The mechanics of an accounting information system can be simplified into a simple equation. Input goes into a process, at which point an output is produced. Input => process => output. An input is some type of raw material. The raw material doesn’t have a specific utility on its own – it has indirect utility. For example, in a system where vegetable soup cans are made, the vegetables are not good in raw form. They have no specific utility on their own – they have indirect utility. In accounting information systems, the raw material is the data. Data is comprised of individual data points.
Individual data point example: at Macy’s in herald square, a person buys a white sweater. The data point will show the sku number, the quantity bought, the color of the sweater, the person working the register, the time of the transaction, etc. One data point won’t have direct utility on its own. There is only one form of utility for an individual data point: it can be used as an audit trail. For example, if a transaction needs to be verified or if additional information is necessary down the road related to the transaction, the data point can be analyzed. This can occur in our example if an unathorized employee discount is given on the transaction.
The process is the component in the accounting information system that transforms the input into the output. A process is made up of a group of subprocesses. The output can be a finished or semi-finished good. In accounting information systems, this output is the information that is used by the end user. An accounts payable system with a three-way-match is a great example of a series of subprocesses. For example, vendors send their invoices to a property management company. These invoices are then matched to open purchase orders (to verify that these goods/services were ordered in the first place). This is the first subprocess and the first match. Then, a project manager or property manager marks these goods as received in the system, adding human verification that not only were the goods ordered and the invoice was correct, but that the goods/services were actually provided by the vendor. This three-way match between Purchase Order, Invoice, and Receipt Report is common in accounting and a great example of subprocesses. The output, or product of these subprocesses, is the Accounts Payable summary of the company. Monitoring accounts payable is an important part of meeting budgetary guidelines, managing operating cash, and operating a business effectively.
Other examples of Accounting Information System (AIS) output is month-end trial balance report, P&L/Income Statement, Balance Sheet, etc. A series of items feed into the process to generate the final output. This output may be semi-finished – for example, the Balance Sheet may require adjustments for depreciation, mark-to-market items, etc.
The process is complete and a final product is available as the output. However, if this process is a repetitive process that must be completed on a monthly basis by the company, there needs to be a way to incorporate feedback back into the process. This is the purpose of a feedback loop. The components of a feedback loop are input => process => output => sensor => comparator => activator => input. You will notice that the final step brings us back to input, which is the purpose of the feedback loop. The diagram below is a visual representation of a feedback loop.
The role of the sensor is to take measurements of the output. The comparator then compares the measurements to an idealized model. The activator triggers a response. An example can be found in Computer Assisted Manufacturing (CAM). American Can was at one point the largest aluminum manufacturer in the United States. At the time, random cans were measured as equipment was not available to measure every can for consistency. However, with American Can at the time only a random sample was taken to test accuracy. American can also they had a steel can division which used the same process. Tens of thousands of cans were produced in an hour. With this volume of work, any lack of alignment in the machinery can cause thousands of defects if not caught quickly. The main goal of quality control with this sort of process is to catch any possible defects early in the process. A system of random testing was therefore developed by American Can. A robotic arm takes a random can. This is the first step in the feedback loop process. The sensor then measures each can. This measurement is then sent to the comparator. The second step is the comparator, which compares the can with the ideal soda can measurement.
There are several possible outcomes that can be produced by the activator in this feedback loop example. One possible outcome is that the measurement taken by the sensor matches the required measurement and is a perfect match. In this case, the activator logs the scenario and moves on. Another outcome is that the measurement is not a perfect match, but is within a previously specified limit known as the fault tolerant limit. This event is logged by the activator, but the system remains in operation as the measurement is acceptable, despite not being 100% perfect. Another option is that the activator logs deviations from the measurement that appear to be well within the range of the fault tolerant limit but are showing a trend toward error. IN this case, the activator generates an email or other notification going to the foreman’s screen. This is not yet an emergency event, but the foreman then knows to adjust the settings of the machine to correct this trend toward error. The final scenario that can be encountered by the activator is that something is completely wrong with the system and the measurement is not compliant with standard. In this scenario, the activator shuts down the system and sends a notification to the foreman. This trips the production process and allows a correction to be made before additional inputs are fed into the process and wasted.
This example was applicable with American Can decades ago. Now, technology has made it possible for 100% of cans in aluminum can production to be tested. The largest producer of cans in the United States currently is Ball Industries (if you look at a soda can, chances are, you will see the Ball logo in cursive somewhere on the can). Ball Industries has 100% basis measured for accuracy. In other words, they test every can.
Quality assurance processes in the form of a feedback loop can be found in numerous examples across various industries. For example, about a two decades ago, it was still possible to find the occasional brown potato chip in Lay’s potato chips. However, through the use of feedback loops, these brown chips were eliminated.
Gathering information from a feedback loop leads to a large amount of data. This data is summarized in a product utilization summary. An example of a product utilization summary from Macy’s, for example, will include how many white sweaters were sold, what color is the one that sold the most, which days/hours did these products sell on (midday, evening, weekend, etc.). This product utilization summary is then compared to the expectation. This is called a budget/variance analysis. This expectation, and the deviations from it, are used to tweak the company’s strategy in the near term (how to dispose of excess inventory) and in the longer term (what quantities of white sweaters should be ordered). Shorter-term solutions related to our Macy’s example include selling excess inventory to an outlet or dropping the price. Alternatively, the manufacturer may ask these products to be destroyed. In Lowe’s Home Improvement, for example, certain appliances returned after a certain season were thrown in the compactor rather than repackaged for resale. Agreements exist between Lowe’s and certain suppliers where certain products are destroyed rather than resold.
These product utilization summaries may be generated hourly for faster-paced businesses in e-commerce. They do not necessarily have to represent a week or a seasonal summary, as our Macy’s example explained.