It is that time again, where we as a profession have elevated enthusiasm for automating redundant tasks and seeking the efficiency promises of artificial intelligence and blockchain.
It is also that time again that I like to ask: “What can go wrong?” For one, fraudsters love automation and seek to realize efficiencies of their schemes too. If they detect a flaw, they will exploit it.
It is disturbing to see those racing to adopt, promote and encourage automation seemingly ignore basic fundamental principles of accounting and fraud. I repeatedly see articles written by accounting or IT professionals that subscribe to the notion that smart contracts, blockchain, and AI will eliminate fraud. One enthusiast that still stands out in my mind is an author that had the debits and credits of cash reversed in an example, showing a cash increase with a credit. I certainly hope that this author is programming the bookkeeping bots.
It is human nature when exploring a new technology to be enamored of the benefits and become vulnerable to the seemingly complicated technology. At these moments, we are most likely to fall prey to fraudsters or those so passionate about their product that they fail to consider all outcomes.
I may seem “algorithm averse” but on the contrary I firmly believe that algorithms are very effective, less error-prone and extremely efficient. However, if programmed incorrectly, or implemented without considering unintended consequences of this power, when things go wrong, they do so very quickly.
Leaving the complex, ‘techy stuff’ to the developers and programmers advice creates a vacuum for illicit actors to fill. I do not propose learning the technology in great detail, however, rarely do I see commentary suggesting practical approaches as to what or how to learn.
Influencers and experts encourage having the conversation with your clients and associates, but what should you say? I believe knowledge is power and you should know how to spot a bot, where to look to detect potential anomalies and what can be done to mitigate risks of negative outcomes. As such, I’m offering these three key tips:
In social media, it is not all about the quantity of followers. In fact, bots often have a limited message and minimal engagement. I understand many succeed in gaining an artificial following, and I suspect that given the continued use it translates to business.
Regardless, it is important to have an actual human monitoring your social media channels for things beyond quantity of followers. I see many accountants with automated accounts. Some of the challenges that can occur with this approach include limited engagement, the potential for a competitor to recognize an unattended account and capitalize on the automation; picking up on key words, publications, and RT patterns.
Chat bots are an efficient method of answering sales questions and providing support. Be sure to have a method of assessing the quantity and quality of information you are giving away. If additional traffic is not translating to conversion and retention, you may want to reassess.
2. Look for Anomalies and Outliers
We tend to look for outliers within the current set of data. Go beyond and look at legacy data and patterns. If you received a record-breaking number of new customers after simply digitizing a promotion, how real are these and what is their longevity?
Consider external and environmental factors when evaluating the appropriateness and outcomes of automation. If vending machine sales are up, what is happening in the cafeteria? If AR collections are up, do not forget to consider if there were human changes in this trend.
I use the same methods and have the same conversations I had with business owners racing to offer Groupon promotions in 2010. One needs to consider the staying power effects to avoid careless decisions in hiring, inventory and customer service. Many small businesses "Groupon’ed" themselves out of business, losing their loyal customers to these promotions.
3. Keep Pace With Trends
Fraudsters are savvy, they work to mimic normal behavior and generally have accomplished this before mainstream adoption. Most successful and large-scale bots know how to beat tools like Captcha. Look to some of the larger firms to see what current methods they are using and their successes and challenges.
Remain conscious of fraudster goals and how they might best accomplish it in your automated world. In terms of occupational fraud and abuse, know that your payroll, AP, AR, expense reimbursement, and asset misappropriation schemes haven’t changed. Know where to find the QuickBooks equivalent of “Print on Check As” type vulnerabilities.
Who is gravitating to the latest trend and why? Over half of all US equity trading is done through quants and high-frequency traders. We are just now starting to see some strategies and efforts to recognize and combat potential abuses. As we do this, are the original players scurrying to another market where they can deploy previously successful tactics?
In short, bots are great tools, but be conscious when implementing to mitigate costly and accelerated missteps.