Google or ChatGPT: where does advertising pay off for your online shop?

New ad placements need a clear commercial assessment

Today, your customers can discover products through search engines, social networks and AI assistants. This creates additional ways to reach your target audience. At the same time, it becomes harder to decide where your next advertising budget is best spent.

The key question is: which channel brings in additional customers at a cost that suits your business?

ChatGPT advertising is now available in Germany, too

In 2026, OpenAI expanded ChatGPT Ads to Europe, including Germany. Following the update on 31 August, access via Ads Manager is available in the 31 announced European markets. OpenAI also describes measurement options via a pixel and a Conversions API.^[1]

According to OpenAI, ads are kept separate from answers, and advertisers do not get access to private conversations. Booking an ad is therefore different from an unpaid product recommendation by ChatGPT.^[2]

This new environment can be particularly interesting when people are clarifying requirements or weighing up alternatives. Whether it generates commercially attractive orders for your shop is a question to test – not a given for every retailer and every product range.

Google is evolving as well

Google is increasingly connecting search, product data and AI-powered experiences. In May 2026, Google introduced new ad formats for AI search, including tests and the further development of existing pilot programmes. Availability differs by format and market.^[3]

Which channel fits which job?

The following overview describes sensible test hypotheses, not a guaranteed ranking.

Why a high ROAS is not yet a good budget decision

ROAS relates attributed revenue to advertising costs. On its own, it says nothing about how profitable the orders are or whether they would have happened without advertising.

A simplified example with fictitious figures:

Channel A looks stronger at first. Under the conditions assumed here, more is left over with channel B. A real assessment has to take cost of goods, discounts, variable costs and returns into account in a way that fits your business.

How to make a channel test meaningful

1. Set the goal. For example, additional new-customer orders at a fixed maximum cost.
2. Define the framework. A limited budget, a suitable observation period and the same definitions of purchase and revenue across all channels.
3. Review the results in full. Together with returns, the share of new customers and purchase delays. With only a few orders, the results remain of limited significance.

Also, platform reports must not simply be added up: several providers can claim the same purchase.

Whether advertising causes additional purchases can be examined with suitable control groups or other robust experiments. The usual attribution of a purchase to a channel does not answer this question on its own.