Relevance

[Couldbe Yue]

Down the last month or so I swear that whenever relevance gets updated (I'd say a couple of times a week max), my sales seem to dip for a few days after.

I haven't the will to check each day and make notes but does anyone else think they've noticed this?

[Czari Zenovka]

I haven't checked relevance - not really sure how to check it - but I'm noticing something another poster said in another thread - I'm seeing swings of people purchasing my less expensive items for a few days...then no sales...then the more expensive items.  Rinse/Repeat.

And when I say "less expensive" - it's one item that I didn't think worth even putting on the MP.  I made it basically as a cabinet to display other things on in my store, then set it at 10L or 15L just in case someone wandered by and liked it.  Recently put it on the MP and was astounded to see it sell!

[mtwtfss71]

You know, I've noticed this as well. I'm not sure if it times with relevance being updated but my sales seem to go like this (over the past 6 weeks): 3 days of inexpensive items selling well, 2 days of almost no sales, 2 days of great sales with majority being expensive items. Rince and repeat. I was wondering if they have some sort of cycle set up with search criteria to be honest.

[Couldbe Yue]

I think it might be a bit scarier than that.  Last weekend I did a search on an item that only returned a couple of 96item/page listings.  I went back to have a look at it yesterday and it was 5 pages - with a whole lot of listings from around #15 onwards that certainly weren't there on the weekend - and a lot of the ones that were shown on the weekend were right at the end.

I know I sell a lot more low priced stuff than before and when I do have a big day of sales usually relevance is updated the next day (only once where it didn't update) and my sales are becalmed and I go back to selling trinkets.

Interestingly my marketplace sales haven't really deviated in month total since just after it finally went live and they abandoned all pretence of putting in any of the other promised phases..  I find it really weird to have no sales days (I do keep track of those because my sales are literally so consistent that a couple of no sale days are never usually made up by month end) as on the marketplace that's my new norm - yet there's no pattern to it (i've been monitoring that for a few years now).

Something is certainly going on and as usual it's not to our advantage in any shape or form.  I can't even begin to guess the pathetic reasoning behind this jerk circle of theirs but then, this is the group of people who couldn't admit they made a mistake when they trashed all those migrated listings and refused to do the only action possible which was backing out the changes and restoring to an earlier time (probably because they're not professional enough to have taken a backup before putting the probably non tested code in) and have lied about a fix being on the way long enough that people have given up in despair.

hey ho.

 

 

 

 

[Deja Letov]

I think the problem is that it's not really a relevance search. It seems to be that their explanation for relevance really has not much to do with relevance and more to do with sales. I have periods of shift as well, and I'm a total number cruncher so I'm also investigating why things are happening. I've noticed that when search gets updated, if I haven't sold something in awhile, that item stays stagnant near the bottom and doesn't sell. But if I happen to sell several of the same item right around when the search updates, then my sales stay up because it ranks higher on the search results page. All of this has nothing to do with relevance and more about the quantity of sales....which is absolute horse crap if you ask me. Why there is a search for best selling, I will never know, because that's fairly close to what relevance is based off.

[Couldbe Yue]

my best guess with relevance is that it's a combination of things, including sales, views (although I'm beginning to think that views are no longer used), pictures with the keywords themselves ranking low on the weighting - which is why you can have a dozen pages of listings returned for a search term and by the time you're at the bottom of page 2 the returns are completely irrelevant then a couple of pages later there are keyword relevant items.

What I'm finding interesting, but I've not really had a close enough look at it to be really confident, is that any activity during the days when the relevance calc isn't run isn't actually included in any future calcs.  ie I've noticed that if I have a run on something and it doesn't update relevance it doesn't seem to end up where I'd expect it to be in the ranking on my shop page after the 3 day window has passed.  If that makes sense.

this has just been one of those things that has been puzzling me for a while but not enough to actually start to make notes to look for the pattern but the seriously scary bit was those missing listings from the search I did on the weekend and seeing the full (I assume) returns yesterday.  if a couple of hundred listings can not be returned then that explains why sales go dead.

 

 

[Spica Inventor]

I look at the relevance standings religiously every day pretty much and it seems to get updated daily usually, but there are still some days that it doesn't get updated. But some things sold are listed sold a day earlier or a day later then they actually where sold I've noticed. Have you noticed any unusual increase of sales around those days listed as no sale days? Like product sold was being pushed back or up for some strange reason.

 

[Couldbe Yue]

the incorrect day stuff could be because the order list works to your current time zone (iirc) but I suspect that relevance is calculated on the pst daily window.  consistency isn't a strong point.

the easiest way to check is to compare the slm list with your transaction history and if it's not that then I'm out of ideas.

Tbh I just glance at the first half of the first page of listings and usually sigh heavily,  Certainly I've noticed that some sales do seem to affect relevance a couple of days after the sale rather than the next day but haven't paid enough attention to come up with even a speculative analysis.  I did notice that a sale I made on another account a few days ago did affect relevance the next day but couldbes account didn't seem to move with her sales.  I wonder if the weightings are such now that some sales just don't affect relevance anymore.

I'm in one of those becalmed periods atm with sales so when they start again I'll pay a bit more attention.

[Medhue Simoni]

The easiest way that I know to check if LL changed relevance, is to check your store. On your store's page, the ranking of the products should reflex your regular selling products. If it is completely out of wack, like mine seems to be now, then LL screwed a bunch of stuff up. I've known to check my store's page for a long time now, but this was reitterated to me by a Linden, and I was told to report anything strange in my store's product ranking when I see it. Of course, when it is wonky multiple times a week, it is hardly worth my time to report it. Right now, my product rankings in my own store are as wonky as I've ever seen them.

Just imagine too, if your own products on your own store's page is wonky, consider how this gets applied to the whole MP.

[Couldbe Yue]

You're right about checking the shop homepage for this.  from what I can see there's a default relevance that is effectively your baseline. Then any sales within the last 3 days are given a higher weighting for the sale depending on the date of the dale (yesterday, day before, day before that - in descending order) - price also appears to have a different weighting if the sale falls within that 3 day window.

for my lower ranking items, views do seem to have an impact on the ranking as even if they haven't sold for a while they'll still shuffle around the ranking but these days views don't seem to do much if an item hasn't already sold at least once.

[Spica Inventor]

1000+ L$ products get a big ranking boost for a few days on the relevance search that is not reflected in how they are ranked in your store area on the marketplace I noticed. I'm not sure if products in the 501 to 999 L$ range get any type of ranking boost as I don't sell anything in that price range. If L.L. wants to push prices higher, I don't really care, but it would be nice if they where more honest about it. ;-)

[Spica Inventor]

The search standings use to be updated every morning at about 1 AM SLT when the magic boxes where the only game in town. I happen to be on SLT also so the dates should match or at least be consistent for mine anyway unless there is some extra weird mystical procedure that's going on periodically behind the scenes covertly of course. ;-)

[Pamela Galli]

Relevance seems to mean "What has recently sold", and have nothing to do with which things are the actual money makers. 

[WADE1 Jya]

---

Pamela Galli wrote:

Relevance seems to mean "What has recently sold", and have nothing to do with which things are the actual money makers. 

---

Doesn't help when 'relevency' is only updated every few days or some such too, right?

I say it seems to mean 'what has recently transfered to an avatar'. In my books, moving items for fractions of pennies (or straight up zero pennies) is not 'sold' or selling anything. Not when you're dealing with a niche market of less than 60000 people anyway. Yet this looks to be what Lindens chase, Maybe in bizarro world free = profits? More likely its set up this way entirely accidental with their usual haphazard abandon & they just wonder glumly 'gee why isn't this working out?'. I truly believe they have no idea what they are doing nor care to find out. We're just not dealing with professionals here.

[Couldbe Yue]

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150 as I had a quick look at my last months sales and the overwhelming majority of them were around that price - a few higher (500-1000) , an even tinier amount of my expensive stuff and almost none of my sub 100L.  So (this month excepted so far) I'm maintaining the same turnover but doing it through increased sales of my lower priced stuff.

I can see where people think it veers between low and high.  After days of selling c150L priced items, today I've sold a few big ticket items and nothing under 500L. The thing is though it could be people's shopping behaviour more than anything LL is doing.

I can't even hazard a guess as to what is going on.  Only LL have the data and I doubt they ever look at it, let alone have a willingness to share.

[Deja Letov]

---

Couldbe Yue wrote:

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150 as I had a quick look at my last months sales and the overwhelming majority of them were around that price - a few higher (500-1000) , an even tinier amount of my expensive stuff and almost none of my sub 100L.  So (this month excepted so far) I'm maintaining the same turnover but doing it through increased sales of my lower priced stuff.

I can see where people think it veers between low and high.  After days of selling c150L priced items, today I've sold a few big ticket items and nothing under 500L. The thing is though it could be people's shopping behaviour more than anything LL is doing.

I can't even hazard a guess as to what is going on.  Only LL have the data and I doubt they ever look at it, let alone have a willingness to share.

---

That's really interesting. My top 3 selling items are 100L (first place), 75L (second place) and 100L (third place). In fact, until you get to about the 8th top selling product in my list, they are all 150L or below and then in 8th place it jumps to 500L. Could be onto something.

[WADE1 Jya]

---

Couldbe Yue wrote:

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150...

---

Blue Linden was so cool & honest. One of the Lindens I wish was still here.

Anyway lol yeah I think you are right.

So what do you do if you think you are worth a little bit more? Say 99

¢ or $1.99... or even gosh $2.99?

I guess I'll continue doing what I've been doing lately. Anything I make that I think is worth more than 60

¢ (minus all their outrageous fees) goes elsewhere (ie everything). Some of my oldest stuff I have marked down to those prices.... but if I thought something I made today was only worth 60

¢ I would probably other trash it, or work on it to make it be better!

Way to grab the bottom of the barrel Lindens. Sorry Lindens I'd love to contribute more to your platform but I gotta have self respect & unfortunately there is bills to pay in my life.

 

[Deja Letov]

---

WADE1 Jya wrote:

---

Couldbe Yue wrote:

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150...

---

Blue Linden was so cool & honest. One of the Lindens I wish was still here.

Anyway lol yeah I think you are right.

So what do you do if you think you are worth a little bit more? Say 99

¢ or $1.99... or even gosh $2.99?

I guess I'll continue doing what I've been doing lately. Anything I make that I think is worth more than 60

¢ (minus all their outrageous fees) goes elsewhere (ie everything). Some of my oldest stuff I have marked down to those prices.... but if I thought something I made today was only worth 60

¢ I would probably other trash it, or work on it to make it be better!

Way to grab the bottom of the barrel Lindens. Sorry Lindens I'd love to contribute more to your platform but I gotta have self respect & unfortunately there is bills to pay in my life.

 

---

Dang it wade...I can never read your posts. Looks like light blue ants across my screen.

[Dartagan Shepherd]

---

Couldbe Yue wrote:

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150 as I had a quick look at my last months sales and the overwhelming majority of them were around that price - a few higher (500-1000) , an even tinier amount of my expensive stuff and almost none of my sub 100L.  So (this month excepted so far) I'm maintaining the same turnover but doing it through increased sales of my lower priced stuff.

---

Whatever it is, I think the market manipulation needs to go. If newbies are getting sticker shock, it's not from virtual goods, it's from tier prices.

Also ... because retention hasn't budged, market manipulation doesn't work for newbies. It may work for maximizing their 5%, though.

During boom times, newbies didn't blink at L$3,000 for a shape.

Also, market manipulation has the opposite effect on merchant profit margin, which in turn makes being a profitable merchant less viable and more an exercise in playing the lottery.

One purpose seems to be creating a spread for merchants. People get "turns" at sales in order to either hide the lack of a previously healthier market or to create the illusion that at least most people are selling "something".

They should probably focus on writing decent software and leave the market alone, but it does seem clear that it's all about influencing the market purchases and not about just providing people the best results.

Masters of the universe, they can't do something unless they overdo it.

I just did a search for "relevance" on the marketplace and the top result was a "spank me harder" product, and that certainly seemed relevant, though.

[Czari Zenovka]

---

Dartagan Shepherd wrote:

---

Couldbe Yue wrote:

Back in 2009, I had a chat to Blue Linden and he said (and I paraphrase here) that LLs aim was to have people sell their wares free or cheaply to encourage newbies to stay.  At that stage I think the their idea of the perfect relevance score was either 300 or 350L on XSL.

I'm wondering now if they've dropped it to around 150 as I had a quick look at my last months sales and the overwhelming majority of them were around that price - a few higher (500-1000) , an even tinier amount of my expensive stuff and almost none of my sub 100L.  So (this month excepted so far) I'm maintaining the same turnover but doing it through increased sales of my lower priced stuff.

---

One purpose seems to be creating a spread for merchants. People get "turns" at sales in order to either hide the lack of a previously healthier market or to create the illusion that at least most people are selling "something".

---

You might have something there, Dartagan.  I do feel like I'm playing the lottery these days as a merchant - will I sell *something* today, will it be another few days of no sales followed by several days of multiple-sales/per day.  Then the particular items I'm selling is not trending like it used to consistently for years.  Every once in awhile there would be an odd item (the item is not odd - just that it's not a big seller ) sold, like once a year.  Now, the "odder" items (ie. not my "mainstays") are the ones that are selling.

 

[WADE1 Jya]

---

Deja Letov wrote:

Dang it wade...I can never read your posts. Looks like light blue ants across my screen.

---

Hey it could be worse I could type like this hehe :catembarrassed:

[Deja Letov]

hahaha that was terrible. lol

[Darrius Gothly]

---

Dartagan Shepherd wrote:

[snip] ... I just did a search for "relevance" on the marketplace and the top result was a "spank me harder" product, and that certainly seemed relevant, though.

---

ROFL!! Too trucking fue Dart.

I have been seeing evidence of this same manipulation myself. Because I monitor Search Rank fairly closely, and because I see the ranks of various competitors side-by-side, I see this same behavior there. One week Merchant A will be on top, the next week it's Merchant B's turn .. and so it goes.

I haven't been paying attention to Marketplace, but I can say that my sales tend to run in spurts. Interesting that we can all gather here and begin to spot those same "Up then Down then Up.." behaviors.

Yup, manipulation is what it looks like .. to be sure. Dang shame if that's what they've decided is the best way to run the Marketplace. Turns it from an exercise in quality, service and good business into a merry-go-round with no golden ring.

[Deja Letov]

Another thing I've been noticing is that your related products seem to adjust your placement as well. i listed a few items and put no related products in them. I did a search term that not very many results came back so I could see it all on one page while i was testing. With this search term 7 of my products were returned along with several other other products that weren't mine. Two of my products had this particular search term in the description and once in the keywords ONLY (nothing in the title or features and the actual inventory name didn't contain it either) BUT I then those two products to have related products back to products that had the actual search term in the title, description, and keywords. Those two products are at the top of that page now for the search term. The actual products that are titled with the search term, description and keywords are still at the bottom. Weird huh? So then I took a couple of the lower ranked products and related them all back to themselves. So they all now have the search term in their appropriate places as well as related back to each other. Nothing much moved. but then I took a product, and in addition to having it relate back to the other products that had this keyword, I also related it back to those 2 original products that didn't have the search term anywhere except the description and keyword. When i did that, the product jumped up in several places. So I guess it would be safe to assume, that relevance also has to do with what you related back to your product...which actually make sense to me for relevance.

[Couldbe Yue]

I never considered that one before but it does make sense in a bizarre way.  I might give that a go if I get a chance on the weekend and see what it does to mine.