Monday, January 4, 2016

Building For Tomorrow: Software Infrastructure

Happy New Year everybody!

The year 2015 was Desert Sky Games and Comics's best year ever.  We aren't quite in the "two-comma club" yet, which happens when you gross at least a million dollars in sales.  ($1,000,000)  But we posted some strong scoreboard and it is deeply fulfilling to be able to run healthy when our local market is supersaturated, mostly with Magic: the Gathering-heavy stores.  It is even more deeply gratifying to have a Magic player community as strong as ours is, mainly achieved by emphasizing fun and casual play, when players have so many options.  Even competitive players do play at DSG, so they must see value in the organized play structure we are offering.  Meanwhile, non-Magic product categories are growing apace.

For 2016, then, our priority is to get on track for future expansion of our market presence.  By late spring we will need to make a judgment call on building a new hub location.  Depending on where available property ends up being, it might replace the existing store if it's close enough, or it might become the new main store and allow us to keep the existing location open as a satellite location.  We also remain interested in smaller satellite locations in unserved and underserved areas of our state.  Down the road, in neighboring states.  Until last week, we didn't think Tucson was in play.  Two stores in Tucson just closed, and both were good stores in their own right.  We think with the right framework we can succeed in that market.

Regardless of where we go, or even whether we expand at all, one thing we need now is a point-of-sale infrastructure that will scale better than Light Speed Retail Pro/Onsite/Client.  We opened with Light Speed and it served us well for years.  Unfortuantely, updates since late 2014 have been mostly of poor quality, shoehorning in new features without resolving existing problems.  They were never able to solve a problem with Web Store v3.2 that made it impossible for about half our customers to complete a transaction due to Error 500 that they couldn't replicate on their end.  That's the ultimate dealbreaker right there.  People can't buy a thing.  When we're in the business of people buying things from us, that's a problem.  Light Speed's Amazon integration was in perpetual beta status and didn't even hit all categories, such as comics.  We've moved away from Amazon as a selling channel, so it's a moot point, but it speaks to Light Speed's dedication to the product.  Bottom line, Light Speed is devoting their resources to their subscription cloud software, which is grotesquely expensive at scales larger than the boutique level.  And with the client/Pro/Onsite version crashing on a near-daily basis, it's no longer reliable enough for us to keep and grow alongside.

We ran Crystal Commerce for our web back-end for two years, from 2012 to 2014.  There is a lot to recommend that software for a store where the beginning and end of their online business is selling TCG singles.  Crystal Commerce is meticulously tailored to exactly that functionality and only that functionality.  As DSG has matured, online singles resale has become a less and less necessary sales vector for us.  It's a race to the bottom for pennies on the dollar and depending on the sales channel you lose anywhere from 2% to 19% of the gross sale price, before shipping, to fees.  Most stores don't even realize they are losing that much because some of the fees are visible and some are not: if you don't log in to your TCGPlayer web dashboard specifically and review your monthly statements, you won't realize how much they're deducting off the top.  You only see the net.  I know many stores see these fees as acceptable because they buy collections of Magic cards so cheaply and it's almost all profit anyway, and this allows that store to avoid spending any money on real technical infrastructure.  You can just use the cheapest PC you happen to have.  But ultimately in that deployment you're depending on there being a never-ending string of people walking in the door to sell you collections at distress rates.  When your purchasing of used merchandise goes from a want to a need, it changes the situation and erodes leverage.  And every new store in your area takes a bite out of that selling traffic.  You end up spending money into Facebook and spending time into Craigslist relentlessly promoting your store's buys.  And God help you if your internet goes out or Crystal Commerce's server is slow or down on a Friday or Saturday, which does happen.  On balance, once a store decides it's going to be more than a corner Magic clubhouse, Crystal Commerce becomes untenable.  This is based on market factors alone, and not even reaching the issues where Crystal Commerce is lacking as a point-of-sale system on functionality.

We are migrating within the next few weeks (I hope, pending various deliverables) to the Microsoft Dynamics Retail Management System (RMS) point-of-sale software, using Diamond's ComicSuite deployment and plug-in.   This is a large commitment, in both money and resources, to a scalable infrastructure that should serve us for years to come and multiple locations.

I am carbon-copying the RMS deployment that a friend of mine did for his two sporting goods stores, where they gross eight digits left of the decimal annually.  So clearly the software is up to the task.  It's not perfect, no point-of-sale software is.  And it's actually End-of-Lifed by Microsoft as they transition to a new build that Diamond figures to have ready for us sometime around 2021.  But there's deep and extensive industry support for it, and ComicSuite as a tool for comic inventory management is best-in-class.

So, critical to me is being able to keep uptime and resilient business continuity across multiple locations whether I am anywhere nearby or not, and being able to administer it all from a central terminal, mine.  This means that each store requires a rack-mount database server.  I bought a Dell PowerEdge for the current DSG, an investment of about $4k after configuration and a RAID-0 of six terabytes (leaving three as the final capacity), hot-swappable and extensible.  Each location will require one of these.  Adding in the rack-mount keyboard and admin monitor, case, power conditioning, and network hardware, we're approaching $5k before even having a single terminal for sales.  That seems like a lot, but if Light Speed crashes on a busy Saturday, I can lose almost that much in sales that day alone.  Crashes are, from my friend's experience, exceedingly rare in a robust deployment such as this.

Then I need three lanes: Two front-of-house and my primary admin lane, which is also the remote lane for conventions.  RMS, unlike Light Speed, offers unlimited non-lane logins.  That is, I can have staff at non-lane workstations doing things like adding inventory or managing comic subscriptions, using the Store Admin app.  Thanks to the cloud Store Operations app, I can access this same functionality across multiple stores.  The Diamond deployment has some issues with multi-store operations at this point, and I don't have to solve those problems yet, but I am confident that with some combination of ComicSuite and Store Operations, I can execute any needed command from somewhere.  So, since Light Speed required Macs and RMS runs on Windows 7, I will be adding three business-configured desktop PCs that each run about half a grand for the base box, and another $80 each for a basic HDMI monitor from Costco.  I'm going to limit the lane PCs to exclusive use for that, and no general computing, to extend their useful life.  Accordingly, we'll keep one of the iMacs up front for lookups and the like.  I already have cash drawers, scanners, and so forth, and my EMV credit card processing is stand-alone.

The remaining expense is for RMS itself.  Diamond has full licensing for the first lane for a combined two grand up front and a few hundred a year for the maintenance subscription.  Since we're doing a three-node deployment and will ultimately be moving to multi-store, we will end up with a bundle price tailored to our exact setup that is less than three times that figure.  (Even if Diamond weren't willing to offer it as such, we could just buy RMS from another authorized reseller on such terms and buy only the ComicSuite module from Diamond.)

Some other equipment is already here and some of what we'll have to buy is incidental to the new deployment -- such as Office365 to replace the Office 2011 Mac that we have now -- but all told this is roughly a $10k technical upgrade.  And it's only a point-of-sale upgrade.  We're standing pat with our security system and our audio/video system, both of which could really use some love, but are much more location-dependent so we'll hold off on those until we know the lease situation moving forward.

I'm hoping to revisit the Building For Tomorrow topic in future articles, where I can talk about security and A/V, but also rack and fixture, other physical plant necessities, and hopefully even topics like a store's human resources framework and convention portable kit.  If you're interested in anything to do with those, or to hear more about our RMS migration adventure as I proceed with it, please feel free to leave a comment or contact.

Have a great week!

No comments:

Post a Comment