Case Study: The IT Person Quit in October. The Holiday Rush Was Weeks Away.

A multi-brand online retailer lost the only person who knew how their technology worked. No documentation, no diagrams, and a network that was down more than it was up.

Code background

The Short Version

Zero documentation

No network designs, no server designs, no vendor list. We rebuilt the picture by working backwards from the invoices.

$18K a month

Cut from communications and data center hosting bills, plus licensing nobody was using.

Password resets

That is what their tickets look like now. Standard help desk requests, not outages.

Code background

What Happened and What We Did

The client

An online retailer with multiple brands and substantial transaction volume, running distribution warehouses in two Midwest states. We do not publish client names. That is how we protect the people we work for.

The challenge

Their internal IT person quit abruptly. It was October, right before the holiday rush. The call we got was simple: we need help, we do not know what he did, and we do not know what to do.

There were no network designs and no server designs. Nobody could tell us what the company owned or who it was paying. What we found was an over-engineered, multi-state infrastructure, custom applications written by the person who had just left, and a network that was down more than it was up.

The first weeks: find out what exists

Every team went in at once.

  • Our senior solutions engineer started locating vendor information and passwords and mapping the infrastructure.
  • Our security operations team dug into the server environment and started scoping their PCI and compliance items.
  • Our cloud solutions engineers went after how they were set up and hosted with a third-party hosting provider.
  • Our service delivery coordinator engaged their vendors and got copies of invoices from the client’s accounting department. We did not know what they had, so we worked backwards: let us see who is getting paid every month.
  • Our software development team took on the custom applications, custom portals, a custom ERP and the EDI feeds from the client’s partners and suppliers. We found the source code. It was not maintained, not commented and not documented, so the team reverse engineered the applications and stabilized them.

The invoices told the story: numerous vendors, duplicated products and significant overspending.

The holiday rush

We met with the client weekly and sent status updates daily. From the start of the Thanksgiving shopping rush through New Year’s Eve, our teams worked around the clock keeping a hosting environment we did not control stable enough to carry their season.

After the holidays: the plan we had been building since day one

The goal was to get them off a patchwork of non-commercial systems and into Azure, stabilize the environment, and stop the spending nobody could explain. Products and vendors were sitting on autopay on a company card. Accounts payable did not know what they were for. Leadership did not either. They had trusted one person, and he was gone.

  • Main distribution warehouse: we gutted the network and rebuilt it. Six network switches, 17 access points, three power distribution units and four battery backup units replaced, plus new switching in a remote building that had been pieced together over time. It is now fully monitored, fully patched and fully managed.
  • Second distribution warehouse: we upgraded network and voice switching, replaced two network switches, and installed power distribution units so our team can control power to the network room remotely. We added backup 5G service.
  • Connecting it all: SD-WAN gives both facilities resilient connections to Azure and to each other, and a Zero Trust backbone provided by Ferrum covers their remote and work-from-home users.
  • Standardizing: every device moved onto our standard tools and security products.
Code background

The Outcome

  • $10K a month removed from communications and connectivity bills. Those bills now run around $4K a month, depending on toll-free usage, which moves with the season.
  • $8K a month in unnecessary data center hosting spend removed that January.
  • $800 a month off Microsoft 365, down from $2K to $1,200.
  • $3K a year in licensing removed for products they were not even using.
  • Added up, that is more than $225,000 a year they are no longer spending.
  • A network that was down more than it was up is now monitored, patched and managed, with resilient connections between both warehouses and Azure.
  • We have a complete picture of their infrastructure, and so do they.
  • Their tickets are now password resets and standard help desk requests.

Microsoft 365: pay for what you use

They were spending $2K a month on Microsoft 365, and nobody was managing the licensing. Licenses nobody was using just sat there, billed every month.

We are now their Microsoft Cloud Solution Provider and manage the licensing for them, and the bill is down to $1,200 a month. The rule is simple: only pay for what you use. Do not over-order. Do not buy things and let them sit. When someone leaves, the license comes off, so they are not paying for it.

This is just the beginning

We are now migrating their home-grown applications into cloud applications, so their leadership team and investors can analyze sales data and customer engagement.

As our dev team says: what is the challenge? Let us talk about how other clients use their data, and maybe that gets your team thinking about what you want to see. What one client does is not what the next one will do. That is why it starts with a conversation.

We have been at this since 2007. We have seen a lot that worked, a lot that failed, and a lot that went through ten versions before the client got the result they wanted. On this one we did not falter, we did not walk, and we did not get scared. We pushed through until their technology worked for them, not against them.

Code background