E-commerce: Aimondo’s Artificial Intelligence against 88% Excel calculation errors
65% of all retailers use Excel for their planning. Artificial Intelligence (AI) from Germany-based Aimondo is used to monitor online competitor prices instead.
65% of all retailers use Excel for their planning, according to studies. This means that in the course of digitalisation, they have only a limited ability to manage their business with real-time information in an agile and forward-looking manner. However, agile management is a decisive competitive factor for both strategic and operational decision-making, especially in online retail. Artificial Intelligence (AI) from Germany-based Aimondo is used to monitor online competitor prices. As a side effect, this results in a constant logic control of one’s own offers. And thus target achievement through error avoidance combined with price optimisation.
Almost all somewhat complex Excel tables contain errors. A single typing error, a wrong formula or a transposed number can have dire consequences. The result: 88% of all spreadsheets with more than 150 lines are proven to contain massive errors. Errors in the price calculation have a direct effect on the profitability analysis. Now you could say: the errors balance each other out – one error increases the price and another reduces it. With opposite effects on profit and loss. However, this is a tempting but misleading assumption. For buyers are the incorruptible corrective for wrong prices.
If the price is too high, the seller in the large assortment hardly notices this and at best draws the conclusion that the article is difficult to sell, lies motionless in the warehouse and he may even take it out of the assortment. The customers buy elsewhere, because they are sensitive to each individual product and know their way around – through their own initiative or with the help of price comparison portals. The effect: lost sales, lost profits and valuable customers who become loyal to competitors.
If the price is too low, the so-called stock rotation rate may be right – but the profit leaves much to be desired and with each turnover of the item affected by a mistake, an unnoticed coverage gap is created. The sum of such unachieved individual profits endangers the strength of the company, it even develops into weakness.
Errors cannot be avoided – but they can be reduced. Let’s go back to the beginning: 65% of all traders calculate with Excel. Let’s take as a vivid example a function of the worldwide spreadsheet programme for which not even a formula or input error by the user can be held responsible. The auto-correction. In the English version, March1 becomes 1-Mar without the user noticing. For the merchant, this error may be inconsequential. The gene researchers, on the other hand, knew the gene March1 in the past – lost it with Excel and now had the effect that up to 30% of all studies are wrong. To determine this, 11,000 scientific studies with Excel attachments had to be examined and analysed. A fatal naming error with major consequences – in addition to the controls and corrections, 27 different genes had to be renamed. What does such a mass error result mean economically? How many products have description details that Excel renames without comment with its programmed knowledge? And how many errors beyond such system problems do humans make in the depth of a file with several references to other files or spreadsheets? Who is going to check all the data points in a grown file with several references and perhaps even different authors?
Research by Ray Panko, professor of information technology management at the University of Hawai’i, has long shown that spreadsheet developers achieve 96% to 99% accuracy when entering information. Unfortunately, for large spreadsheets, a cell error rate of 1% to 6% is certain to produce incorrect results. Can cell error rates (CERs) really be that high? General human error research has shown that humans inevitably have comparable error rates in simple but non-trivial cognitive tasks. The problem is that cognitive mechanisms get it right in almost all cases, but fail a few percent of the time.
For years, systemic or human errors affect day-to-day sales and distribution – influencing everything from purchasing to the balance sheet. And this is the case for 65% of all traders. In a market environment of countless individual products for which around 85% of all buyers cite price as the most important purchase decision criterion.