In the initial promotion stage of its advertising business, OpenAI has offered high prices comparable to advertisements on popular video programs such as the National Football League (NFL), and its pricing is much higher than competing social media products such as the Meta platform. However, unlike Meta or Google, OpenAI will not provide advertisers withDetailed data on ad-related query feedback, and cannot tell whether the ad prompted ChatGPT users to make purchases, visit the website, etc.

OpenAI CFO Sarah Fryer and CEO Sam Altman
OpenAI CFO Sarah Fryer and CEO Sam Altman

Although OpenAI may launch such data services in the future, it will only require building more complex advertising technology tools, which will take some time. This also means that if OpenAI wants to build a business landscape that can compete with leading advertising platforms, there is still a lot of work to be done.

A media buyer who serves some advertisers revealed that OpenAI has made it clear to early advertisers that it will provide data on ad impressions (that is, the number of times an ad is viewed) and total clicks. An OpenAI spokesperson also confirmed that advertisers will receive macro-level data insights such as total ad views, which is similar to the type of data provided by TV networks.

But in the past 20 years, the core reason why Google and Meta have surpassed the TV industry and become the world's largest advertising sales platform is that they can provide advertisers with refined data and help marketers measure the return on advertising investment.

Industry insiders believe that as time goes by, advertisers will expect OpenAI to introduce more advanced advertising technology, which will not only provide targeted data such as portraits of ChatGPT users who saw the advertisement, but also track whether the advertisement directly promoted the transaction of goods.

Olivia Corey, chief strategy officer at marketing analytics company Haus (who previously worked at Netflix and Quibi), said: "When an advertising product is still in its infancy and demand is strong, there is really no need to rush to improve these features. But as the product continues to mature, advertisers will inevitably exert more pressure on the platform to supplement data capabilities."

Earlier this month, OpenAI announced that it would be adding ads to the free and minimally paid versions of ChatGPT in the coming weeks. The push into advertising comes at a critical stage as OpenAI plans to raise tens of billions of dollars in new funding. The company aims to generate $11 billion in annual revenue from non-paying ChatGPT users by the end of next year, and ad sales are central to achieving that goal.

The above-mentioned media buyers said that in order to finalize the first batch of advertising orders, OpenAI’s corporate cooperation team chose to contact potential advertisers directly instead of negotiating through advertising agencies. The first batch of signed customers include not only OpenAI’s existing enterprise users, but also well-known large enterprises in multiple industries. In contrast, Meta and Google rely heavily on their large numbers of small and medium-sized business customers for their advertising revenue.

The person also revealed that OpenAI’s first batch of ads will be displayed below the ChatGPT reply content, with a target pricing of approximately$60 CPM. Analysts note that this price is on par with advertising pricing for streaming targeted ads and top TV properties like NFL live broadcasts, while Meta’s CPM is typically less than $20.

Analysts and marketing executives believe that given ChatGPT’s large user base and the fact that users often reveal clear interests and preferences during conversations, many advertisers are likely to be willing to pay this premium for the first batch of advertising spaces. For example, when users search for open-ended needs such as "high-quality luggage suitable for travel" on ChatGPT, they often have not yet targeted a specific brand. This type of traffic is extremely attractive to advertisers.

However, most advertisers may not invest large budgets until they are sure that ChatGPT ads can actually drive sales. The Information Daily previously reported that the first batch of advertisements was just a test for OpenAI, and subsequent platforms may adjust their strategies based on test results.

Brian Stanpek, founder of the artificial intelligence marketing startup Evertune AI (who previously worked at the advertising technology company The Trade Desk) said: "The most critical task for ChatGPT next year is to build the link between advertising and sales conversion, that is, to open up the downstream links of the marketing funnel."

In fact, it is not unprecedented for companies not to be equipped with complete marketing technology tools in the early stages of advertising business. About ten years ago, Facebook (now Meta) initially sold ads mainly based on impressions as the core indicator, and then gradually launched tracking tools such as ad clicks, app downloads, and product purchases.

Advertising and e-commerce analyst Andrew Lipsman pointed out: "This was the development path of Facebook in its early years. In the first few years, the platform always focused on the two selling points of user reach scale and exposure. It was not until 2015 that Facebook completely transformed and improved its performance tracking capabilities, and its advertising business also ushered in explosive growth."

An OpenAI spokesperson said that the company has established a dedicated advertising sales team to promote related businesses, and did use the enterprise cooperation team to carry out customer development work in the early stage. He also emphasized that the advertising pilot project is not only for existing enterprise customers.

Another advertising industry source revealed that OpenAI is currently cooperating with large advertising agencies. These agencies usually develop long-term media placement plans for big brands. The purpose of the cooperation is to tap more new advertisers. The above-mentioned media buyers also mentioned that OpenAI has not yet cooperated with advertising technology companies for advertising space sales or delivery management.

"It's not difficult to get test budgets from advertisers, but if you want to increase annual advertising revenue to $10 billion or $15 billion, you have to develop a joint business plan with large partners like Omnicom and Coca-Cola," Stanpek said.

E-commerce linkage potential

If OpenAI wants to further improve the measurability of advertising effects, there is another way to deeply bind advertising with the checkout function added by ChatGPT since September last year. In theory, the combination of advertising and checkout within the chat interface is expected to give rise to a “see now, buy now” shopping advertising model.

Currently, ChatGPT’s checkout function can realize the monetization logic of “sales sharing”: merchants do not need to pay for ad clicks or exposures. Only when users complete purchases through ChatGPT, they need to pay commissions to OpenAI. Starting this week, most merchants in the United States who use Shopify to build their websites can choose to connect to ChatGPT to sell goods. After the free trial period expires, they need to pay OpenAI a 4% share of the sales.

In the early days, the ChatGPT checkout function was mainly targeted at small online stores and direct brands. Such merchants usually relied heavily on advertising from Meta, Google, and Amazon to acquire customers. Olivia Corey's clients include direct brands such as Mejuri and Jones Road Beauty. She said: "For these merchants, accessing ChatGPT is as simple as a 'one-button switch', with almost no trial and error costs, so they are willing to try any new channel."

However, shopping within chatbots is still a new experience for consumers, and consumption habits have not yet been formed. Some brands are skeptical about the sales conversion effect of ChatGPT checkout function.

Yoga Acharya, president and chief operating officer of protein bar brand Mezcla, said that the brand’s online sales mainly rely on the Amazon platform. He questioned: "If users are just searching for information about protein bars on ChatGPT, is this considered a high purchase intention? Or are they just doing product research?"

He also added: "Our customer acquisition cost on Amazon is indeed high, but when users search on Amazon, they often have a clear purpose of purchasing, and the conversion efficiency will be higher."