When I last wrote, I discussed the roll-out of eBay’s new search engine, Cassini (read more here). This ground-up rebuild was designed to be far more intuitive than the original (and very literal) Voyager search engine that it would replace.
I am not afraid to admit that there are a couple aspects of Cassini which provide major improvements to searching abilities within the eBay system. First, when a search is performed, Cassini is designed to dig deeper into a listing. It looks at not only the title, but within item specifics and descriptions to find keywords. The depth to which it digs may vary, depending on the number of items listed carrying a particular keyword. Another benefit arises when sold items are searched. Cassini provides a full 90-day history, opposed to 14 days previously provided by Voyager. We find sold-listing searches very helpful while doing research for listing our own items.
Beyond these two benefits, we find most other aspects of Cassini concerning. From our perspective as long-time eBay sellers, Cassini views buyers as eBay’s primary customer. The system is built with advantage tipped heavily toward the buyer at the cost of item visibility for the seller. In theory, if eBay provides better matches for buyers, sellers will sell more. By pre-sorting, pre-qualifying and mechanically interpreting each search performed the buyer, eBay believes that they are providing better search results and creating more sales. Unfortunately, this isn’t what seems to be happening.
In their quest to provide a more targeted search result, eBay is, without question, leaving the searcher with fewer results to sift through. What happens to all of the possible matches that have slipped through the cracks in Cassini, making them virtually invisible to the buyer? They will undoubtedly remain unsold. Since the implementation of this program (in addition to other problems behind the scenes, which we discuss here, here, here, here, here and here), there have been widespread reports of falling sales. eBay’s own sales reports verify that things aren’t looking as well as they had expected. (More on that here)
So who is eBay’s primary customer? From my perspective, it should be the us, the sellers. It is the seller who generates sales, creating profit for eBay. That would make the buyers our primary customer. As a seller, if my listings are being filtered out of search results in order to provide a “best match for the buyer, how does that benefit me? And how exactly does that benefit eBay?
In a presentation about Cassini’s development, Hugh Williams, architect of the program shared that the name Cassini came from a satellite launched in 1996. He jokingly said that it is a “Wonderful satellite. The only secret that I’ll tell you about Cassini is that it is stuck in orbit going around Saturn, but we’ll pretend we don’t know that”. (Watch the full presentation here and read our discussion in regards to that presentation here).
Isn’t it ironic that eBay’s Cassini seems stuck in orbit around buyers, at the expense of the sellers?
Sadly, this is only the beginning of Cassini’s woes. The intuitive search engine assumes that it knows the intent of the buyer, eliminating choices which it believes to be irrelevant. In reality, Cassini is offering the buyer fewer choices. An algorithm has given the power to determine what items it “thinks” are being searched for. What if the search engine gets it wrong, and the eliminated items are actually what the buyer was looking for? That buyer may be forced to take their money elsewhere, and both eBay and the seller suffer loss.
Another major problem with eBay’s current search engine is its assumption that certain listings are irrelevant from day one. As I stated earlier, all listings are “pre-qualified” when they first enter the eBay system. Cassini determines, based on data alone, the likelihood that each item will sell and the price that it is most likely to fetch a sale. Basically, if you tend to price your items high, your listing is automatically tagged as “un-sellable”. Once listed, that item must prove its relevance with watchers, views and clicks. As a seller who tends to price their items at the high end of the pack, I ask the following: how does Cassini factor in that my item, with the same basic title and description, has been professionally cleaned? Does it know that my price includes the cost of shipping? Does it think about the high level of care I take while packaging each item with the best packing materials? How does it account for the high level of customer service I provide with each transaction? These are choices that were previously left up to the buyer.
When we look at this aspect of the Cassini system, is it catering to bargain sellers at the expense of the higher-quality sellers? I have openly admitted that my prices tend to run higher, but is it possible that this is justified due to the value of added benefits? Does eBay also lose out on a more successful and profitable sale? In eBay’s system, is my item downgraded because it is not relevant, or is it not relevant because it was downgraded from the beginning?
More problems arise when we realize that Cassini assumes that an intuitive approach is better, thus should always replace a literal approach. According to Hugh Williams, “The way search works today is that we look to deliver the best experience that we can for the majority of buyers that are out there. I’ll wholeheartedly acknowledge that on the edges, we’ll make mistakes. Some buyers come in with different intents. We may not get it right, but we might get it right for the majority of buyers. For those buyers where we missed, you know we have all the great refinement features down the left that I talked about earlier, and of course there is opportunity there to modify the query as well”.
One has to wonder, what would motivate a buyer to modify their search if they have no idea that they are not getting full possible search results? They have no reason to modify their query if they are unaware of additional search options. As a result, they receive inaccurate search results and the items available are not properly represented.
Yet another problem arises when Cassini assumes that my previous behavior reflects my future behavior and responds accordingly. In one presentation, Williams gave a personal example of the weakness of this assumption. He had a friend who’s wife had a baby. He searched on eBay for a baby gift. For him, this was a one-time query for a need that went away immediately after his purchase was made. Yet for the next several months, he continued to receive information in his searches, because the system assumed that he was interested in baby items. Cassini makes erroneous assumptions based on misinterpreted data. As sellers, we experience this on a regular basis. Since we research every item we sell on eBay (and elsewhere) before we list it, Cassini assumes that we are interested in purchasing all of those items. As a result, we receive skewed search results based on inappropriately applied behavioral data. Based on historical data alone, a computer cannot accurately determine the intent of the individual.
Another problem? Cassini is designed to work well with commodity goods (items which are frequently available for sale), not for one-offs (such as rare antiques and collectibles). It is much easier to gather usable data for items that are sold on a regular basis. For example, data concerning a specified iPhone at a specified price is easy to compute and determine sell-ability. It is not so easy to determine the sell-ability of a one of a kind antique piece of furniture at a specified price based on set figures of data. Williams addressed the issue in this way: “It’s a much more challenging problem when a user lists something that’s one-off...but none-the-less, we work very hard on this problem and we do a pretty good job of it”. Unfortunately, this is the only response we have found available to this problem. eBay was built on sellers who sell one-offs, yet we are now working with a system geared toward sellers of commodity goods. It just doesn’t work. It explains why so many veteran sellers are leaving the marketplace.
Yet another problem in regards to Cassini is the fact that those who control Cassini assume that all businesses should grow slowly. Any rapid growth can’t possibly be handled properly, and therefore should be discouraged. They also assume that if your sales are down, it is likely something you are doing wrong. Williams stated:
Some examples I see of sellers having issues are sellers who are trying to scale. So, somebody who’s a really fantastic small seller, running at a small scale, who tries to become a very large seller, often has growing pains as they go through that process…Often those guys come and say, hey, something’s gone wrong with search, “are you doing this? and “are you doing that?” The real issue turns out to be they are having trouble with scaling. They’re typically having trouble with trust. They’re having stock-outs, they’re having trouble with shipping, they’re not able to respond to their questions as quickly as they perhaps once were. So often they’re looking for a very specific answer that’s an eBay answer, but unfortunately the answer is a much broader answer, often, it’s about their business.
So, it seems eBay is doing something to protect us from growing more than we can handle. But how exactly does this occur?
In my own situation, I doubled my eBay inventory in a twelve month period of time. Sales were increasing dramatically. The data eBay had at their disposal indicated that I was increasing my inventory at a rapid rate. What wasn’t included in their data was the fact that, during that time period, I transitioned from doing eBay part-time to full-time. Additionally, I added seven employees to build my business. This enabled me to provide a level of customer service even higher than I already provided. Yet to eBay, it looked like I was taking on too much, too fast. As our sales took a dramatic nose dive, I can only assume that they took whatever steps necessary to slow my growth. Granted, there are many other factors I have presented in previous blog entries that have contributed to our stagnant sales, but this question remains unanswered: does eBay control the rate of growth for businesses and is this justified?
This brings me to my final question: could Cassini ever be reversed? It seems like so much has been invested into the creation of an “improved” search engine that it would take a significant event to reverse its path. This project was a major venture for eBay. At least one hundred engineers and several-hundred-person support staff worked for more than a year to develop the program. What is the chance they would abandon Cassini after that much was put into its development?
Behind the wheel of Cassini there is pride. There is power. eBay finds itself sitting on all of this data. With that data comes perceived power. To leave that data unused would be viewed as a waste. They mine the data because they can. They can’t imagine not mining the data. It’s what engineers live for.
There are two factors, possible glimmers of hope, which lead me to believe the project might be vulnerable. The first is the fact that the architect of Cassini left eBay before the roll-out of the program was even completed. Secondly, eBay is in the process of reversing many aspects of another major policy just released last year: the feedback system.
For the moment, I set my hope in the possibility of change.
See the rest of our eBay series here:
Part I - Why are My eBay Sales Down?
Part II - eBay Takes a Hit
Part III - Google Takes Aim
Part IV - eBay Alternatives?
Part V - Fixing the Defects in eBay's Defect System
Part VI - Breaking Up is Hard to Do: eBay's Split from PayPal
Part VII - Is eBay's Cassini Really the "Best Match"?
Part VIII - Is eBay's Cassini Stuck in Orbit?
The Plan, Part I - Positioning for More than Survival
The Plan, Part II - Expanding Our Reach Beyond eBay
Frankly, I am not fond of monkeys. They affect me the way spiders and snakes affect other people. The flying monkeys in “The Wizard of Oz” and the rogue monkeys in Robin Williams’s “Jumanji” were menacing to me, and I closed my eyes so I wouldn’t have to see them.