Announcement: New E-commerce Product Search Engine Implementation
Comments
-
Nate Merritt (Logos) said:
Inaccurate search result counts is one of the limitations of the new system.
This is not the issue. If the new algorithm can't count large numbers, simply say ">100". One of the problems is the restriction: if one scrolls down to the last one, there's no "more" button to get the rest. But this only applies to some searches.
All those ordered by progress, by price and by savings need to be correct orderings of the whole catalog. Many users, myself included, use those all the time. Of course you can cache those somewhere and have the first 1000 of those delivered first for selection (which will miss the for me most relevant aspect of the ordering by price: dynamic price based on my ownership).
Actually, when ownership really is an issue, you got the wrong eCommerce provider. This is not a shop where we might buy the same product again and again. We always buy the things we don't yet own.
Have joy in the Lord!
0 -
I just keep asking myself, in all due respect but as a long-standing customer, what are we getting out of all this? There are a number of deficiencies that may not go away. What do we get - first page loads 1-2 seconds faster? Counts of resources are not more accurate, facets don't work better/hopefully they will work as good eventually, searches don't seem more accurate..
I re-read the first announcement - We have heard your feedback over the years about our custom e-commerce product search implementation and your desire for better search results. Therefore, our development team has been working this year to migrate to a best-in-class third-party e-commerce search solution
"Better search results". IMHO better search results means more accurate and includes accurate result COUNTS. This does NOT say feedback shared that the initial page(s) should come faster, especially at the cost of inaccuracy, which violates the premise (more accurate results) of this effort!
It seems like the whole crux of this thread are issues, acknowledgements, and intentions to improve the SEARCH RESULTS to be AS GOOD as before.
So really - what do we get out of this as customers? It seems like a misguided effort to me that is not directed at what feedback shared with Logos. It certainly isn't an improvement to me, I am hoping eventually it can be as good albeit with compromises. Very disappointing.
0 -
While this system is so badly broken in so many ways, can we please go back to the old system, at least until you've fixed most of the problems with the new one? Please!
0 -
Don Awalt said:
So really - what do we get out of this as customers?
We get search-lite, leaving us in the demi-monde.
Search-lite saves them computer costs, as each of us does not get a custom search. Owned Already does not work as first click, so does not
customize one's search.
And they want a monthly fee payment.
0 -
Don Awalt said:
"Better search results". IMHO better search results means more accurate and includes accurate result COUNTS. This does NOT say feedback shared that the initial page(s) should come faster, especially at the cost of inaccuracy, which violates the premise (more accurate results) of this effort!
It seems like the whole crux of this thread are issues, acknowledgements, and intentions to improve the SEARCH RESULTS to be AS GOOD as before.
Right.
I hadn't tested this yet, but just for fun I thought I'd try it to see what all the fuss is about. The experience inspired a couple thoughts.
1) When I get lost while driving my car, I'll just drive faster knowing that I might not get to where I want to go, but at least I'll get there faster.
2) In fact, if there results are going to be wrong, why not just throw up a static web page. It'll be wrong, but at least it'll be super fast and more predictable.
My comments might not help much, but it does express my bewilderment in the change foisted upon us. If it's a third party thingy they're paying for, I wonder if they're still in the warranty period.
0 -
Nate Merritt (Logos) said:
If you apply more specific filters or search queries after applying the "Unowned" filter you will likely see more accurate counts because there are fewer products for the system to count.
Note that the returned results are accurate. You should not be seeing any owned products when applying the "Unowned" filter. You should also be able to page through many pages of results without issue.
This is not right. I have a particular commentary on my wish list. If I search for "Unowned" "Commentaries", I get 214 results and as I scroll through the 4 pages, my wish-list commentary is not there. There is no ">" to click on at the end to get more result, and I cannot make it appear by adding more filters such as the publisher or author, because the correct options are not listed. Adding filters like "Live", "English", "Single Title", "Old Testament" still does not do it. Some of those give more results than the facet promised, but the commentary that I'm looking for is never there. In no case are you showing me all the relevant results for my search. Your 1000 limit cuts it out early in the process so that it never comes up. If I did not already have this item on my wish list, I would not be able to find it by searching. Hiding relevant products like this seems counter productive for a company that makes its money by selling books!
0 -
One item that has long frustrated me with the store (and every area of Logos that has filtering) is sorting the options by count rather than alphabetically. If I'm looking for dictionaries I should be able to look under 'D', not '320'. Oh, wait...I just refreshed the screen and now it's '309'.
0 -
Paul Gibson said:
One item that has long frustrated me with the store (and every area of Logos that has filtering) is sorting the options by count
Unfortunately, it is the industry standard that facets are sorted by count and a search box provided to quickly find lower count items. Yes, it is frustrating ... but it would also be frustrating to have it work differently than your other applications. I have no solution. I personally use facets sparingly as a result.
Orthodox Bishop Alfeyev: "To be a theologian means to have experience of a personal encounter with God through prayer and worship."; Orthodox proverb: "We know where the Church is, we do not know where it is not."
0 -
Lew Worthington said:
1) When I get lost while driving my car, I'll just drive faster knowing that I might not get to where I want to go, but at least I'll get there faster.
2) In fact, if there results are going to be wrong, why not just throw up a static web page. It'll be wrong, but at least it'll be super fast and more predictable.
Ha ha! Great analogies.
0 -
The new search engine is simply bad. Your explanations don't help. We need accurate results not whatever the engine wants to spit out quickly.
It seems you are trying to solve problems almost no one has raised. In the process you've broken a system that, while not perfect, worked.
Roll back the change, please, then get a group of users to tell you what they would really like to see improved in search and see what you can do to deliver that.
Please.
Pastor, North Park Baptist Church
Bridgeport, CT USA
0 -
Not defending Nate, but in fairness, their issue SEEMS to be the servers are struggling with high volume (Nate's initial comment). Their solution SEEMS to be, to deliver a high-value subset per user, thereby delivering results quicker (during high use).
And the logic isn't there. Though putting production changes out there with obvious bugs is odd for Faithlife? Pressure from?
"If myth is ideology in narrative form, then scholarship is myth with footnotes." B. Lincolm 1999.
0 -
We used to type an author's name and get quick matching options. Same with most kinds of searches. Now, there is no 'drop down' options appearing and results are terrible, general and mixed. No suggestions, no matching author (or anything) name, nothing. It's like doing a very general search on g*ogle.
What happened to the search? It has become horrendous!
All the best in Christ.
0 -
Mateus de Castro said:
We used to type an author's name and get quick matching options. Same with most kinds of searches. Now, there is no 'drop down' options appearing and results are terrible, general and mixed. No suggestions, no matching author (or anything) name, nothing. It's like doing a very general search on g*ogle.
What happened to the search? It has become horrendous!
I do get a "dropdown" menu when I type an author's name.
But I agree with you that author searches are terrible now. It used to be the default would be to take you to that author's page with a listing of books by that author below the box with their bio. That doesn't seem to work anymore, even if I put quotation marks around the author's name.
Only if I then click on the facet for that author does it finally show me the author's page & bio box.
This is a step backwards.
0 -
Thanks for the reply. Strangely, I don't get the drop down options. I get nothing relevant in the search. I've been trying for days on Logos, Verbum websites and nothing of the sort anymore. Searching has been a drag. It was really useful when you didn't remember names perfectly.
I hope the search goes back to what it was. Used to be one of the best, now it seems like adapted for marketing purposes, but actually driving me away. I used to spend some time searching and updating wishlists. Now I just give up.
P.S. - not sure there's a thread for this, but wishlists could get some updating in the organizing capabilities.
0 -
Logos, if you are listening, please take us back to the old search. The consensus here seems to be that whatever the imperfections of the old search, the flaws of the new search are far worse.
0 -
Logos is "fixing" the Search Engine in preparation to starting a new subscription plan.
But don't poke it too soon, it's only half-baked.
Seriously, I think they might fix it. A third-party...party...party....I don't know if they will fix it.
Wait in silence. Something might happen.0 -
NB.Mick said:Nate Merritt (Logos) said:
Inaccurate search result counts is one of the limitations of the new system.
This is not the issue. If the new algorithm can't count large numbers, simply say ">100". One of the problems is the restriction: if one scrolls down to the last one, there's no "more" button to get the rest. But this only applies to some searches.
All those ordered by progress, by price and by savings need to be correct orderings of the whole catalog. Many users, myself included, use those all the time. Of course you can cache those somewhere and have the first 1000 of those delivered first for selection (which will miss the for me most relevant aspect of the ordering by price: dynamic price based on my ownership).
Actually, when ownership really is an issue, you got the wrong eCommerce provider. This is not a shop where we might buy the same product again and again. We always buy the things we don't yet own.
I can appreciate that you are seeking to create a more efficient sales page that doesn't tax your servers as much as they have in the past. This is commendable.
But please do not limit my access to the entire catalog. Surely there is a way to do this in a manner such as that which NB.Mick suggests. When I search for items that are unowned by me, I want to see all of them so that I can choose. I do not want the system to “suggest the best products for you based upon your previous search history.” I have never liked that from other online retailers, and I definitely will not like it when it comes to my personal choices for the books that I use in my study and ministry. There are times when I enjoy searching page after page for some unknown gem that I might purchase. Trust me, I have yet to find an online retailer that can do that better than I can myself.
Please do continue to send me ads about certain books that you think I might be interested in, and I will continue to look at them and ponder them; but please do not limit or choose which books I may or may not look at in the online storefront.
Considering performance to be more important than accuracy or consistency is so counterproductive to what I need when I am browsing for books to purchase.
Again, I appreciate the need to seek to improve, but I have to say that this is not improvement. Please, I ask sincerely, return to the old storefront.
Above all these things, walk in love, which is the bond of perfection. - Colossians 3:14
0 -
Rosie Perera said:
Only if I then click on the facet for that author does it finally show me the author's page & bio box.
Question: How fast can one get to author's page + bio box? >> Experiment: Which names get to author's page + bio box?
answer Yes: the name gets there....................answer No: the name does not get there
Yes: J Piper, John Piper
Yes: John Frame....................No: J Frame
Yes: N T Wright....................No: Tom Wright..........[Everyone series has Tom Wright on the cover]
Yes: T Schreiner, Tom Schreiner....................No: Thomas Schreiner, Thomas R Schreiner
: I looked at title page of 6 Schreiner books I have..........One listed as Thomas Schreiner
..........Five listed as Thomas R Schreiner
Say Person 1 shows Person 2 a softcover at a Bible study. #2, new to Logos, goes home and looks up Tom Wright. This leads to Tom's author page + bio box. That's the way it should be!!! Today, this connection doesn't happen.
I'm sure these uneven results above predate this week's boondoggle. But they do add to the carnage.
Logos: Why not straighten these out [and others like them] as you repair the search. Fix the catalogue, too, please. Thank you.
0 -
Sam Shelton said:
I can appreciate that you are seeking to create a more efficient sales page that doesn't tax your servers as much as they have in the past. This is commendable.
But please do not limit my access to the entire catalog. Surely there is a way to do this in a manner such as that which NB.Mick suggests. When I search for items that are unowned by me, I want to see all of them so that I can choose. I do not want the system to “suggest the best products for you based upon your previous search history.” I have never liked that from other online retailers, and I definitely will not like it when it comes to my personal choices for the books that I use in my study and ministry. There are times when I enjoy searching page after page for some unknown gem that I might purchase. Trust me, I have yet to find an online retailer that can do that better than I can myself.
Please do continue to send me ads about certain books that you think I might be interested in, and I will continue to look at them and ponder them; but please do not limit or choose which books I may or may not look at in the online storefront.
Considering performance to be more important than accuracy or consistency is so counterproductive to what I need when I am browsing for books to purchase.
Again, I appreciate the need to seek to improve, but I have to say that this is not improvement. Please, I ask sincerely, return to the old storefront.
This is an important point by Sam. I may add that if Logos diminishes its research profile in exchange for a more 'popular readings' seller, it greatly loses value to many. When I do a search, I want to find what I don't know; rare books; new perspectives etc. If a theme comes back crowded with 'popular books' on the search, Logos becomes another book seller that, on the side, has great books for research and study. But those I must find on my own, by looking for it exactly, not by discovering on a more focused search.
0 -
Mateus de Castro said:
if Logos diminishes its research profile in exchange for a more 'popular readings' seller, it greatly loses value to many.
Amen to that!
I really dislike "popular" as a sales tactic. I don't care what lots of other people are buying. I want to find what *I* would want to buy. To quote Brian in "Life of Brian" -- "You are all individuals!" Let's not chant back "We are all individuals!" but rather get to buy what is appropriate for who we are and what we are interested in. Say "no" to mass society. Even if it does make more money for the sellers. It's insidious.
0 -
Rosie Perera said:Mateus de Castro said:
if Logos diminishes its research profile in exchange for a more 'popular readings' seller, it greatly loses value to many.
Amen to that!
I really dislike "popular" as a sales tactic. I don't care what lots of other people are buying. I want to find what *I* would want to buy. To quote Brian in "Life of Brian" -- "You are all individuals!" Let's not chant back "We are all individuals!" but rather get to buy what is appropriate for who we are and what we are interested in. Say "no" to mass society. Even if it does make more money for the sellers. It's insidious.
Indeed! Remember what happened to "The Great Courses" just the other day. Profile matters! They used to be about 'academic-level content for everyone', but became 'Wondrium', tried to 'popularize' their content and profile, and the people hated. They became just another site and had to revert back to "The Great Courses" and promise quality again.
Maybe Logos should take it as a cautionary tale and stop pushing 'how to make money' books (I should have seen the changes as I saw those on pre-pub and my searches) and crowd our searches only (!) with 'popular' books.
0 -
Mateus de Castro said:
We used to type an author's name and get quick matching options.
Thank you for pointing out this regression. We did have as-you-type suggestions for author and publisher names in the previous search implementation but they were missed when creating the new implementation, which only supports title suggestions. We plan to add this feature back to the new search.
0 -
Jordan Litchfield said:
Logos, if you are listening, please take us back to the old search. The consensus here seems to be that whatever the imperfections of the old search, the flaws of the new search are far worse.
I'm not part of the Commerce team, so I'm not directly involved in this project.
But I want you to know that the team is certainly listening to all your feedback and are doing their best to figure out a solution.
Part of the challenge is that the new system typically performs noticeably better for simple searches than the old system. So, the team is scrambling to see if there's a way in which we can retain this search improvement without sacrificing accuracy when you drill down through the facets.
And if it's any consolation (I suspect it won't be!), I'm a customer, too, and this change has disrupted the way I search for books too! So be assured that we understand your pain firsthand and appreciate how disconcerting it is when the search erroneously tells you it's given you all 37 books that match your criteria, but you know for certain that there are at least 12 more it's missed.
0 -
Mark Barnes (Logos) said:
I want you to know that the team is certainly listening to all your feedback and are doing their best to figure out a solution
Thanks for your comments, Mark.
This project should not have been released till it ran well. Internal testing + fixing is the key. To me this is obvious, but to Logos t'was oblivious.
0 -
Mark Barnes (Logos) said:
Part of the challenge is that the new system typically performs noticeably better for simple searches than the old system. So, the team is scrambling to see if there's a way in which we can retain this search improvement without sacrificing accuracy
Thank you, Mark. You have just explained why I find the new system much more accurate than the old. When I use the Logos site, I am usually looking for something quite specific - the old system gave me a lot of crap results and often missed an exact title match. Except for the monthly free ebooks, I rarely use facets. I feel the new system was implemented before the kinks were worked out - I've been in that position myself where external people set unreasonable implementation dates.
My biggest problem with the storefront is that the software still misses books I consider critical to Logos and to building a market for an expanded Logos community. The reason it misses them it that Logos doesn't carry them. Something like The Religion of Whiteness: How Racism Distorts Christian Faith by Michael O. Emerson & Glenn E. Bracey II needs to be in Logos NOW when adult book discussion groups are using it. [Yes, I bought it from Amazon] Next year, the hot discussions will have moved elsewhere.
Orthodox Bishop Alfeyev: "To be a theologian means to have experience of a personal encounter with God through prayer and worship."; Orthodox proverb: "We know where the Church is, we do not know where it is not."
0 -
Thanks Mark, for helping to put our minds at ease. If it is any consolation to you, it does mean something for me to remember that you are a customer as well. That is why we were so excited when you first came on board with Logos.
Above all these things, walk in love, which is the bond of perfection. - Colossians 3:14
0 -
My limited "tests" of the new search seem to be mixed.... I may agree with the possibility that searches more often than not have been more accurate. Though they seem to be much slower, which may be tied to the increase in accuracy.
The release before being ready fits into the current Logos we are dealing with.... As does the response that all the feedback is being listened to and the solution is being worked on.....(Though Logos.com search function is something that carries much less weight for me as a user)
Then again..... I'm still waiting for an up to date Forum that can be navigated on a mobile friendly platform - most other forums have had that capability for many years.... Considering the amount of free user tech support that the Forums provide - you'd think it be worth the investment of time and money is needed.... But that's probably still being worked on LOL
Improved Forums functionality would be a bigger plus than website search and AI related projects for this user....
Logos 10 - OpenSuse Tumbleweed, Windows 11, Android 16 & Android 14
0 -
Mark Barnes (Logos) said:Jordan Litchfield said:
Logos, if you are listening, please take us back to the old search. The consensus here seems to be that whatever the imperfections of the old search, the flaws of the new search are far worse.
I'm not part of the Commerce team, so I'm not directly involved in this project.
But I want you to know that the team is certainly listening to all your feedback and are doing their best to figure out a solution.
Part of the challenge is that the new system typically performs noticeably better for simple searches than the old system. So, the team is scrambling to see if there's a way in which we can retain this search improvement without sacrificing accuracy when you drill down through the facets.
And if it's any consolation (I suspect it won't be!), I'm a customer, too, and this change has disrupted the way I search for books too! So be assured that we understand your pain firsthand and appreciate how disconcerting it is when the search erroneously tells you it's given you all 37 books that match your criteria, but you know for certain that there are at least 12 more it's missed.
Mark, IMHO no one has really answered a question I and others have posed - why? Faster searches? No one asked for that! Why are we going through all this, for feature(s) it's not evident we have asked for? The overwhelming response here is that the old search was better, albeit with some issues. Why is this happening? It seems to me there is some underlying motivation driving this.
Think of it - in the face of all these complaints, the replies are to work on implementing improvements. There seems to be no plan other than plow ahead. Very disappointing from a customer perspective. And who let this get released in this state? Was there insufficient testing maybe? It's in production now? For shame.
It really seems Logos is plowing ahead to implement a new system for reasons of their own, and customers are suffering. How about a little transparency here?
There are without a doubt good people working at Logos, but this was a bad plan, bad implementation and testing. Someone should be held accountable for rolling this out prematurely. And for what reason, we continue to ask.
0 -
Mark Barnes (Logos) said:
Part of the challenge is that the new system typically performs noticeably better for simple searches than the old system. So, the team is scrambling to see if there's a way in which we can retain this search improvement without sacrificing accuracy when you drill down through the facets.
Thanks for posting Mark. I agree that the revelance sorting has improved. I do not notice a performance difference. I strongly dislike the limitation on results.
I was once an expert in database internals with a focus on performance. I strongly doubt that the 1000 row restriction is necessary. It looks like your developers are making rookie mistakes, and it is very frustrating.
0 -
Don Awalt said:
It really seems Logos is plowing ahead to implement a new system for reasons of their own, and customers are suffering. How about a little transparency here?
Sounds like reducing server costs. I can't believe better searches, even claimed simple ones.
"If myth is ideology in narrative form, then scholarship is myth with footnotes." B. Lincolm 1999.
0 -
Nate Merritt (Logos) said:
Thank you for pointing out this regression. We did have as-you-type suggestions for author and publisher names in the previous search implementation but they were missed when creating the new implementation, which only supports title suggestions. We plan to add this feature back to the new search.
Thank you! It was hard to believe that the option was gone. Good to know it will be back. Hope sooner than later, but I'm glad.
0 -
Sorry for my ignorance, I have no idea what this feature is. What I know is that I can just search for a resource by name and that is all. Would you please explain to me what it is apaart from what I assume it is?
Blessings in Christ.
0 -
Tes said:
Sorry for my ignorance, I have no idea what this feature is. What I know is that I can just search for a resource by name and that is all. Would you please explain to me what it is apaart from what I assume it is?
Tes, it's not a feature. What you assume it is is exactly what it is. It's just that they have changed the way search is implemented "under the hood" (or inside the code that we don't see).
So the user interface for search in the store hasn't changed at all. You still search for a resource by name (or by the name of an author, or by any keyword if you're looking for books on a particular topic). You still might get better results if you put the title in quotation marks (same as before).
What has changed is how it computes that search. Supposedly it's faster now, though to be honest I had not noticed it being annoyingly slow before, so I don't know why they needed to upgrade the algorithm. That was apparently their motivation for changing to a third-party search function from what they had before. But as you've read in this thread, it has introduced some inaccuracies that are bothering many people.
To be absolutely clear, what we are talking about is the Search box that appears at the top of logos.com:
You previously asked for a training video for this. There is nothing to it. Just type in text in that search box and hit Enter. If you know the name of a specific book you are looking for, type in the title (ideally in quotation marks) such as, "The New Testament and the People of God".
Or if you just know of an author and want to find books by him or her, type in the author's name (again, ideally in quotation marks), such as "N.T. Wright".
If you are looking for books on some topic, type in the topic (maybe in quotation marks if it is a complete phrase), such as Pharisees or "universal salvation".
You can also search for publishers, which I do often if I'm looking to see if there are any other books from my favorite publishers that I don't own yet, for example "Herald Press".
You can also use the Filters on the left-hand side to narrow down the search results. Sometimes you'll hear people talking about "facets" -- these are just the items under the sections titled Topic, Resource Type, Author, Status, Language, Publisher, Genre, Christian Group, etc.
Ownership Status and Geographic Availability are not technically "facets" -- but they do affect the filter.
When you select a facet, it will show up at the top of that left sidebar area to remind you that you're filtering on that facet:
Topics you can only select one of. The other facets with checkboxes next to them you can click however many you want in each section, so for example, you can filter the results to show all books on a given topic from any of your three favorite publishers.
I hope that helps. Sorry, but there is no video showing how this works.
0 -
Thank you Rosie you have wonderfully made it clear.
Blessings in Christ.
0 -
Oh, and I forgot the obvious: if you click the X on a facet at the top of the sidebar, it will remove that facet/filter.
0 -
Frank Sauer said:
Improved Forums functionality would be a bigger plus than website search and AI related projects for this user....
Watch this space!
0 -
Don Awalt said:
It really seems Logos is plowing ahead to implement a new system for reasons of their own, and customers are suffering. How about a little transparency here?
I think the Commerce team is trying to be transparent, but I'm happy to add some additional detail.
Logos has been in business for 32 years. We have a LOT of systems that have been built in-house and have evolved over time, and our store is a big part of that. When we first started selling books online, we were one of the pioneers. Developing an in-house system was the only option!
But there are some really hard problems to solve in e-commerce, and there are now a lot of specialist providers who are better at solving those problems than we are. We want to be a company that specializes in Bible study, not a company that specializes in building eCommerce platforms. Maintaining a completely custom eCommerce platform that doesn't use any off-the-shelf components is an expensive business that diverts developer time away from our core products. Moreover, it's expensive to train new developers to work on that system because literally no one outside the company is familiar with it.
So, the Commerce team has been looking at ways in which we can replace some (probably never all) of our commerce platform with customized off-the-shelf components. This transition to Algolia search was part of that process.
The team had been quietly A/B testing the new search engine in the background for some time. That means some searches were sent to the new search engine and some to the old one. They measured various signals to see if there was an improvement, and they found that the new engine was faster and displayed more relevant results when people typed in text in the search box. As a result, on average, people found the books they needed more quickly with the new engine than the old engine. As a result, it was decided to go ahead with the new engine.
What wasn't clear from those tests was although the new engine was measurable better for the majority, it was significantly worse for a few, and could feel entirely broken when drilling down through the facets. That's what has now surfaced through this thread, and this is what the team is wrestling with.
I agree that this has been a poor experience for customers with large libraries who drill down through results to find what they need. I'm in that category myself. The team now understands how important that is for us. But, I'll repeat, the motivation is that we want to provide a great buying experience for everyone while being a company that specializes in Bible study, not a company that specializes in building eCommerce platforms.
0 -
Thank you for that explanation, Mark.
0 -
Steve Adams said:
It looks like your developers are making rookie mistakes, and it is very frustrating.
I agree it's frustrating, but it's not about developers making rookie mistakes. It's about trade-offs, and probably we've got the balance of those tradeoffs wrong.
As the post above this one explains, we're no longer using only a database to power this search, but a respected third-party search engine (Algolia). Doing so has gained us increased relevancy but at the expense of completeness. That was the tradeoff, and what the team is now trying to do is to see whether it's possible to create a scenario where some don't have to sacrifice quite so much in order for everyone to gain the benefit of greater relevancy.
0 -
Well I finally learned what a facet is. Thanks Rosie, above.
Re in quotes: Was this best practise weeks ago, or is it only now with the new search function??
Ideally, it should work perfectly well without quotes. SIBKIS: see it big, keep it simple. Quotation marks complicate things, + show a lack of thought for the customer.
0 -
Mark Barnes (Logos) said:
The team had been quietly A/B testing the new search engine in the background for some time.
I had been noticing the problems for some time too, but refreshing the page enough times normally worked. If the team had done their testing less quietly, we may have been able to give them the necessary feedback.
I know it's not your baby, but if it were mine, I would have backed out the change temporarily. Are you aware of why that has not been done, despite some significant customers being very unhappy? Is there any chance of that being done now, until the bugs are fixed? Thanks.
0 -
I'm not the fancy user, with the large library. More like basic user with specific needs. I can understand an a/b might work using categories, publisher blurbs and even similarity matches. Logos could have written that, years back.
But as it stands, it's like asking your child to go find the TV remote. Before, he just looked around ... everywhere. Now, he looks around in likely places. And stops. No remote!
It illustrates the absolute necessity of search-trust. Not 'trade-offs'.
"If myth is ideology in narrative form, then scholarship is myth with footnotes." B. Lincolm 1999.
0 -
DMB said:
I'm not the fancy user, with the large library. More like basic user with specific needs. I can understand an a/b might work using categories, publisher blurbs and even similarity matches. Logos could have written that, years back.
But as it stands, it's like asking your child to go find the TV remote. Before, he just looked around ... everywhere. Now, he looks around in likely places. And stops. No remote!
It illustrates the absolute necessity of search-trust. Not 'trade-offs'.
a/b means what?
Great search analogy!
0 -
scooter said:
a/b means what?
https://en.m.wikipedia.org/wiki/A/B_testing
It can be considerably more sophisticated (eg customer-targeting, margin maximizing, supply-chain optimization, etc).
"If myth is ideology in narrative form, then scholarship is myth with footnotes." B. Lincolm 1999.
0 -
DMB said:scooter said:
a/b means what?
https://en.m.wikipedia.org/wiki/A/B_testing
It can be considerably more sophisticated (eg customer-targeting, margin maximizing, supply-chain optimization, etc).
Thanks. I'll dive in, shallowly.
0 -
Steve Adams said:
I know it's not your baby, but if it were mine, I would have backed out the change temporarily. Are you aware of why that has not been done, despite some significant customers being very unhappy? Is there any chance of that being done now, until the bugs are fixed? Thanks.
There is a chance of that being done, yes. It's under active discussion. In situations like this there it's often a difficult question as to whether the team invests time in rolling back, then fixes the main issues, then deploys again, or whether they just fix the issues while it's live. Given that many users were benefitting from the improved relevance, I don't think that was an easy decision to make – especially when the full range of issues took some time to understand.
0 -
The most frustrating aspect of searching on the website for me is that if one word is misspelled, the item being searched will not show up.
If I search what the bible really says about God's heavenly host, I get the correct result.
If I misspell "the" as "hte," I get no results.
So what hte bible really says about God's heavenly host returns no results.
Interestingly, the software has this same issue when I search my library.
I would think the search would be smart enough to suggest the correct title based on all words given.
Disclaimer: I hate using messaging, texting, and email for real communication. If anything that I type to you seems like anything other than humble and respectful, then I have not done a good job typing my thoughts.
0 -
Joseph Turner said:
The most frustrating aspect of searching on the website for me is that if one word is misspelled, the item being searched will not show up.
If I search what the bible really says about God's heavenly host, I get the correct result.
If I misspell "the" as "hte," I get no results.
So what hte bible really says about God's heavenly host returns no results.
Interestingly, the software has this same issue when I search my library.
I would think the search would be smart enough to suggest the correct title based on all words given.
I agree that this is frustrating, and I deal with this categorically across many things. For example, Spell-Check on my phone constantly seems to have silly issues like this.
That said, I was able to quickly and simply resolve this specific issue. So, if you search what hte bible really says about God's heavenly host you'll find that you get your expected results.
0 -
Steve Adams said:
Here is another illustration of the 1000 product limit being applied too soon.
If I search for all "Unowned" products, ordered by "Price (lowest first)", I should get the cheapest 1000 products that I don't own.
Instead, you fetch 1000 products in database order, exclude the ones I already own, leaving only 306, and then sort them by price.
The count IS accurate. I can only scroll through 306 products. Most of the cheap books that I do not own are not shown.The team has been focused on the 1,000 result limit, as this seems to be the largest issue raised here. After continued digging and testing we were able to find a way to raise the limit from 1,000 to 20,000 which is the system maximum. We expect this greatly improves your ability to discover unowned products.
We greatly appreciate your continued feedback, and are continuing to work to improve your search experience.
Some additional notes:
- Speed was not necessarily the motivating factor for this change. Speed was discussed as a reason that our search solution provider implements a 1000 item limit in certain instances. We've received numerous reports of irrelevant search results over the years, and relevancy is a measured improvement as Mark noted. As Mark also noted, it was certainly not the intent to improve relevancy at the expense of important use cases for users such as those in this thread who are understandably frustrated by the negative impact this had on expanding one's already extensive library by discovering more unowned products.
- In addition to relevancy, our previous solution was difficult get insights on, and difficult for us to improve. Our new solution makes it easier for us to self-identify problem areas, and solve problems. e.g.,
Michael Fisher said:SSST and Short Studies in Systematic Theology are now synonyms and a query like ssst justification will bring up the mentioned title in results. We'll be working to further improve cases where acronyms vs full phrase (or opposite) is used in titles and/or descriptions to improve results.
Michael Fisher said:That said, I was able to quickly and simply resolve this specific issue. So, if you search what hte bible really says about God's heavenly host you'll find that you get your expected results.
0
