Tracking the true growth of e-commerce has become a kind of statistical puzzle in recent years. On the one hand, the sheer pace of digital transformation is obvious—anecdotes abound, headlines proclaim record-breaking sales, and there’s a general sense that online retail is outpacing every other form of commerce. But policymakers, economists, and anyone charged with understanding structural change need more than anecdotes. The real question is how to translate broad trends into hard numbers, especially numbers that allow for meaningful cross-jurisdictional comparison.
ISIC code 4791 (“Retail sale via mail order houses or via Internet”) offers an appealing starting point. National business registers increasingly rely on this code to categorize new entrants whose primary activity is selling goods online. In principle, this should provide a clear lens for measuring the expansion of e-commerce: simply count the number of new registrations under 4791 year by year, and one can produce a tidy chart of digital retail’s ascent. In practice, the process is more nuanced, with several important complications.
First, there is the matter of business registration practices. Not every country applies ISIC codes with the same level of consistency or rigor. Some jurisdictions update their business registers regularly, requiring companies to confirm or amend their primary activity; others do not. In the absence of such discipline, the 4791 code may be over- or under-used, especially if it is easier for a company to default to a broad “retail” code elsewhere. This inconsistency introduces bias, making international comparisons fraught unless one accounts for the specifics of each registry.
Even within a single jurisdiction, the story is rarely straightforward. A growing share of new firms operate hybrid models—part online, part physical. A “brick-and-click” business may register under a general retail code if its physical store is established first, then add e-commerce later, or vice versa. Others may cross-list, registering under multiple codes either for tax purposes or to maintain flexibility as business models evolve. This makes year-over-year comparisons of 4791 registrations tricky; an uptick may reflect administrative changes or shifts in self-identification as much as actual market growth.
So, how can one adjust for these complications? The first step is to dig deeper into the registration data itself. Where available, examine secondary codes, business descriptions, or declared activities beyond the primary ISIC listing. Some business registers maintain historical records that show the progression of codes attached to a firm over time. Tracking such changes helps to distinguish genuinely new e-commerce firms from established retailers expanding into online sales.
Next, consider supplementing registration data with other indicators. For example, national tax records can sometimes be disaggregated to show sales attributable to online versus offline channels. Survey data—though less timely—can shed light on the prevalence of hybrid business models and the proportion of revenue derived from e-commerce. Even web-scraping or commercial business directories, while not official, can help to cross-validate trends seen in the registry data.
Another useful technique is to look for discontinuities or sudden spikes in 4791 registrations and investigate the cause. Did a change in business registration policy, tax incentives, or e-commerce regulations occur around the same time? Such events can produce artifacts in the data that mimic real growth but are actually administrative in origin. Adjusting for these episodes—by flagging them in the analysis, or by constructing alternative time series that smooth out temporary distortions—leads to a more accurate trend line.
Policymakers interested in using ISIC 4791 registrations as a proxy for e-commerce growth should also be wary of cross-listing. Where possible, filter out businesses that hold multiple active ISIC codes, or at least note their presence in the analysis. In cases where hybrid businesses are significant, a simple count of 4791 registrations is insufficient. Consider a weighted approach, where firms are categorized based on the proportion of activity conducted online versus offline, drawing on survey or sales data as available.
International comparisons introduce yet another layer of complexity. Some countries may aggressively promote e-commerce registration, even subsidizing the process or providing fast-track approvals for digital-first businesses. Others may lag, either due to outdated registry systems or regulatory inertia. Before drawing conclusions about relative growth rates, analysts must normalize for such differences—whether by adjusting for population and overall business formation rates, or by anchoring their analysis to broader macroeconomic indicators.
Ultimately, the value of ISIC 4791 is in its ability to provide a standardized, if imperfect, view of digital retail’s trajectory. The process of measuring e-commerce growth, however, is not a matter of simple arithmetic. It requires a willingness to interrogate the data, to question anomalies, and to adjust methods as new information emerges. The trend is clear—e-commerce is expanding rapidly—but the details matter, particularly for policymakers who must design interventions, forecast tax revenue, or understand the broader implications for labor markets and urban infrastructure.