An e-commerce website is trivial. That’s why Home Depot, Target and all the other big Amazon competitors have it nailed.
Seriously, I don’t understand how so many major players are so bad at it. Search that doesn’t work (heck, Home Depot can’t consistently get their search bar to display if an ad blocker is running) or can’t find the item you’re searching for even if it’s there. Item displays that are absolutely awful (the item I’m currently looking at and probably want to buy is far below the fold). Delivery that is “FREE” but mysteriously is not free in the cart. The list goes on. Shipping address verification that wants to “correct” your address to something wrong in an infinite loop.
Sometimes I feel like Amazon’s competitive advantage is that its website is so-so in a field where everyone else’s is truly awful.
Beyond a certain scale (when you're becoming more of a logistics company and less of a store front) there are a whole heap of rather tricky problems sitting there waiting to bite you in the arse. Preventing exponential cache growth is my favourite, especially for b2b. Search is also highly non-trivial, again especially b2b.
For a company like Home Depot or Target, there is an actual finite inventory of sellable objects in the physical possession of the retailer, and this inventory is small by computer standards. Any remotely reasonable full-text-searchable database should be able to search it accurately and efficiently.
That leaves items in the supplier’s inventory (or made to order) and third-party listings, I suppose. To the extent that items in the supplier’s inventory are curated, they should still be finite and in a database. Made to order is more complex, since the actual number of distinct orderable items may well be effectively infinite, but the number of item listings is still finite.
Third-party is a mess, and this is where I think many large retailers mess up. A large part of the value proposition of a store like Home Depot or Target or the in-store part of Walmart, etc is curation and first-party responsibility: if I buy electronic stuff at Home Depot, I expect it to have appropriate safety listings and generally not be utter crap. If I buy housewares at Target, I expect them not to be made out lead. By allowing third-party crap into their sites, these companies dilute the very thing that makes them better than Amazon. (And they apparently mess up their websites, too.)
At the very least, search should be entirely reliable for first-party listings.
you can think of b2b as "every customer has access to different products at a different price" and optionally "they want to integrate their ERP with the ecommerce vendor's ERP". Also that the concept of customer is not what you might think it is.
I’ll ignore the ethics and wisdom of a major retailer showing different customers different pricing, especially a retailer that operates physical stores with price tags on the shelves. Instead I’ll point out that, if you run a successful business (physical retail) and you want to add a secondary business (online sales of the same goods plus some other stuff), then getting the simplest model working well should be a much higher priority than adding fancy complications (bespoke pricing, for example).
yeah b2b is pretty much at the wholesale end, where everyone has different accounts with their own pricing negotiated as part of the contract and so on. If you go into an electrical wholesaler for example, you can buy at consumer prices, but the tradies will definitely be getting some kind of negotiated price depending on the scale they're operating at.
From my experience in this space b2c is kind of a subset of b2b where you can treat the c as a single customer (see - users and customers are related but not the same) in the rest of the b2b mess. But totally for the simple case - a retailer who sells a particular kind of thing - the basic solution is pretty trivial. Search is tricky once you start moving from single product category into general catalogue. Then you're going to want things like rules to feed back search results from customer behaviour (one example of a few curly ones). Algolia seems to have solved that kind of problem adequately for now given the assumption of a certain level of quality of data in the licence holder's back end systems and aggregation code. Then there's things like variants (e.g. same product different colour / size) where it's tempting to make a naive implementation that won't scale and will be a maintenance nightmare, or where your data modeling fits the arbitrary case instead ...
another one to consider is that product classification should be thought of as arbitrary efficient representations of a hierarchy subject to structural change at any time. And that different customers can see different hierarchies. They're all simple problems once you start thinking about them int he right way, but getting them into a coherent architecture with minimimum spaghetfication is challenging. I'm definitely interested to hear about good agencies working in this space (not because I'm hiring).
Is there anything to search not working because the platform has allowed itself to be gamed with key word stuff to the point of simple things like a title being useless in and of itself?
Places like Target and Home Depot do not have an unlimited number of suppliers for the same object. So each object has one page and not a varying number of pages for the same product for different vendors.
And yet Home Depot is regularly unable to find items using its internal search. So I fall back to using a web search engine to find things on Home Depot’s website.
Seriously, I don’t understand how so many major players are so bad at it. Search that doesn’t work (heck, Home Depot can’t consistently get their search bar to display if an ad blocker is running) or can’t find the item you’re searching for even if it’s there. Item displays that are absolutely awful (the item I’m currently looking at and probably want to buy is far below the fold). Delivery that is “FREE” but mysteriously is not free in the cart. The list goes on. Shipping address verification that wants to “correct” your address to something wrong in an infinite loop.
Sometimes I feel like Amazon’s competitive advantage is that its website is so-so in a field where everyone else’s is truly awful.