It’s been a while since I have blogged so I figured that it was time for me to give an update on what we’ve been working on around here lately. I don’t know how impressed you will be with what I (and Carter) have done, but we put quite a bit of work in this summer. So for your quick reading, here is my selection of the top 5 technologies that we have been working on.
Tags! Tags Everywhere!
Once upon a time our product entry system was simple; you would add a product to the site, add images and a description, and assign categories. This system seemed to work fine for many years however after a while we began to notice that not everything had all of the data that it needed to have on it. Some products would have wonderfully written descriptions (and some not so wonderfully), but they didn’t even answer the simple questions like “what color is it?” or “what is it made out of?”. To fix this issue we had to reevaluate EVERYTHING about how we were putting data in the system. The biggest issue with how we were doing it was that were all caught up in writing descriptions and setting good images, and forgot that computers (like the ones that actually run the stores) aren’t very good at reading or seeing.
So after much system tinkering we created the “Product Tagging System” that we use today. The idea behind the tag system is simple (albeit somewhat stolen from WordPress), there exists certain things (frequently in adjectives) that describe products. There is a “Red” tag, a “Cotton” tag, and a “Shirt” tag. All of these tags help us define what a product is in ways that the system can understand. Although out computer systems will never need wear shirts or which ones are fashionable, it can not easily separate the items in our store that are shirts from the ones that are not. Building upon this tag system we have setup schemes of tag applicability and requirement (such that shirts require sleeve length, but sleeve length is not applicable to books), and a system of tag inheritance, such that if a shirt is red, all its children products (the specific sizes) are also red.
And of course we pass all information back to the users, and it is shown in the nice little specifications box at the bottom of the product pages Click here to see the box on the Black Onesie
Filters and Search
Thanks to Google, when people come onto our site their eyes are automatically drawn to that little text box that says search. They begin to expect it, but more importantly they expect it to get them to the products they are looking for without a need to understand anything about the way our system is set up. In the past, if you did a search for “shirts” or “shrt” you would get nothing because the word that was included in the name of the products was “shirt”. If your spelling wasn’t perfect, or you pluralized when you shouldn’t, or we happened to call them “body coverings” (ok, maybe that’s a bad example) then you would get no results. And zero results meant unhappy customers and zero sales
To fix this we changed the way our system handled search requests, instead of simply just looking for text matches in the name of a product, we wanted the system to somewhat understand what the user was searching for. If someone did a search for “camo shrts for girls” we needed a system that would break this down and try to understand it. After some thinking we realized the only thing that the system really understood was tags (discussed above), so we started there. The system as it exists today is smart enough to break that search down into a search for products that are “Camouflage” (another word for camo), “Shirts” (the proper spelling), and “Female” (our system’s classification of gender). All this put together lets us parse the query and result proper results. It’s no Google, but it’s better than what we had
We have been trying to sell bundles to our customers for years, but until now it has been a serious pain for us. Due to constraints in our system we had to setup each bundle we created in multiple systems and they had to be static kits (containing specific parts). And the inventory management on them was a nightmare. It seemed like as soon as we made a kit, one of the parts would go out of stock and we would have to replace it, or put the kit out of stock.
Skip forward to today. We are now in the process of rolling out a new bundle technology onto all our sites (starting with Baby-n-Toddler) that allows us to create bundles like this Argyle Bundle. We are excited about this technology because it lets us offer bundled discounts, allow users to pick what parts they want, and adds the parts to the cart instead of the bundle itself (which simplifies inventory like crazy). Even if you aren’t excited about the technology itself, you will be excited by what it allows us to sell like this Personalizable Select-Your-Color Onesie or this Born to hunt vinyl iron on that you can choose what onesie you bundle it with.
More Reliable Inventory
In a previous post, I mentioned that inventory integrity is really important to us here at RNK, and I am happy to say that inventory has never been managed better. Due to the history of the company, we use Stone Edge Order Manager (a crappy little piece of Microsoft Access code) to manage our inventory in our warehouse, and a centralized database on our web server (referred to around here as just “Central”) to manage the inventory number displayed on our websites (I think we have 7 at the moment). The synchronization between these two used to be a flat file spreadsheet that would be sent over FTP on a periodic basis, and there was serious issues with latency of the process (it could take up to a day for numbers to make the full path from SEOM to our Miva stores). Those days are now behind us though because our server now has inventory submission servlets that allow us to send inventory directly from SEOM to Central with no need for file saving and transfer. This may not sound like exciting news to the outside world, but we are certainly happy.
Feeds and Google Ads
This summer, sales have not been as great as we wanted them to be so we decided to play around with some advertising. For several weeks we worked to get our systems populated with the necessary data and mapped out into a feed that Google wanted so that we could successfully send out a Product Feed for Google Shopping. You have probably seen the little ad box that shows up on Google when you search for a product, and we are quite happy that our products now show up in there.
Now that we have all this data and technology built, we hope to get our products visible in even more places online so that we can get products to the people who want them.