If retailers had been to rely completely on latest historic information, they may conclude that UK customers purchase their annual provide of bathroom paper and pasta in March. Why? In early March 2020, UK customers had been anticipating the nation to enter lockdown. This drove panic shopping for, particularly for these staples. Immediately that info stays in corporations’ buyer insights and advertising and marketing databases alongside different predictive purchaser behaviours like turkey gross sales in December or pumpkin gross sales in October. However which behaviours are prone to repeat, and that are anomalies?

The difficulty continues within the (fingers crossed!) post-Covid-19 interval as some sectors get pleasure from a considerable bounce again. Ought to corporations, having seen 50-100% year-on-year progress, consider it to be a development and plan for a similar within the coming yr? Doing this is likely to be wise, however it might equally imply missed forecasts and extra inventory sitting on cabinets.

And the issue extends past provide and demand and buyer engagement information. For instance, I see anecdotal proof of digital occasion attendance ranges dropping off alongside hesitancy to decide to in-person vent attendance — an unwelcome double whammy.

A colleague lately advised that whereas it is likely to be apparent to folks what information represents a development, it is probably not so apparent to an algorithm. This difficulty is highlighted in Forrester’s 2022 International Advertising and marketing Survey, the place respondents recognized the quickly altering enterprise setting making historic information meaningless as the highest measurement and analytics problem.

Advertising and marketing leaders should rigorously contemplate how you can interpret and use information which may be anomalous. Contemplate this instance from my first job: Whereas some shoppers had been visiting my firm’s community operations centre, the large digital community map lit up with purple warnings. The guests had been excited to a community disaster response first-hand; nonetheless, the employees carried on like nothing was taking place. Once they queried the presenter, he stated that what we don’t need in a community centre are the proverbial headless chickens who react to information with out pondering (this was pre-ML/AI). Diving right into a community and making quick modifications primarily based on a short-term change within the information was typically detrimental. The coverage was to attend to see if the information level turned a development and solely then take motion. As if to show his level, the purple alerts began disappearing from the display screen and the community went again to regular.

There isn’t a magic method that entrepreneurs can apply to determine what to do with the final three years of buyer and market information. All companies are completely different, and every will want to consider their very own circumstances and what impacts the pandemic had on their clients’ and markets’ buying patterns. Primarily based on this info, corporations can determine whether or not the information that you simply’ve captured stays legitimate, must be calibrated, must be quarantined, or must be erased. What’s essential is that you simply make acutely aware selections to keep away from sleepwalking into automated data-driven errors that might significantly affect enterprise efficiency or steadiness sheets.