B2B buyers are deciding before they speak to vendors, says Tug’s Michael Ozburn
Michael Ozburn of Tug argues that marketers need to shape how AI tools describe their products and get sharper at routing buyers between self-serve and sales.
A growing share of B2B buyers now choose what to buy without ever contacting a vendor, and businesses need to find ways to influence those decisions much earlier, according to Michael Ozburn of Tug.
Ozburn says the modern B2B SaaS buyer has had enough of the traditional sales pitch, pointing to figures from Apollo.io showing that 69% of B2B buyers now finish their purchase process before speaking to sales, and 61% would rather buy without a rep at all.
Product-led growth is spreading quickly across the industry, he says, with ecommerce checkouts, self-serve tiers and free trials taking over from the gated demo as the standard entry point. Small and mid-market decision-makers in particular no longer accept a sales call as the cost of getting in. “They expect hands-on access to the product first, and they’re getting it.”
The bigger shift, in his view, happens before a buyer reaches a pricing page.
AI tools take over early research
Ozburn cites Forrester’s 2026 Buyers’ Journey Survey, which questioned roughly 18,000 business buyers worldwide. Before contacting any vendor, it found, 55% use AI tools to compare suppliers, 54% research products and 47% put together internal business cases.
Buyers are using AI assistants and LLMs for the early work once done by a sales rep, he says: assessing needs, carrying out the first evaluation and drawing up a shortlist. By the time they will speak to anyone, they have already decided what they need and which supplier might deliver it.
Buyers would rather get impartial information about a product than endure an upsell, he argues. “An AI assistant doesn’t have a quota to hit. A sales rep does.” Offered the choice, most SMB decision-makers will pick the conversation that is not trying to close them.
For marketers, he says, the self-education stage used to be something sales teams could steer through discovery calls and guided demos. It now happens where marketing has much less visibility and control. “The evaluation is still happening. It’s just happening without you in the room.”
Managing how LLMs describe products
Ozburn says most marketing teams have yet to catch up with the consequence: if buyers ask an LLM to assess a product before talking to its maker, the model’s answer matters as much as the company’s own website.
Teams now need to actively manage how AI models present what they sell, what value they offer and how they compare with rivals, as they have long done with SEO.
Whether it is called GEO, generative engine optimisation, or something else, he argues the mechanics matter less than the discipline: keeping public information accurate, consistent and up to date wherever an AI model might draw on it. “Get it wrong, and you’re not just losing a search ranking. You’re losing the pitch before you knew one was happening.”
Balancing self-serve and sales-led routes
Ozburn stresses that sales-led growth is not dead. Instead, marketing teams need to be better at deciding which buyers go where. Some segments are ready to go straight into a free trial, while others still need a conversation with a person to close.
The task is to identify each group and design ad campaigns and lead-form workflows that separate them, without pushing a self-serve buyer into an unwanted call.
He says this calls for continuous testing rather than a one-off decision. Marketers should A/B test which segments convert better via a self-serve trial or a traditional demo, and shift the balance as data arrives.
Treating it as a fixed rule risks losing buyers at both ends, he warns: self-serve buyers abandoning sales calls they were funnelled into, and high-touch accounts left alone on a free trial they would never have closed without help.
Judging results on unit economics
According to Ozburn, a dual-motion model should not be judged on lead volume. What counts is what happens after conversion: deal size, churn rate and the ratio of lifetime value to customer acquisition cost (LTV:CAC).
A surge of self-serve sign-ups is worthless if those customers churn in their second month, he says. The approach only works if it is measured on figures that show whether the business is genuinely healthier, rather than just busier.
He argues that buyers have already taken charge of how they evaluate software, and that companies which design around this, instead of resisting it, will end up with “a pipeline that scales and a bottom line that holds up.”


