All across the country, there were gas stations managing inventories of gasoline using a "daily dip book" - a paper book, located at each gas station, that was updated manually after a person dipped a big dipstick into the underground tanks that fed the pumps by which commuters fill up. Ideally, yesterday's closing amount, plus deliveries, minus sales, should equal today's closing amount.
Every once in a while (more often than most companies wish to admit), there are discrepancies between what's on the books, what's in the ground, what went into a gas tank, and what was paid for.
In the old days, an investigator would have to fly to the location and audit the paper book. Understanding the cause of the discrepancy was added cost above the lost inventory.
If a gas station lost 3,000L in "a day", it showed up like that. 3,000L in a day. What we needed to understand is how this gas station lost those 3,000L. And, it's amazing just how many different ways a company can lose inventory. Each gap represented a different risk package, and had to be mitigated accordingly.
Technology, meet methodology
The company invested in and installed automated gauges into the underground tanks, and the data begin to stream in from across the country. No more need to fly investigators out to the site in question, and all the costs of airfare, hotel, car rental, etc. were eliminated.
The company also assembled a team in Toronto to formalize how we could use this data, of which I was a part.
We could pull down data for any gas station that had been retrofitted with the new electronic gauges, and had, on our desktop monitors, all the inventory and sales data we needed, in near real time. And that was, basically, what they thought the advantage would be - access to the daily dip book without the costs of travel.
While they continued to refer to the inventory as the daily dip; and even though the software that received the data stream summarized the data daily, our team's "ah-ha" moment came when we realized we could modify the parameters, turn off the daily summary, and pull down the raw data by hour...and pictures began to emerge.
There were some gas stations that showed what appeared to be the same volume discrepancy, consistently, every hour, throughout the day, regardless what else was going on at the gas station. We understood this to indicate a leak near the bottom of the tank. This type of loss gap represented environmental risk.
Some gas stations showed that same volume discrepancy, every hour, until the tank reached a certain level. This indicated a leak towards the top of the tank. After sufficient sales dropped the volume of gasoline in the tank below the location of the leak, the leaking stopped. This loss gap was an environmental risk of lesser extent than the bottom leak.
The first real profile we identified was the whole 3,000L occurring within a single hour that just happened to correspond with a delivery event. In other words, the delivery paperwork said they were dropping, say, 25,000L...but they only dropped 22,000L. This loss gap was not an environmental concern, but it was a potential fraud/theft concern.
There were also discrepancies that corresponded to sales, which is to say, the amount of loss corresponded with the volume of throughput initiated when customers squeezed the nozzle to pump fuel product into their vehicles. If the meter that captures the volume of product was improperly calibrated, it would register less fuel than had actually been pumped, and the company then was charging less price for the fill up than the value of the inventory that had been pumped. This loss gap represented lost revenue and lost profit.
Oh, wait, there's more. We also observed that one grade of gasoline might have lost inventory while another grade gained the same amount of inventory, within the same hour, during a delivery. We could thus determine that, during delivery, high octane grade gasoline was dropped into the regular grade tank, but the delivery was booked to premium. So, the premium gauge would show no change, its book would show new total volume, and thus the discrepancy. The loss risk here is that the station was selling premium product at regular price. Customers wouldn't have known, but if they found out, they might appreciate getting premium gasoline at regular price. But the company lost revenue/profit, selling a premium product at a regular price.
Now, the opposite, was a worse situation. If the delivery driver dropped regular gasoline in the premium tank, the company would be selling regular gasoline at premium prices. Sounds like a win, right? Yeah, no, way wrong. Because if customers found out they'd paid a premium price for a regular product, the company would be exposed to lawsuits. That PR risk was not at all worth beating customers for a few more cents per litre.*
And thus, we developed the company's first hourly dip book for remote, wetstock inventory discrepancy diagnosis, providing 24x times the granularity that empowered the company to identify the discrepancy profile and apply the appropriate mitigation.
While the initial objective of the project - eliminating the travel costs associated with discrepancy diagnosis - our approach raised the value of the data to provide more insight than they'd previously sought, insight by which they developed a risk mitigation platform that empowered their becoming exponentially less exposed to losses (PR, environmental, legal, financial) and exponentially more responsive and proactive in responding, in a timely and appropriate manner.
The value? Lots of millions of dollars.
Technology alone will not maximize a company's potential. Technology is a tool, which is only as good as the skill of the person wielding it. We took some good tools, developed a methodology to put them to better use, and produced sharper output than the company originally thought possible.
Just another example of what we say, and do, here at Spherical: your data...more profit.
* I hasten to add, I was never aware of any gas station that exhibited this particular loss gap. We formalized a set of theoretically possible loss gap profiles that extended from the data, whether or not we ever observed any gas station that actually exhibited any of the possible loss gap profiles. As is the case with most any data landscape, there was a bell curve demonstrating that a few profiles covered the majority of losses, while some profiles were never observed. The company was able to proactively understand the extent of loss risks; develop appropriate mitigation for each, and quickly apply the appropriate mitigation should the respective loss profile be identified at any gas station.