and explore what the effect these actions would have if User Behavior is a factor in Fiverr.

When I describe an action as Positive it means I am suggesting it improves ranking position. If I say Negative it would lower ranking position.

If lots of people are checking where their gig ranks in results for certain keywords (eg. “Logo design”), what would the effect of all these people checking where their gig ranks be?

  • If they look at the results but don’t click on anything there is likely to be little effect on any of the gigs in the search results - possibly an ever so slight negative as no gig enticed the buyer to click.

  • Now, if they did click on one then it would show Fiverr that out of the whole page of results, that gig was the preferred one. This could be a positive for that gig BUT if the user then goes back to results or searches Logo Design again it would be a Negative.

  • If the user clicks but doesn’t go ahead and buy the gig then it would be a Negative. This is what Google refers to as a conversion rate (not to be confused with Fiverr’s conversion rate which is slightly different).

  • If the user clearly read the whole gig description (noticed by the time on site, scrolling actions etc) then that could be a positive as it seems like the gig is at least of interest to the person who searched for Logo Design.

  • It is likely that there would be a failsafe for if the same person does this type of thing (searching but not ever buying) to try and stop the algorithm being manipulated. Google has this.

Hyper Speculation Time:

IF this is a factor on Fiverr, it COULD account for why gigs seem to rotate so much. Thousands of people searching for keywords and never buying could make a big difference and make gigs fall and rise seemingly for no reason.

It could also be why it happens that we sometimes get a lot of inquiries from one part of the world on a certain day or few days. _If a gig gets a lot of positive engagement from hundreds or thousands in one country then perhaps the gig is shown higher in that country.

If a well-performing gig with a high ranking suddenly drops for no apparent reason it could be a casualty of simply not being clicked by these searchers/phantom buyers. It wouldn’t happen with just a few people doing it but with a couple of hundred in one category I would imagine that would make a difference. I want to explain the rationale on this one in particular.

Let’s say my gig is ranking on top of Page 1.

Out of 100 buyers searching for this service, I get 25 clicks to my gig each day and 5 people buy the gig. That gives a click through rate of 25% and a conversion rate of 20%. Those are quite good metrics and what you would expect of a high performing Page 1 ranked gig.
Now, let’s say that 100 Sellers who are searching the same phrase to see where they rank and how many orders the tops rankers have. All of a sudden there are 200 searches and let’s say 25 of those Sellers click through to see my orders in queue (because they want to rant against capitalism on the forum). This means my click-through rate stays at 25% but my Conversion rate drops to 10%.
Suddenly, through no fault of my own or my buyers, my performance seems to have dropped! Fiverr’s algo thinks “hmmm Eoin is not doing as well as before - let’s drop him down a bit and boost the guy on Page 2 who has only two sales but a 50% conversion rate up for a while.”


Ok, that’s enough random speculation for now I think.

Please note again: This is all pure speculation with no evidence whatsoever.

Link to comment
Share on other sites

Fiverr dont have this on its algorithm. It is far more simple: higher price of Gig And completed orders = First page in search.

I’m not looking to discuss the theories that have been discussed a billion times before. This is simply looking at a completely different possibility. I am definite that your theory is far more simplified than reality though I’m sure both those things help.

Link to comment
Share on other sites

My own theory is that the Fiverr search employs the chaos theory principle. If a butterfly passes wind in Bermuda, I get orders.

Practically, this would mean that to get orders, I would have to execute a very complicated butterfly catch and release program, during which I feed butterflies in my care lots of legumes. Sadly, the only wildlife were I live currently amounts to just me. 😦

I do think user behavior has a lot to do with ranking on the seller side of things. I notice that as soon as I turn down a request to help someone, my orders and messages flat-line. As for user behavior as far as the search is concerned, I just get headaches thinking about it. .

Link to comment
Share on other sites

But yes, higher price and higher priced completed orders have a heavy impact on the relevance tab filter on search.

higher price and higher priced completed orders have a heavy impact on the relevance tab

I don’t think so. When I use search from time to time I see a lot of 5$ orders on the first page and only 2-3 higher priced gigs. (With or without a lot reviews)

Link to comment
Share on other sites

My own theory is that the Fiverr search employs the chaos theory principle. If a butterfly passes wind in Bermuda, I get orders.

Practically, this would mean that to get orders, I would have to execute a very complicated butterfly catch and release program, during which I feed butterflies in my care lots of legumes. Sadly, the only wildlife were I live currently amounts to just me. 😦

I do think user behavior has a lot to do with ranking on the seller side of things. I notice that as soon as I turn down a request to help someone, my orders and messages flat-line. As for user behavior as far as the search is concerned, I just get headaches thinking about it. .

If a butterfly passes wind in Bermuda, I get orders.

I have so many questions!

  • Is there a specific butterfly?
  • What if more than one does?
  • Do they need to be facing a certain direction?
  • Do you received orders from certain locations based on the direction?
  • What were you doing when you discovered this?
Link to comment
Share on other sites

If a butterfly passes wind in Bermuda, I get orders.

I have so many questions!

  • Is there a specific butterfly?
  • What if more than one does?
  • Do they need to be facing a certain direction?
  • Do you received orders from certain locations based on the direction?
  • What were you doing when you discovered this?

Is there a specific butterfly?

It can be any butterfly.

What if more than one does?

I get more orders. - Duh!

Do they need to be facing a certain direction?

No, but they must pass wind discretely. Butterflies are a very proud species.

Do you received orders from certain locations based on the direction?

Butterfly farts in Bermuda typically result in orders from New Zealand and Australia. UK and European butterfly methane releases typically result in orders from the U.S. and Canada.

What were you doing when you discovered this?

Naturally, I was quite drunk. However, within a short amount of time, I was able to scaremonger much of the EU butterfly population into leaving the EU because of Brexit. Now that they have arrived in the UK, I should start seeing an influx of new orders.

favicon.icoITV News

Thousands of painted lady butterflies spotted in July as UK experiences ‘influx’

Tens of thousands of painted lady butterflies were spotted over two days in July thanks to a migration “influx” in the UK, according to environmental charity Butterfly Conservation.The butterfly is a common immigrant to the UK from the continent each...

Link to comment
Share on other sites

That’s an awfully long opening post.

Pfft, not even 1000 words. Back in the day that was an opening paragraph not a post.

I blame Snapchat for short attention spans. Careful or I’ll throw up a topic on that too and that would be far longer than this.

Now get out of my yard!

Link to comment
Share on other sites

I’ll contribute some speculation.

You yourself click on your top of page 1 gig > no algoreaction (yes, that’s a word since now)

Another seller, who never bought a gig clicks on your top of page 1 gig (wants to see what you’re up to, copy-cat you, whatever > very slightly positive algoreaction

Another seller who sometimes buys gigs clicks > a bit more positive algoreaction

A buyer who more or less regularly buys gigs clicks > positive algoreaction

A nominal Top Buyer clicks > more positive algoreaction

An actual Top Buyer (let’s assume the algo differs between levels of buyers just as it does between levels of sellers) clicks > very positive algoreaction

Discobot clicks > you either get your gig denied, or a pretty temporary badge, depending on :robot:

(edit: forgot to mention, user behaviour in this case is the - former - buying behaviour of people who click gigs, in case it wasn’t clear)

Link to comment
Share on other sites

I’ll contribute some speculation.

You yourself click on your top of page 1 gig > no algoreaction (yes, that’s a word since now)

Another seller, who never bought a gig clicks on your top of page 1 gig (wants to see what you’re up to, copy-cat you, whatever > very slightly positive algoreaction

Another seller who sometimes buys gigs clicks > a bit more positive algoreaction

A buyer who more or less regularly buys gigs clicks > positive algoreaction

A nominal Top Buyer clicks > more positive algoreaction

An actual Top Buyer (let’s assume the algo differs between levels of buyers just as it does between levels of sellers) clicks > very positive algoreaction

Discobot clicks > you either get your gig denied, or a pretty temporary badge, depending on :robot:

(edit: forgot to mention, user behaviour in this case is the - former - buying behaviour of people who click gigs, in case it wasn’t clear)

Ooooh! I like these a lot.

So the people who have nothing else to do all day but try and boost their gig by clicking etc actually don’t make much of a difference!

That’s ingenious!

Link to comment
Share on other sites

Is there a specific butterfly?

It can be any butterfly.

What if more than one does?

I get more orders. - Duh!

Do they need to be facing a certain direction?

No, but they must pass wind discretely. Butterflies are a very proud species.

Do you received orders from certain locations based on the direction?

Butterfly farts in Bermuda typically result in orders from New Zealand and Australia. UK and European butterfly methane releases typically result in orders from the U.S. and Canada.

What were you doing when you discovered this?

Naturally, I was quite drunk. However, within a short amount of time, I was able to scaremonger much of the EU butterfly population into leaving the EU because of Brexit. Now that they have arrived in the UK, I should start seeing an influx of new orders.

favicon.icoITV News

Thousands of painted lady butterflies spotted in July as UK experiences ‘influx’

Tens of thousands of painted lady butterflies were spotted over two days in July thanks to a migration “influx” in the UK, according to environmental charity Butterfly Conservation.The butterfly is a common immigrant to the UK from the continent each...

Now that they have arrived in the UK, I should start seeing an influx of new orders.

I thought they had to be in Bermuda for it to work! 😳

If they look at the results but don’t click on anything there is likely to be little effect on any of the gigs in the search results - possibly an ever so slight negative as no gig enticed the buyer to click.

On the contrary, what if this is a slight positive for all gigs that have the keyword the buyer searched for (just because there are at least buyers searching for those keywords as opposed to keywords that never get searched)? :thinking:

Or maybe there is no net change at all because the slight positive and the slight negative (that you’ve mentioned) cancel each other out.

Link to comment
Share on other sites

What if you leave your computer open on your gig page over night?
What if you click your gig yourself repeatedly every day?
What if you buy a gig?
What if you go without checking in every day?
What if you don’t deliver the order until you only have two hours to go?
What if you contact customer support more than once a month for any reason?
What if you never change anything or edit anything on any gigs?
What if you edit something on your gig more than once a month?

And finally, what if Mercury retrograde made our sales drop? It actually was in effect the last three weeks of July, which coincided with my largest sales drop in history. It was a doozey.

giphy.gif.9db4e8408b913dd5af54a467594f6b0f.gif

Link to comment
Share on other sites

Ooooh! I like these a lot.

So the people who have nothing else to do all day but try and boost their gig by clicking etc actually don’t make much of a difference!

That’s ingenious!

No, no, the algoXperts don’t do genius, they only do empirical data, you know. So, after they analyzed countless “Like my gig and I like your gig” forum threads and found out that sellers who like other sellers’ gigs don’t actually go and buy them one day, they concluded that example_seller123 wouldn’t ever actually buy their own gig either, never mind how often they click it, and then told the algorithm to exclude both gig likes as well as clicks on gigs from the gig owners. 🙂

Link to comment
Share on other sites

Unfortunately, you’re wrong. I have seen lots of level 2 with more than 2k orders and good reviews + pretty high prices - page 3-6.

Unfortunately, you’re wrong. I have seen lots of level 2 with more than 2k orders and good reviews + pretty high prices - page 3-6.

You don’t know the number of cancellations (completed orders) they have.

Link to comment
Share on other sites

Unfortunately, you’re wrong. I have seen lots of level 2 with more than 2k orders and good reviews + pretty high prices - page 3-6.

You don’t know the number of cancellations (completed orders) they have.

@misscrystal

That’s right but I still don’t believe that price affects something because actually, no ones knows the answer of their algorithms so this topic should be closed. I will better do buyer requests instead of trying to solve enigma code here… 🙂

Link to comment
Share on other sites

no ones knows the answer of their algorithms so this topic should be closed.

Personally, I prefer the threads for which there is no answer, typically, they are the only ones worth reading for anyone who read the ToS and Seller Help Center articles, because usually, the threads with an answer consist of a question and the answer, in form of one or more people telling OP what they’d already know if they’d read ToS or Help Center info, not really as exciting nor fun as threads without answers like the traditional algorithm discussion happenings. 😉

These threads close themselves anyway and die a natural death, to be accidentally revived 5 years later by someone who didn’t notice the date of the last post, to then get mercifully locked forevermore by some unicorn called Mod.

No need to close them, really, just because they don’t have an answer. Answers are overrated. Some also have a short half life. 🙃

Link to comment
Share on other sites

no ones knows the answer of their algorithms so this topic should be closed.

Personally, I prefer the threads for which there is no answer, typically, they are the only ones worth reading for anyone who read the ToS and Seller Help Center articles, because usually, the threads with an answer consist of a question and the answer, in form of one or more people telling OP what they’d already know if they’d read ToS or Help Center info, not really as exciting nor fun as threads without answers like the traditional algorithm discussion happenings. 😉

These threads close themselves anyway and die a natural death, to be accidentally revived 5 years later by someone who didn’t notice the date of the last post, to then get mercifully locked forevermore by some unicorn called Mod.

No need to close them, really, just because they don’t have an answer. Answers are overrated. Some also have a short half life. 🙃

I’m with you on this miiila. It’s interesting to ponder things we will never get an answer to. I love mysteries.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...