Маркетологи США массово внедряют синтетические данные на основе ИИ для принятия стратегических решений

James Whaley is the CEO of OvationMR. I’ve observed many brands across various industries increasingly use synthetic (AI-generated) data to make major decisions, such as which new features to add to a product and which new service lines to add to their offerings. For instance, a dog food brand might ask an LLM to simulate respondents who are dog owners in their mid-20s to mid-30s living in metropolitan U.S. cities rather than survey people in that demographic. A main reason I’ve heard marketing leaders give for doing so is a fear that traditional market research methods, such as surveys and online panels, can be corrupted by AI-generated or AI-assisted responses. Therein lies the paradox. If marketing leaders trust AI to answer strategic questions in a synthetic environment, why distrust AI when it’s used to generate or assist responses in traditional market research methods? Key Reasons Driving Synthetic Data Adoption For a 2025 report, SAS, an AI and data solutions company, surveyed “marketers who are decision makers at organizations across a range of industry sectors.” The survey drew respondents from “300 organizations worldwide, ranging from SMBs (small and midsize businesses) to enterprises with 10,000+ employees.” The survey found that 18% of respondents were currently leveraging generative AI for synthetic data in marketing, while 54% intended to do so in the future. In my view, a few key reasons are driving synthetic data adoption in marketing. First, there’s budget pressure. If marketing leaders and other stakeholders see the promise of cheaper answers in synthetic AI tools and are working with shrinking budgets, they may see synthetic data as a more viable solution. There’s also the appeal of getting answers faster. Traditional market research can take days, weeks or months. Synthetic data can potentially deliver insights in as little as a day. “Good enough” accuracy is another factor. How many decisions really require a 95% confidence level? I’d wager that few do. For some marketers and other stakeholders, a directionally accurate answer may be sufficient for certain decisions, particularly when speed and cost are priorities. Synthetic data solutions can offer a faster, more efficient way to get those directional insights. Finally, at some organizations, leaders may be pushing for AI adoption, seeing it as a strategic priority that can improve margins and boost competitiveness. A department that doesn’t use AI could risk being seen as behind the curve. The Pros Of Synthetic Data Synthetic data can add value to organizations when leaders use it strategically. For one, synthetic data can be an effective tool for data reduction. It can help marketers and stakeholders organize data and narrow down their focus areas before undertaking live research. Synthetic data can also support concept generation, such as drafting message pitches mapped to customer personas and exploring which pain points a product or service might address for different personas. Additionally, synthetic data can help with prescreening product and service concepts. For instance, marketers could use synthetic data to eliminate product line options that clearly don’t align with historical customer attitudes, provided that the organization’s historical customer data was one of the inputs used to generate the synthetic data. The Cons Of Synthetic Data However, synthetic data also has its disadvantages. If leaders overrely on synthetic data, those disadvantages could have bad directional and financial consequences. A key shortcoming of synthetic data is the risk of sycophancy. A user can iteratively prompt an LLM, potentially influencing it to return an answer it thinks the user desires. Marketers and other stakeholders can end up with the answers they want to hear, not need to hear, as they conduct market research. Another drawback of synthetic data is that it can be backward-looking by design if its main input is a company’s historical data. It can show leaders what’s in the rearview mirror but not what’s ahead in the front window (as traditional market research methods such as surveys and focus groups can). Other inputs for synthetic data also have disadvantages. For example, AI-generated assumptions are just that—assumptions. Then there’s what I call the robot vacuum analogy. Like a robot vacuum, synthetic data can follow an optimized path without accounting for situational awareness and changing conditions. For instance, it can tell a marketing team at a snack company that customers are likely to highly value a certain ingredient in its line of chips, not accounting for the fact that supply disruptions have made it difficult to acquire that ingredient. When companies rely too heavily on synthetic data, they risk missing out on two important opportunities. The first is the opportunity to align with what real customers actually want. A synthetic data tool might take a list of, say, 10 product features and narrow it down to the top three, so marketers can present the top three to customers and have them pick their favorite. But who is to say that customers would have even selected those three product features as the top three to begin with? The second is “the big surprise”—insights from real customers that reveal new pain points and, in turn, drive true innovation and shifts in competitive strategy. People can share opinions shaped by their authentic, real-world experiences. Synthetic data doesn’t have authentic, real-world experiences. How Leaders Can Balance Using Synthetic Data With Traditional Market Research Methods Leaders should balance using synthetic data with traditional market research methods. They can leverage synthetic data for data reduction, concept generation and prescreening. As for traditional market research, they can use methods that fit their unique needs, such as surveys for forecasting and ethnographic studies to capture behavior consumers may not self-report accurately. More broadly, leaders should embrace the “AI-in-the-loop” approach rather than the “human-in-the-loop” approach. They should elevate human judgment, context and nuance instead of letting them take the backseat to AI. Leaders and their teams should adopt the mindset that “it’s a great day for discovering why.” Because ultimately, discovering “why” is the point of market research. And good market research comes down to sound methodology that can uncover meaningful insights. There doesn’t have to be a paradox. Leaders don’t have to treat synthetic data and traditional market research as an “either-or” choice. There’s room for both. Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?