It’s no wonder Google has reversed AI and once again put links back at the top of its pages and moved the AI summary down the page and in importance. – The decision came right after the company lost half a trillion dollars on the stock exchange in a matter of hours.
An early comparison of online advertising shows a brutal difference: Visits from Google search ads generated far more sales revenue than visits from paid ChatGPT ads. The difference is even greater when you factor in what advertisers actually paid.
Real Madrid bring in tennis icon Rafael Nadal to set the course
Marketing researchers Maximilian Kaiser and Christian Schulze found that a visit from a Google search ad brought in an average of $2.45 in sales revenue. A visit from a paid ChatGPT ad, by comparison, yielded only 89 cents.
About 1.17 percent of Google’s search ad visits ended in a purchase – compared to 0.37 percent for ChatGPT ads.
Most had zero sales
Among the 102 advertisers with complete revenue data, a whopping 82 percent recorded zero purchases from their ChatGPT ads. Using publicly reported ChatGPT ad prices, the researchers estimated that Google search ads generated about 6.6 times as much sales revenue per ad dollar.
The number measures revenue and depends on the prices used in the calculation.
The researchers used two types of comparisons because a click and a sale tell different stories. Revenue per visit shows how much money the resulting traffic brought in on average. Revenue per ad dollar also takes into account what it cost to attract the traffic – which is why the reported ad prices are important.
The average can hide biases
The figures also show why the average can hide a skewed result. A group of advertisers can make money from a channel even if most of the individual companies in the sample do not have sales from it. The finding applies to the advertisers for whom the researchers had complete income registers.
A customer coming from a search ad may already be looking for a specific product. Someone who clicks on an ad while chatting may find themselves at a completely different point in their decision about what to buy. The comparison shows a difference in outcomes, but allows for more explanations for how and when people shop.
Raises questions about the business model
The findings also raise a question of how a chatbot pays for a free service. Advertisers need enough sales to justify spending, while those asking for help need to be able to see when a recommendation has been paid.
The working paper also looked at unpaid store links in ChatGPT responses. After the advertisement appeared, stores received an estimated 6 to 8 percent fewer visits from these links. The percentage of visitors who made a purchase remained about the same. That difference is important for retailers, because a drop in traffic can affect sales even if the visitors who arrive are just as likely to buy.
Neve Campbell on “Scream 8”: “I’m going to be in it”
Early phase – a lot can change
The comparison captures an early period of ChatGPT advertising. Prices, placements, and the composition of advertisers can all change. Search and chat also provide customers with different ways to explore a purchase.
The study measures how these channels performed in the observed rollout – with a separate question for OpenAI on how to increase ad revenue while keeping users confident in the recommendations they receive.
Source: Frankfurt School of Finance & Management.
