Can Social Media Enhance an Audit?

If you were working, traveling, shopping, or doing pretty much anything this past summer, the mere mention of CrowdStrike or Microsoft Windows might trigger some strong emotions. As some of you may recall, on July 19, 2024, a software update from cybersecurity vendor CrowdStrike caused most Windows computers to fail, and social media’s reaction to the outage was as widespread and talked about as the outage itself.

As these failures occurred, social media filled with images of Microsoft’s infamous “blue screen of death,” along with the general sentiment of unhappiness from users. On the flip side, social media posts from MacOS and ChromeOS users were far less negative. In fact, one Chromebook user, Debanish Achom, took to X (formerly Twitter) and posted: “Today, I laugh at those who said getting a Chromebook that boots up in five seconds was a bad investment.”

Setting aside the debates over which operating system is best, there’s much we in the accounting profession can learn from the CrowdStrike event.

From the auditing side of things, for example, research from the University of Illinois Urbana-Champaign explored whether user-generated social media posts can help inform auditors’ expectations for companies. Specifically, researchers Andrea Rozario, Ph.D., assistant professor of accountancy, and her colleagues Miklos Vasarhelyi, Ph.D. (Rutgers University), and Tawei (David) Wang, Ph.D., MBA (DePaul University), looked at whether consumer-generated posts can enhance the accuracy of preliminary analytical procedures.

WHAT AUDITORS CAN LEARN FROM NONFINANCIAL INFORMATION

Nonfinancial information, such as social media posts, can help auditors understand a company’s financial health, risks, and future performance. “It’s an interesting topic because of the ongoing debate in the audit community about how new, externally produced information could potentially enhance financial statement audits,” Rozario says.

Regulatory agencies, including the Public Company Accounting Oversight Board and International Auditing and Assurance Standards Board, have also recognized the growing importance of nonfinancial information for auditors, especially in the planning and substantive testing stages of the audit. Social media posts, like those on X, LinkedIn, TikTok, or Yelp, provide real-time insights into consumers’ attitudes and behaviors and represent a specific type of nonfinancial information that could be useful for auditors. This is especially true for consumer-generated social media posts related to companies in business-to-consumer industries.

“We found that consumer posts expressing interest in products or brands (for example, a person posting on social media that they want to buy an iPhone) improve the precision of auditor-developed revenue expectations in analytical procedures,” Rozario says. “These findings show that new sources of external nonfinancial information can provide incremental value to auditors.”

THE ‘WISDOM OF CROWDS’ THEORY

As motivation for their study, the researchers drew on a theory known as the “wisdom of crowds,” which says that the average prediction from a large group of people can sometimes be more accurate than predictions from individual experts due to the diverse perspectives and knowledge within the large group. Common examples of this include “guessing games,” where a large number of people are asked to provide an estimate on something, like the number of jelly beans in a jar or weight of an animal at a state fair. With guessing games, the average of all of the individual guesses is often surprisingly close to the actual amount.

If this theory holds, social media information, when properly aggregated, could provide insights about company performance. For example, if the average post about a product reflects an intent to purchase that product, then it would be reasonable for auditors to predict increased revenues for a company. On the other hand, because social media is typically not monitored or filtered, the resulting information may not be useful. For example, posts made by the companies themselves could be biased. Additionally, the increasing number of bot accounts on social sites could lead to data that doesn’t reflect the attitudes and preferences of real people.

Because data in social posts can be noisy, the researchers used measures compiled by LikeFolio, a software company that analyzes social media data to accurately predict shifts in consumer behavior. As noted by Rozario and her colleagues’ research, LikeFolio specifically “captures tweets that are posted by individual consumers that express a recent past purchase, or future purchase, and positive or negative sentiment about a product or brand.” LikeFolio uses its database of products and brands (along with keywords related to products and brands) and combines it with text-mining techniques to capture posts that relate to either consumer interest or consumer sentiment about a specific product or brand.

Additionally, posts about the company that owns the product or brand are excluded from the research. For example, posts about Apple’s stock or CEO would be excluded, but posts about Apple’s iPhone or Mac computer products would be included. The researchers note that “because tweets are extracted at the product or brand level and then aggregated at the company level, they can provide a more complete picture of aggregate consumer behavior.” In other words, posts that relate to consumer interest or consumer sentiment about a product or brand reflect the “wisdom of the crowd” about that product or brand.

CONSUMER INTEREST VERSUS CONSUMER SENTIMENT

Next, the researchers used linear regression as an example of a preliminary analytical procedure that could occur in a financial statement audit. A benchmark model was used to predict revenues based on both company-specific and macroeconomic information (e.g., lagged sales, accounts receivable, and gross domestic product). Then, the researchers estimated two more regressions: one with LikeFolio’s Tweet Consumer Interest measure added to the benchmark model, and one with LikeFolio’s Tweet Consumer Sentiment measures added to the benchmark model. The study included 1,824 firm-quarter observations (where each firm-quarter represents quarterly results for one company) from 76 companies in 20 business-to-consumer industries during the years 2012-2017.

Overall, the researchers found that information about consumer interest in products and brands—not consumer sentiment—improves the predictions and error-detection performance of analytical procedures of firms in most consumer-facing industries.

“The general idea is that consumer-generated postings expressing interest in products or brands can be useful in predicting company performance, so our theory extends to other social media platforms, like Facebook and Instagram,” Rozario notes. “However, something to consider is that some social media platforms may impose restrictions on how much data can be accessed, which could make it more challenging to extract this information.”

INSIGHTS FOR CORPORATE FINANCE TEAMS

Consumer interest information may have benefits for corporate finance professionals too. “While our paper focuses on the benefits to auditors, the implications extend to management,” Rozario explains. “Consumer interest data from social media can provide management with real-time insights into product demand, brand perception, and trends in the market. This kind of information can help companies make more informed decisions about inventory management, marketing strategies, and revenue forecasting.”

Overall, Rozario and her colleagues’ research shows that social media provides real-time insights that auditors and others can leverage to keep pace with fast-changing consumer behavior. While still an emerging area, their research offers a glimpse into how innovative data sources can become useful tools for both accountants and auditors.

When asked whether information like this would have been helpful during her time as an auditor, Rozario says: “Yes! Considering the potential of external nonfinancial data to correlate with an audit client’s business activities, I believe this information would’ve been highly valuable in enhancing the risk assessment process.”

Related Content:

• What Drives Difficult Auditor-Client Interactions? The relationship between an auditor and their clients is very complex and often met with conflict. Here, researchers explore these dynamics to identify areas for improvement.
• What Happens When Audit Meets AI? New research on artificial intelligence in audit examines the impact when human auditors trust—or don’t trust—the machines.