Why Businesses Should Optimize for AI Search Before ROI Is Clear

In a recent LinkedIn post, Neil Patel discusses why businesses may be overlooking large language models (LLMs) as a way to reach people making purchasing decisions. Patel’s post argues that uncertainty about return on investment is keeping companies from optimizing their presence on tools such as ChatGPT, Gemini and Claude.

“People aren’t sure if LLMs like ChatGPT are really useful for driving sales.”

The ROI question is holding companies back

According to Patel, businesses cite the difficulty of measuring returns as a reason to avoid investing in visibility across LLMs. He says this hesitancy persists even as consumers report using the tools. Patel cites a figure from Julius Nyambok that roughly 16% of people say they use an LLM, with usage higher in countries with greater GDP, including the United States.

That figure is presented in Patel’s post as a signal of growing adoption, though it does not by itself establish how often those users rely on LLMs to choose products or how much sales the tools generate. The distinction helps explain the central tension in his argument: interest may be visible before its commercial impact is easy to measure.

A gap between consumer behavior and business activity

Patel points to what he describes as a mismatch between consumer use and company preparation. “When we look at our data,” he writes, 14% of people use LLMs for purchasing decisions, while only 3% of companies participate in optimizing their businesses to be found on those platforms.

“14% of people use LLMs for purchasing decisions, yet only 3% of companies participate in optimizing their businesses to be found on these same LLMs.”

Those percentages are figures Patel attributes to his organization’s data; the post does not specify the sample or methodology. Even so, the comparison frames his concern: companies may be slow to respond to a behavior that Patel says is already influencing some purchase decisions.

The SEO comparison—and its limits

Patel likens the current uncertainty around LLM visibility to the early days of search engine optimization. In his view, businesses once faced a similar challenge proving SEO’s return, but the widespread use of Google made discoverability important.

“Hard to prove ROI, but you knew everyone uses Google. So if you weren’t on it, you were making a big mistake.”

The analogy is Patel’s case for acting before measurement becomes straightforward. It suggests that businesses should pay attention to where customers seek information, even when attribution is imperfect. But the comparison does not establish that LLM optimization will deliver the same scale or economics as SEO. Patel’s post instead raises a strategic question for marketers: how much evidence of sales impact do companies need before they adapt to a changing discovery channel?