I'm recoding a lot of price search features lately. There's one thing Steph and I talked about today that I would like to get a little feedback on before I jump into making it reality.
For those of you that use the cart functionality here would you like the ability to work with multiple carts? That way you could have an 'active' cart and an array of saved carts. I see that there are some people using the Kickback program to place multiple orders with different stores on a regular basis. Would it me nice to plan your orders and save them as a cart? Think of it as a wishlist of sorts that you can recall to be your active cart and then transfer it to a store.
Just a thought. If you like it or think it's a silly idea let me know. I'd hate to spend time working on something that no one wants to use.
BTW, this is just one of many new features. I'm not ready to post the rest of the list just yet though.
it's a great idea. I'm generally using my cart as a wishlist, and when I see some DVD that looks interesting I simply add it to my cart to be able to check it out some time later. So the content of the cart can sometimes become very much, and after checking which store is the cheapest for which DVD, for placing an order I have to set the DVDs I'd like to order from other stores to zero-quantity and increase the quantity again once I've transferred my cart to the store and deleted the items just transferred from the cart. This can become quite time consuming.
I'd still like to use the cart as a big wishlist, so the perfect functionality - in my eyes - would be to have one general cart one could use as a big wishlist and another cart for every store I'm ordering from. Then it should be possible to move the items to other carts which then can be transferred to a store.
Although I guess it's quite some work to code I'd appreciate it a lot
I'm also looking forward to the international shipping costs being added to the price comparison. I'm actually doing this manually at the moment and sometimes $15 less shipping costs makes the difference that gets me to order from another store than the one the price comparison calculated as being the cheapest.
I envision the multi-cart functionality to be much like a copy/paste operation. The idea is that you can save your current cart to Cart 1, make changes to the working cart, save the modified cart as cart 2, make changes, save as cart N. Later you can recall cart X to become your working cart. Cart X will not be changed unless you save the working cart over the contents of Cart X again. If I do it this way I won't have to set a limit on the number of carts you can maintain.
I forgot to mention that when you save a cart you will be able to assign a text description for it's contents/purpose to make it easier to recall it later.
Drew, What a great idea. That would save alot of time. Would it be possible that you have your ADT Cart (big wish cart) that you could see maybe the top three cheapest prices? and be able to select which ones go to the store of your choice? Maybe that idea with the top three cheapest stores might separate the other more expensive store ??
I'd rather not mix in price information with the cart contents. I know at most stores they show you the items in your cart, the quantity and a price. It would become very confusing for most users if I listed a single item in the cart many times with different prices. That's what the total order comparison page is for. I think it's a more logical place to display pricing information.
Your multiple cart idea, and the way you have described implementing it, sound very useful to me.
The last time I placed a mass order, I found that I had to go to my master list, and systematically remove all the items from it which I didn't want to order right then, or could be more cheaply ordered from a couple of different online vendors.
Then I had to re-enter the deleted items when I wanted to get them from a different ventor.
One thing you haven't mentioned I don't think but would be very good is this. In addition to the ability to rename the whole cart, you should allow moving a particular item to a different cart.
So in addition to being able to delete it, or decrease it's number of copies wanted to 0, one could choose move it to another cart. (For a different store, or a particular store one wanted to order from right then.) Ideally one could move a highlighted group of titles to the same other cart all at once.
I also mentioned the possibility to move the items from one cart to another in my last post. Sorry if this didn't get clear. In fact I wrote exactly the same message as dexplorer did
Just wanted to emphasize again that I'm with him concerning the cart features.
I basically keep everything in my large "wishlist" cart, then when I'm ready to order, I transfer the whole thing over to the price check, so I can have a glance at what things are going for at my fav stores.
I then check stock, and transfer the whole thing across. Once I'm in the store, I trim the stuff that's out of stock, the stuff that I'm waiting on price drops for, and then begin the cull of the things I actually want to get, eventually ending up with an order I can actually afford.
As long as the ability to do this doesn't get changed, (and it appears that it will not), then I'm a happy lil' camper.
Just wanted to throw in my two cents -- I think this sounds like an excellent idea, especially with the 'move to other cart' functionality described by a few other posters. I often put together a big list, but find that if I split it between two stores, I can save a lot of money. Then the pain in the ass part begins -- removing and adding and rechecking to make sure I didn't fuck up and whatnot. This would eliminate all of that and only increase my slavish devotion to ADT as the Mecca of all true porn fans everywhere!
Drew, I would like to be able to update the quantity of more than one title at a time within the comparison cart. Rather than "+ - X" options that work immediately upon clicking them, it would be nice to either have a box where the quantity could be manually entered or have the "+ - X" buttons be the type of button that saves the status until an "Update Quantity" button gets pressed.
I realize that some of the need for this process could be replaced by the multiple cart option, but I think tht while playing games with exploring different pricing combinations between multiple stores, that one cart (that is just easier to update quickly) could be useful.
Wow it's been a really long time since the original multicart concept was tossed around. Here's an update:
I've been working on building the mother of all comparison functions off and on for the past month. I was working towards creating a comparison function that would automatically determine the best mix or orders from based on the contents of your shopping cart. The idea was that if you had 10 items in your cart the comparison function would automatically determine and split the cart contents into the most econimical orders taking into consideration coupons, shipping, tax, Kickbacks, etc. The concept works great for carts with up to 6 items. Beyond 6 items the recursive nature of the necessary logic starts to take a while to do its thing. I've scrapped the mother of all comparisons idea for the time being until I can figure out a better method.
So now I'm revisiting some of the original discussions to be sure I'm covering everything with the rewrite of the shopping tools.
One thing that I would like to do is combine the shopper profile data and the alert profiles. This will enable me to do cart-contents based alerts in the future. (i.e. tell me when this order is available for $X or less or when store Y carries all of these items.) Combining these features would require a login step to modify cart preferences and shopping preferences. (A persistent cookie option could keep you logged in.) Alert profiles and shopper profiles are not currently assosicated in any way in the database. Does anyone object to me creating a single profile?
So are you thinking of combining the multi-cart idea and the alerts? Could be quite interesting.
Also, I'm gonna do a little digging for you and find out if anyone has an efficient algorithm for doing your mother of all shopping carts price minimization. I know other engines do this so there has to be a way to do it without killing the box. Anyway, it can't hurt to look.
Other engines do not do what I was attempting to do. I checked because I was hoping to find something to learn from. Why reinvent the wheel if you don't have to? Because of the complexity of prices, tax, shipping, coupons and Kickbacks I had to recursively calculate the total order cost for all possible combinations of N items at each store carrying those items.
As an example, consider free shipping. Some stores do it for orders over $X.XX. At some stores it's free all the time for the slowest method. Other stores free shipping is a function of the number of items in the order. (i.e. 4 or more items and shipping is free.) Sometimes free shipping is in the form of a coupon good only for orders over $X.XX. With shipping costs being a fairly significant portion of the overall order cost for many retailers you can see why this is so important.
Now think about building the logic to facilitate placing an order for items A,B,C and D at 20 different retailers each with their own shipping cost structures. Lots and lots of iterations. I may still do it at somepoint and have it be a function that's calculated offline with the results sent via email. That way the lack of decent real-time response isn't such a big deal. In the mean time I have some different ideas.
I did some more thinkering on the ultimate comparison function late last night and then again this morning. For a cart of 14 items there are 16,383 possible basket combinations that could potentially go through the total order comparison process. At 12 items there are 4095 combinations. At 10 items there are 1023 different sets.
I'm contemplating continuing on the ultimate comparison path but need to know what a reasonable number of items is to compare. No more than 10 would be ideal. 12 is OK. More than 12 gets a bit heavy on the database.
How many items is a reasonable limit?
[Link]
DenverDon Benevolent
Firmly Embedded in Depravity 5914 Posts 7/02
Posted - Mar 14 2003 : 12:04PM
Drew,
10 would seem to be a reasonable number as I think that most times a person is going to pick from the lower 5 or 6 anyway.
Would the number of iterations make any difference for the connection speed relating how your connection is made, such as dialup versus cable? I would think that the iterations are server specific and probably have no relation to type of connectioin.
I think maybe I didn't explain it properly. As an example: You have 10 items in your cart and click the UC (ultimate compare) button. The server will do a total order comparison for every possible combination with the various stores in the price search that carry the items in your cart. The purpose is to figure out the absolute lowest cost to purchase these 10 items. It's not going to provide the results like the current comparison cart where it shows the results of buying these 10 items from the same store.
Example: You're looking to buy items A,B,C,D. The UC will not necessarily tell you that buying these items from Store 1 will be $X.XX and $Y.YY from Store 2. It will figure out the best combination of orders resulting in the lowest out of pocket expense for these items. It's more likely that you will find that buying A & D from Store 1 and B & C from Store 2 results in a less out of pocket expense than buying all four items in one order from either Store 1 or 2.
Also, to answer your question, the processing is all done on the server end and is not affected by the user's connection type.
10 would seem like way more than enough. I can think of a time or two when I ordered a half dozen DVDs, but I don't think I've ever ordered more than 8 at once.
[Link]
DenverDon Benevolent
Firmly Embedded in Depravity 5914 Posts 7/02
Posted - Mar 14 2003 : 1:08PM
Thanks Drew,
And I certainly did misunderstand, or misread what you said.
Anyway, I would still say that 10 would be a reasonable figure. I think it would cover the major portion of most shopping/buying excursions.
Should the user have the option to include/exclude certain stores from participating in the UC comparison?
With a store filter option if your not interested in shopping at hotfuckingxxxdvds.com they won't be included as a potential option.
[Link]
DenverDon Benevolent
Firmly Embedded in Depravity 5914 Posts 7/02
Posted - Mar 14 2003 : 1:55PM
Hmmm,
Now that would be a clever option.
If for example you clicked on the UC to do a comparison, and an intermediate window popped up where you could click on the stores that you want to use for comparison (and with an 'all' button choice too).
If a shopper, for example, only deals with 4 or 5 stores, there might be little use to include all, all the time.
Also, if he only selected 4-5 stores, it might reduce the total number of combinations that are required from the server.
I was thinking more along the lines of keeping the user's store preferences as part of his/her user profile. That way I could also filter the price search result pages to only include the stores they are interested in.
The catch here would be that if there is a less expensive option available I would include a note on the UC results or price search results page along the lines of "* This item/order available for $X.XX less using different stores. Click here to expand the search to include these stores."
I really like the idea of leaving out certain stores, even though I have yet to find an ADVDT affiliate I wouldn't use again. I use that option all the time when I check out mainstream stuff over at dvdpricesearch.
Positively, the ability to filter certain stores from the potential results set is a desirable feature.
[Link]
DenverDon Benevolent
Firmly Embedded in Depravity 5914 Posts 7/02
Posted - Mar 14 2003 : 3:17PM
Drew,
A thought just occured to me. Correct me if I'm wrong.
I think at present the price search doesn't require a person to create a profile.
If selective store/retailer information were part of the user profile, wouldn't that create a huge number of profiles that have now become 'members' but end up with 0 (zero) postings because they are not in the fourms, but only in the price search.
Forum profiles and shopper profiles are not related. i.e. You can be a member of the Kickback program without having a forum profile.
UC users will need to have a profile because the UC results may not be immediately available. I'm not sure how long it's going to take to process the UC. I'm working under the assumption that it will be at least 60 seconds. I can't have a web page processing for that long. It would kill server performance for the rest of the site. Ultimate Comparisons will be "submitted" to an external processing queue. I'll also need to limit this queue to one pending submission per user profile. When the results are available you'll get an email notification and they will be visible through a page in your personal shopper profile area.
Using an external processing step will enable me to offload the processing to a different computer if it's too much of a burden. Initial testing on a 700 MHz single processor system shows the set generation step takes 100% of the processor's cycles for 2.5 seconds on a 10 item cart. (1023 combinations) This is before any database accesses even start up.
What you are describing sounds much like what I go through with my multi-page spreadsheets when I plan an order. Although I often work with up to 40~50 titles, my eventual choices are usually narrowed down to 4-to-8 titles so the 10 items per cart per UC search seems like a perfectly acceptable limit - especially considering that we will eventually have multi-cart capabilities.
Being able to check preferred vendors is an excellent option. May I suggest that you create a list in the profile where checked is the default and we would have to uncheck those we want eliminated. That way, when new vendors are added, they are already in the mix, bringing them to our attention.
Also, if it would help, receiving the results via email seems perfectly acceptable - why bother generating a web page?
Drew, Maybe having a limit to the combination of titles and stores would be the perfect way to go.
I know for me that I usually have a large number of discs in my cart, but would only use a few stores in the "Ultimate Price Search". On the other hand, someone who typically only has a few titles may want to check all stores.
It might be a good idea to set the limit at the level of the multiplication of the titles * stores. That number could be around 60 or so. That would allow 5 titles @ 12 stores or 12 titles @ 5 stores.
While an increase/decrease in the number of stores does affect the amount of processing time, it's portion of the process is minimal when compared with the number of titles in the cart.
5 titles at 12 stores completes much faster than 12 titles at 5 stores. I haven't run that one specifically yet but I would guess than it's over 5x faster.
The current total order comparison takes into consideration a user's default shipping preference. (i.e. slow, medium - 2 day, fast - overnight) In the current system if a store doesn't offer medium or fast shipping options the comparison still includes the store in the results but uses their slow shipping rates as the default. The shipping method used in the calculation is indicated by the superscript and if the method doesn't match the preferred method there is a special character there to denote the preferred method was not available.
My question: Is this the way it should work? From my perspective I think this is a little unfair to the stores that do have the faster shipping options because their totals will be higher than the slow methods of their competitors. I'm leaning towards coding the new ultimate comparison to not include stores in the results when a shopper is looking for suppliers with 2 day or overnight delivery available. Before I finish writing it I figured I had better not make this assumption.
Don't almost all outlets offer different shipping options?
I always choose the least expensive method, regardless of its speed; however, I can understand that others are willing to pay for next day and 2nd day delivery.
If I were coding the system, I'd have it compute based on the user's default shipping method where available from a store and for the least expensive shipping method where the default method is not available from a store. Plus, when the default shipping choice is not available, I'd code the result to include "S&H?" in bold text to alert the user that the shipping choice is not his default.
You mentioned earlier that we might have the ability to restrict the search to favorite stores (a sort of checked-box subset of all the stores available on ADT). It seems to me that it would be easy for the user to check the stores that offer overnight shipping if that is one of his criteria. He could always run the search utility twice, amending his store list based on the initial search result.
To me, the whole point of this exercise is to lower the cost of my porn addiction. Restricting choices generally doesn't help lower costs.
A number of stores in the price search do not offer express shipping options. What's even worse is that some stores that only have one method (slow) also provide it for free. This biases the order totals even more in favor of the slow-only store than the store(s) providing the desired shipping methods.
I'm looking at this from the perspective of the shopper that wants the best deal for an order placed today that will ship today and arrive tomorrow. What I don't know is how many people are out there that actually shop this way.
Personally, I'm cheap enough that I always go with the slow shipping, so it wouldn't matter to me. Is it a problem to put the stores that don't offer quick shipping at the bottom of the list, similar to how the stores that don't carry some of the items in the cart would get moved down based on not having the full stock. Maybe make shipping like another item??????
How about this: figure the search based on the option for shipping chosen by the user - whatever that method may by. Show results for that method where it exists, show results for the least expensive shipping method where the preferred method does not exist. Then, for each store represented in the result, show the shipping charges applicable (if any, $0.00 if none) and the method on which they are based (preferred shipping or least expensive shipping)) as line items in the report. That way, you get the best of both worlds.
As I pointed out earlier, if you do incorporate a preferred vendor option, the user can always mark those vendors who offer overnight shipping if that is what he wants.
I'll be including the preferred vendor option too.
I think I'll write it in a way that if someone has a faster shipping type selected as the default then then it will only display results for the stores matching the option selected. But I will also include a note on the results page indicating that the results are only for the shipping method selected and cheaper order totals may be available by using a slower shipping selection. Along with the note will be a button to recalculate the results with the slowest/cheapest shipping method selected. This note/button will not be present if the preferred shipping method seelcted is the slow/cheap method.
I finally sat down and started working on this today after being away from it for 6 weeks due to various distractions. I'm 95% done now.
Now that things are much further along I've got a pretty good idea of how much drain this comparison puts on the servers. It's significant enough that for even a small cart of 5 or 6 items that I'll need to process these comparisons as offline and then fire off an email to the shopper when the results are available.
To give you an idea a 6 item cart takes about 16 seconds, an 8 item cart averages 32 seconds. A 10 item cart takes about 2 minutes 25 seconds. If I run simultaneous comparisons (simulating multiple shoppers comparing at the same time) these durations increase almost linearly as a function of the number of users. (two users=double the time and double the load).
So the new thought process is that you will "submit" your cart contents (up to 10 items) to be compared and within a few minutes you will receive an email containing a link to the results page. I can also include a link to your ultimate comparison results at the top of the view cart page.
Are you thinking of this as a new option, while keeping the existing methodology? Or are you thinking of this as a replacement to the existing price comparison process?
If possible, I think that keeping both methodologies available would be good. I know that as I add titles to my cart, I like to see who has what (without much regard to price at that point). Sometimes I will also buy something from 5-Star even if it is a little more expensive because it will get me to the total cost that I need for their 15% discount to kick in for Kickback redemption purchases (therefore, lowering my net cost) . For those instances, the existing methodology would work well for me.
Once I have accumulated titles in my cart, and am getting ready to order, the new process would be something that I would want to submit.
I'll be keeping both. Actually, I'll be tweaking the existing process to take advantage of some of the logic built into the new stuff.
New things to add to the existing comparison: preferred stores list filter, include/don't include Kickbacks in total, integration with multi-cart features, new coupon types and additional free shipping calculations.
The current comparison doesn't calculate free shipping coupons or standard free shipping offers when they are a function of the order subtotal. The new version of the existing comparison will.
