Tales from the Trenches: IT Reporting During Peak Retail Periods
This post was written by Numerify’s Director of Product Marketing to offer firsthand insight into the challenges many retailers face when relying on IT reporting during peak retail months.
I may work in marketing now, but in the past I was fortunate to spend some time in retail on the technology side. I worked for a major retailer on their multi-branded web presence, managing performance engineering and capacity planning teams. My role was to ensure the systems could handle peak loads during the holidays. Since I had performance data around most of the systems, I was also tasked with building daily dashboards and presenting those dashboards to senior leadership during peak holiday season.
The giant spreadsheet I produced consisted of aggregated front-end web traffic, web sales numbers, system data (server, storage, networks, caching, CDN, application, middleware), file transfers to the warehouse system, the warehouse management application, order flows through the systems and orders shipped, along with a handful of additional metrics. I placed all of this data into a spreadsheet that was printed on a single 11 x 17 paper, with each metric represented in a row/column and color-coded as red, green, or yellow. I had no time to dig into what the numbers meant, because the spreadsheet took so long to build. Each morning I reviewed this spreadsheet with the CIO, CTO, and the head of development, infrastructure, and warehouse management.
What Is Peak Season, and What Does It Mean for Retail?
I’m stating the obvious here, but the holiday period is everything to retail! Almost 20% of the U.S. retail industry’s annual sales come from the Christmas holiday shopping season. At the retailer I worked for, that number was even higher.
All strategic decisions, marketing tactics, and technology investments are set up to support key shopping days like Black Friday, Cyber Monday, and Super Saturday. These dates can make or break a retailer’s revenue results for an entire calendar year. For my previous company, the holidays were by far the highest periods of steady traffic to our sites — and more traffic means more $$$!
All tech projects were scoped during late winter and had to be delivered by the end of October. If you were late on your project, you would work overtime until it was finished, pushed to a post-holiday time, or scrapped altogether. Only emergency changes could be made to systems between November and January. If you needed a change, it had to be extremely important to risk system downtime – so you really needed to get your act together!
The Limitations of the Spreadsheet
- If a file was missing or in the wrong format, the dashboard build would fail and I had to spend time troubleshooting the issue
- I had no time to do any type of analysis, because I was wasting time compiling data
- I had no idea who owned the data and had to hunt down owners if files did not appear on time
- There were no explanations for missing data
- I could not answer questions around the data such as “How did high network utilization impact sales?” and “If this has happened before, what did we do to fix it?”
- All data was normalized so I could not see any outliers, and trending was inconsistent and error-prone
- The process was extremely stressful!
My Key Takeaways
Peak planning and holiday uptime is critical. The retail technology stack is complex, especially when you factor in infrastructure, APIs, middleware, pick-pack-ship processes, POS systems, etc. That complexity continues to rise amid shifts in customer attitudes. Customers have many retailer choices, and a poor experience means lower sales and damage to the brand.
Having an IT analytics solution would have been a huge help for me. But more importantly, having access to key data in a single data warehouse is invaluable. When you automate data collection and transformation, you have time to use that data to make decisions. I survived peak season and sales continued to rise, but I always wondered if there was more we could have done. Could we have prevented outages or business disruptions? Could we have sold more items and delighted more customers? Now I know there are applications out there to help answer and address exactly these types of issues — and even eliminate the process of building those pesky spreadsheets.