Author:  dsimpsonugcs  Posted:  Jan 28, 2024 21:45  Subject:  Re: Variants Thread - January 26  Viewed:  53 times  Topic:  Catalog Cancel Reply	BrickLink ID Card dsimpsonugcs (37) Location:  USA, California Member Since Contact Type Status May 3, 2019 Buyer
In Catalog, Admin_Russell writes:    The Research The following is an example of the some of the research that has been done to help determine the course of action we are taking. When it comes to Frosted Bricks, we took a look at the price guide, the inventories that contained these special bricks, and we studied the nature of the bricks, at least as far as we could. What we found was that bricks listed under the normal (umbrella) entries sold on average for over twice the price, and the volume of sales was 17 times of bricks listed under the frosted entries. In addition, sellers listing under the main entries didn’t have to spend the time sorting out their parts. So from a marketplace standpoint, these frosted entries are simply an inefficiency (see image below). Apologies for the multiple long posts. I finally figured out on my own how the BL admins collected and analyzed their posted research data. It isn't as bad as I thought, though it does still contain a glaring math error that heavily skews the results. In any case, I don't believe the provided data can support the conclusions the BL admins are pushing. Actually, I'd caution against taking nearly any conclusions away from strictly looking at this data. It is generally low quality due to the fairly small number of sales and the very high variability of sales prices. The underlying data is what it is and BL's high pricing variability is well known. So, the data is actually just gathered from the pricing guides. Anybody can collect it, so I grabbed the latest for all of the frosted and main variants in trans-clear. The sales counts listed are for all completed sales of each listed type over the last 6 months (except there seems to be a bug that leaves around a small number of sales older than 6 months). The BL admins split the sales of the main variant across each frosted variant of the same type. I list them together once. So, total sales volume of the main variants is 15.88x the frosted variants. Conclusion: common parts outsell rarer parts. Not exactly enlightening. I don't know the prevalence of these frosted variants amongst all these trans-cleat parts in existence, but if it less than 1 in 6, then I'd say sales of the frosted variants are doing quite well on their own. They are not a minor market "inefficiency." For the pricing the admins used the lot average price. I think the quantity average price is more appropriate to use, but I included both for completeness. The admins then calculate the average of the ratios of the average frosted vs main lot prices. This is mathematically incorrect, and the result is wrong and misleading. I correctly use the geomean. Prices of the main variants are about 1.21x of the frosted variants (range is 1.11x-1.31 due to rounding of average prices). The geomean of the data from the admins is 1.47x. The biggest difference between the two is a single sale of two 3037f for $0.17 each. The data is low quality. It is not safe to draw price related conclusions from it. I think the admins are trying to push the idea with their data analysis that selling variant frosted parts under the main catalog entry would net additional revenue for sellers (i.e. they will sell for more at higher volumes!). That analysis is flawed, but even if it were true, the difference is quite minimal. If every frosted variant sold during the last six months was sold at the average main variant price the difference would be $90.13 (+/- 11.94 due to rounding). That's a total of $0.50/day across all sellers. Shall we take up a collection platter? BTW, I'm all for sellers not sorting out frosted variants if they don't want to but removing them from the catalog is not the correct answer. The admins seem to think it is and have provided data to help justify their decision. That data is bunk and the glaring math errors bring into question any other data they may be using but haven't shared with everyone. The better plan would be to solve this as part of the UX. More specifically, to list variants under the main entry (the checkbox system). The use of variants would then be opt-in for both buyers and sellers. After such a system is nearing completion, then the decision of which variants should be recognized or not can be revisited.

Message is in Reply To:

Variants Thread - January 26 - Admin_Russell

Hello again everyone, My initial post on this subject earlier this month has garnered almost 700 replies, which is right about the level of the BrickArms threads a few years [...] (4 months ago, Jan 26, 2024, to Catalog)

135 Messages in this Thread.
(Message tree supressed because there are more than 50 messages in this thread)
show message tree