What changed
Working paper comparing paid ChatGPT visits with paid Google search and other channels on tracked store revenue
When
Data June 29, 2025 to August 22, 2026; paper dated October 2026; reported November 2, 2026
Who it affects
Kaiser (WU Vienna) and Schulze (Frankfurt School), data from Grips Intelligence; 387 sites with paid ChatGPT clicks, 102 with full sales tracking
What to do
Work out your own parity price per click before raising ChatGPT bids for the holiday season

Key takeaways

  • Paid ChatGPT visits returned $0.89 and converted at 0.37%, against $2.45 and 1.17% for paid Google search, in US data to August 2026.
  • The break-even click price against search is about $0.26, while reported bids run from $1.72 to OpenAI's suggested $3 to $5.
  • Organic ChatGPT referrals convert near search, so optimise for citations while capping paid tests at a parity bid.

Revenue per paid visit, US traffic

  • Paid ChatGPT0.89$
  • Paid Google search2.45$
  • Organic ChatGPT3.27$
Source: Kaiser and Schulze working paper, October 2026, via PPC Land

What the study measured

The paper, titled Can Advertising Fund Free AI, uses Google Analytics e-commerce data from Grips Intelligence. The organic panel is 1,898 sites with about $16 billion in annual sales. The paid panel is 387 sites with at least one paid ChatGPT click, of which 102 have full sales tracking. It is a version 1 working paper, not yet peer reviewed, and Kaiser is a scientific adviser to Grips, which the paper discloses.

Ads reached the US on February 9, 2026, then eight more countries through early August. The EU had no ads until August 24, two days after the data closed, which gave the authors a control group. The data records only last-click visits, with no prompts or impressions.

ChatGPT ads rollout

  1. Ads launch in the US

  2. Australia, Canada, New Zealand

  3. United Kingdom

  4. Study data window closes

  5. EU ads begin

  6. 89 countries, 100,000 advertisers

The paid numbers: $0.89 against $2.45

In US traffic, a paid ChatGPT visit converted at 0.37% and returned $0.89. A paid Google search visit converted at 1.17% and returned $2.45. Within the same sites, ChatGPT orders were $74 smaller, and against direct traffic paid ChatGPT was the only channel short on all six measures tracked.

The distribution matters more than the average. Of 87 advertisers that bought both channels, 84 earned less per ChatGPT visit. Seventy-three recorded no ChatGPT sale across 18,726 sessions, while 14 sites produced all 570 purchases. One advertiser accounts for 66% of paid ChatGPT purchases and 76% of US paid ChatGPT revenue.

Organic ChatGPT is a different story. Unpaid referrals converted at 1.31% and returned $2.48 per session, and $3.27 in US traffic, the highest of five channels measured. Those visits fell an estimated 6% to 8% in the 13 weeks after ads launched.

So what: the $0.26 parity price

The authors ask what a ChatGPT click would have to cost to match search's return per ad dollar. The median search cost per click in accounts with cost data was $0.72. Scale that by $0.89 over $2.45 and the parity price is about $0.26, with a range of $0.05 to $0.57. The largest publicly reported account, run by the agency Opascope, paid $1.72 per click over 15 days in June 2026, more than six times parity. OpenAI's help centre suggests a starting maximum bid of $3 to $5 and says it has no benchmarks yet.

At $1.72 a click and $0.89 per visit, each dollar of spend returns about $0.52 of revenue before product cost. Opascope's account did better, with 2.35% conversion and about $89,000 of revenue on $60,000 of spend, but that is one account chosen to scale. The lesson is that the average advertiser paid search prices for a channel returning a third of search revenue per visit, and a few took almost all the sales.

Worked example with illustrative numbers

These figures are invented to show the method, using the study's averages. A Shopify brand with a $240 average order and 40% gross margin tests $3,000 of ChatGPT ads at $1.72 a click, which buys about 1,744 visits. At 0.37% conversion that is 6.5 orders and about $1,550 of revenue, or $620 of gross profit against $3,000 spent. The same $3,000 in paid search at $0.72 a click buys about 4,167 visits and, at $2.45 per visit, about $10,200.

At the $0.26 parity price, $3,000 buys 11,538 visits and, at 0.37%, 43 orders and about $10,270, which matches search. So the question is whether your bids can sit near parity and still win impressions. If not, the budget earns more in search or Google Shopping until it does.

Find your parity price

  1. 1Revenue per visit

    ChatGPT and search, last 90 days in GA4

  2. 2Divide

    ChatGPT figure over search figure

  3. 3Multiply

    By your median search cost per click

  4. 4Bid there

    Cap the ChatGPT max bid at the result

What sellers should do this month

  • Analyst: in GA4, compare revenue per session for paid ChatGPT, paid search and organic ChatGPT over 90 days, excluding ad-review crawler sessions. Metric: three figures on one sheet.
  • Marketing lead: calculate your parity price as ChatGPT revenue per visit divided by search revenue per visit, times your median search cost per click. Metric: one figure per click.
  • PPC manager: in ChatGPT Ads Manager, set the maximum bid near parity rather than the $3 to $5 suggestion, in its own budget. Metric: average cost per click at or below parity.
  • Finance: cap the test at a fixed sum, such as the $5,000 to $10,000 Opascope suggests, with a stop rule on revenue per dollar. Metric: the rule is written before launch.
  • SEO owner: protect the organic side, which converts near search, with the AI search optimization guide and the brand mentions in AI search explainer. Metric: organic ChatGPT sessions month over month.

When this does not apply

The paid panel is mostly services, software, education and finance, while the organic panel is mostly retail, so a retail brand may sit anywhere in the range. Long-consideration purchases are penalised by last-click tracking, EU advertisers are not in the paid data, and conversion-optimised campaigns from June 5, 2026 are not analysed separately. OpenAI has reported one e-commerce advertiser with three times return on spend over 28 days, so individual results can beat the average.

Where PPC Spider helps

We set a payback rule per channel from margin after fees, so a new channel earns its budget rather than borrowing from search. Follow the Shopify ecommerce news channel for the next data point, or request a free Amazon audit if your traffic lands on Amazon.

Sources

Get the next change before your competitors do

We send the marketplace changes that matter each week, with what to do next.

By subscribing you agree to our privacy policy. Unsubscribe any time.
Want this done for your account?

We run these playbooks on client accounts every day. Start with a free audit.

Get a free audit