Union Bagel POS Analytics

Point of sale (POS) systems capture customer transactions during operating hours. Each system is placed at the front of every restaurant, retailer or business. Even mobile POSs can be found in taxi and delivery services. The industry has slowly been making improvements in its technology features and lowering the service price point. Since maintaining a server in house is technically challenging the majority of businesses use the SaaS model of POS. The majority of entrepreneurs do not spend their days fantasizing about logging customer transactions so they arrive on a decision unprepared and often exploited.

Hastily setting up a POS is a mistake because it is not convenient to change services. There are short term hardware leases built into the contract so timing the change is critical to avoid paying early termination fees. Employees will also need to be trained on a new UI. Each POS service markets itself as easy and all powerful but the reality is all the extra services have add on fees which aren’t included in the initial basic subscription.

Dashboards are provided to the business as part of the service. A store of 4 employees will get the same basic dashboard as a chain of stores with 500 employees. The one size fits all approach makes answering fundamental business questions impossible. At first glance all the charts and tables of information appears to be good. The dashboard is often considered better than nothing but it is actually worse. They are all designed to up-sell services. Compare the price and performance of each merchant POS and it becomes clear they are converging towards a common script.

I put together a few charts for a friend of mine who operates a artisan bagel store with less than 12 employees. Every day he sold from 1,000 to 2,500 transactions. Drilling into the data we found some interesting patterns which makes the business decisions much more intuitive. A few high level examples:

Monthly Sales Revenue

Season plots for monthly sales revenue show year over year change. Monthly gross sales are indicated in the y-axis, with all values removed to protect the owner. The y/y change calculation would be trivial to make: it compares each total sales for the month from the previous year. To keep up with increasing rent, cost of goods and labor a 5% average growth is desirable.

Season plots are useful for detecting seasonal trends and unusual patterns. The increase in sales revenue after JAN 2016 was aggressive. Also there are some annual cycles that show up in comparing 2016 and 2017 (not shown). The winter low season (Nov-Mar) and the summer (Apr-Sep) peak season.

Monthly Sales Items

A monthly sales by item table showed buying patterns to help with inventory. A table is made for total sales and transaction counts for each menu item on a monthly basis. See  column ‘totalGrossM’ to see the totals sales for that item. The blue bar size corresponds to the dollar amount, so bigger is better.

This list is over 1400 lines long so the results are less than intuitive. To overcome this the table has several features. Major trends can be filtered for the owner (not available here) using:

  • Counts (N) are colored yellow when counts are 100 or greater, gray otherwise.
  • Page 1-144 can be manually clicked through. The bottom right on table indicates this feature.
  • A search bar is available for filtering by column, e.g. search using “Bgl W/CC”.
  • Each column has an arrow available to interactively sort the order.

The amount of entries can be adjusted using the drop down on top left of table. Click on it to make an adjustment on the number shown.

Customer Order Times

This plot aggregates customer order times to help with organizing employee shifts. Each weekday density plot is approximately equivalent with respect to shape; i.e. there is a peak at The results are not very informative but that may be because they are aggregated together for all transactions from 2015 to present. Drilling down, e.g. monthly or seasonally, would be informative.

 

Discussion

A lot more can be done with this information. Drilling down further into the details is often necessary after making a first pass at the high level questions. Setting up marketing campaigns is simple to design. The ROI on spray and pray promotions is difficult to quantify using the basic POS service. A data driven approach will dovetail conveniently with business interests.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s