Jump to content
Sign in to follow this  
JustinMichael Torok

Marketplace Review Questions

Recommended Posts

Hello,

In the reviews tab for a marketplace listing there seem to be some functions that aren't well explained anywhere.

Does anyone know what the "Was this review helpful?" yes/no buttons are for or if they do anything?

When you comment on a person's review, do they get notified in any way?

Is a review less than 5 stars considered negative like most commerce websites? If so, are customers made aware of this in any way?

Lastly, do reviews impact an item's listing position in the Marketplace and, if so, how?

Share this post


Link to post
Share on other sites


JustinMichael Torok wrote:

Hello,

In the reviews tab for a marketplace listing there seem to be some functions that aren't well explained anywhere.

Does anyone know what the "Was this review helpful?" yes/no buttons are for or if they do anything?

When you comment on a person's review, do they get notified in any way?

Is a review less than 5 stars considered negative like most commerce websites? If so, are customers made aware of this in any way?

Lastly, do reviews impact an item's listing position in the Marketplace and, if so, how?

I do not know anything for sure, all is shrouded in mystery, but I suspect reviews impact rank. 

Share this post


Link to post
Share on other sites

JustinMichael Torok wrote:

Is a review less than 5 stars considered negative like most commerce websites? If so, are customers made aware of this in any way?

 

I can't get my head around the idea that anything less than 5 stars (assuming out of 5) is to be treated as negative.  4/5 is a level of positive above the median in anyones book.

Share this post


Link to post
Share on other sites

Pamela Galli wrote:

I do not know anything for sure, all is shrouded in mystery, but I suspect reviews impact rank. 

That is correct. Search listing ranking on MP is based on two factors and - as far as I can see - only two factors: sales figures and review ratings.

Sales figures are weighed against time and price - recent sales count for a little bit more than older ones and the higher the price, the more ranking boost each sale gives.

There doesn't seem to be any weighing or such when it comes to reviews. The more stars your listing has, the higher it ranks, pure and simple.

 

Edit: Forgot to mention, if two listing have exactly the same amount of "rakning points", the oldest of them is ranked the highest.

Share this post


Link to post
Share on other sites

My anecdotal evidence supports what Chinrey offered.

There was a time when my top selling product was on the first page when searching for just the letters "RLV" and then suddenly it changed.  No difference in my sales numbers but searching just on RLV then returned about 2 pages of items which were all in about the L$2000 price range (give or take a bit) and then the next 10 pages were almost all exclusively L$0 items.

This gave rise to me thinking that they factored in that items that were "affordable" but of a high enough price would yield more commission for LL but then throw in a bunch of freebies to make everyone else happy.  That's certainly what it looked like because had it been that freebies by sales number were higher, then they'd always have been higher before this sudden change.  Don't get me started on the fact that nobody "sells" a freebie in the first place!

At this point, out of curiosity, I decided to create the L$1,000,000 RLV pose ball with only that as the title and keywords as I didn't want to add any other potential skew.

It didn't appear at the front of the "RLV" only phrase listings so I figured it was because it was outside the qualifying price range considered affordable to tempt.

However, on the day that it did sell 27 times, it rocketed to number one "best seller" on MP.  What we can infer from this is that "best selling" isn't by quantity as others will have sold far more in that time so "best selling" results must be a product of price x quantity and i'd expect that they do the same for relevance in results.

(of course they may have changed things since then)

Share this post


Link to post
Share on other sites


Pamela Galli wrote:

How do you know these things, Chin? i have heard speculation but nothing for sure.

Mainly by testing and observing but also by talking to others who have done the same. I'd rather go too much into details about our methodology here - it would be too long a post for a start - but here are some simple examples:

The fact that the most recent listings are listed last is easy to demonstrate. Just create a new listing, wait for a few minutes for it to show up in search and then do a search for one of the keywords. It will always show up last. But the moment you sell a copy it jumps quite a bit up on the ranking, overtaking all those listings that never made a sale at all. (After that initial jump, however, it's a struggle to get it up to a place high enough people might actually see it in search results...)

An experienced MP merchant once told me he never removed listings from his store. No matter how outdated they were, his old listings would always outrank and outsell the more recent ones. I've done a few more tests in addition to the simple observation above and they all confirm that new listings rank lower than old ones. It's not a very strong factor though and it seems it's not part of the ranking algorithm per se but simply that listings that can't be separated by ranking points are listed unsorted - in the order they appear in the database.

This also affects updated listings btw but I have not been able to determine if that is permanent. If/when you update the description of a product, it will drop significanly on the ranking. This may only be temporary though, in the occurences I have observed, the updated listings seemed to return to or close to their original position after a day or so. Need some more testing to determine if updating does any permanent damage to the ranking.

 


 

The significance of reviews/stars is also fairly easy to see. A while ago I had a bed for sale at my store. (Nothing fancy, just a basic one sold at a symbolic price to offer newcomers and others on a strict budget a good low LI/high LOD alternative to the high prim old freebies and low LOD new mesh garbage they usually have to litter they homes with.) The bed was available in different colors. The white silk one was the most popular and soon outsold and outranked all the others. But the the pink version got a positive review and suddenly it was the highest ranking. The white one was still the bestseller and soon caught up but the pink one retained its ranking position relative to the other variants and also to beds by other makers.

So yes, reviews do affect search ranking and although I need more data to be sure, the evidence I have indicate that the ranking points you gain that way are permanent, unlike the ones that come from sales.

What I can't say is if/how the different ratings affect ranking. I don't buy reviews, I sell mostly to the consumer rather than collector market so I don't have a fan base and for some weird reason I don't seem to have many enemies in SL either. That means I don't get many reviews at all. The one four star rating I have seemed to have less ranking effect than the five star ones while the only one star rating I ever had didn't seem to matter for ranking at all. That's not nearly enough data to draw any kind of conclusion though.

 


 

Sales figures...

I have a series of rose bushes in different colors at my garden and landscaping store (they're just about the only items I have there at the moment, I really should do something about that...) They were listed at the same time and are quite similar so it was easy to see how their rankings changed according to how well they sold. However, one of them is sold at a symbolic giveaway "promo" price and it is of course the bestseller. it never seemed to get as much ranking boost as I thought it would though. At the beginning a single sale of one of the other was sometimes enough to outrank it for a while.

It was another merchant (I'd rather not name him) who explained it to me. He is the kind of person who, just like me, has to know how things work and he has of course had far more time to experiment and gather data. He had noticed the same and found out that price matters, the more an item costs, the more each sale counts towards ranking points.

It's not just a straight money-towards-ranking-points relation though. Even freebies gain ranking point through "sales" and twice the price doesn't mean twice as much rank gain per sale. In my store I'm selling a basic unfurnished skybox at a giveway price. I think it must have sold more than all my other houses combined but at 10 Lindens it doesn't bring in much money. Even so, it's ranked as no. 2 in my store right now. (Btw, this makes me wonder just how much BDR, Larry and Alessandra have sold of their dollarbies to get top rankings in such a competitive field as applier clothing. Their sales figures must be astronomical!)

I was going to give an old and lengthy example how sale dates affect ranking but I just had a look at my store and found a fresh one instead.

I have two five bedroom family houses for sale, Coniston Villa 01 and Coniston Villa 02. Same price, same amount of reviews etc. No. 1 has been for sale for quite a bit longer and has been selling slowly but steadily. No. 2 was added later and didn't sell at all for a long time. I was actually thinking about removing it from the store. But recently it has started to pick up speed and at the moment it is the best selling of the two. It has also become the highest ranked of the two even though the total sales figure is still considerably lower. The only explanation I can think of for that is that recent sales count for more than older ones.

 


 

The last question is of course, are there any other relevant factors?

It's impossible to say for sure but I haven't found any and I haven't heard of anybody who has either.

  • There certainly is no algorithm to evaluate a listing's relevancy to the search phrase.
  • Keyword density does not matter
  • It doesn't seem to matter which data field the keyword is in - as long as it's one that's actually searched of course
  • There is no indication that click-through ratio matters

Oh, and in case any conspiracy theory fans are reading this: No, Linden Lab does not tweak the search results manually to favor specific merchants. Even if they had been dishonest and stupid enough to try (which they aren't), they simply wouldn't have the time for that. ;)

 

 

Share this post


Link to post
Share on other sites

Thank you very much Chin and Sassy. I pretty much ignore market wide search rank, it is so irritating, and have since LL destroyed MP search (dont get me started). I only look at how things in my store are ranked -- and yes, recent sales have a lot of weight, which is easy enough to notice. 

Share this post


Link to post
Share on other sites

Could it be like YouTube, where basically all reviews help towards the ranking because they count as interaction. I would hope not if the average review is less than 3. That wouldnt make sense. I think a 1-2 star rating is considered negatove, 3 is literally just OK and 4-5 is positive. 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...