It's like choosing between a rock and a hard place: Will Google cannibalise its own highly profitable search engine advertising business with Gemini, or will it prefer to lose ground in AI? One thing is clear, however: the massive competition from generative AI is threatening Google's business model and forcing the giant to disrupt itself.
The most successful business model in the digital economy – with a volume of around 300 billion US dollars worldwide – is under pressure: search engine advertising. We may even witness a disruption. Artificial intelligence means that online search engines as we know them today are likely to be replaced by answer engines in the future. Not immediately, of course, but possibly gradually. For the first time, the search algorithm is no longer so superior that Google automatically outstrips its competitors. On the contrary, Google itself is faced with the key question of how far it wants to compete with its own business model in order to remain viable in the future.
How quickly this development will take place and whether AI assistants will triumph depends on users, but also on the quality of the answer engines. ChatGPT, Google Gemini, Microsoft's Copilot & Co. are already working today, but according to completely non-transparent criteria. From a marketer's perspective, this means that products and brands do not know why or whether they are listed in the answers. And users do not know why the AI gives them certain recommendations.
If we analyse the topic in more detail, there are at least three relevant perspectives:
1. The user's perspective
From the user's point of view, a response engine is more convenient than a search engine. For one thing, you no longer have to type in the question. You can also speak it. The answer then comes via voice output. However, it is also significantly more limited in its recommendations. While the search engine spits out pages of hits, the response engine limits itself to a few hits.
And this is where the first dilemma arises: why does the AI-supported answer engine recommend product A and not product B, with which the user has had better experiences in the past? Why does it list the more expensive product and not the cheaper option? Why are American or international products and brands listed more often than national and regional ones, or not at all?
These are all questions for which there is currently no answer. Why the systems give certain answers (and on what research basis) is largely unclear.
A concrete example. We feed four AIs (Bing's Copilot, ChatGPT 4.0 from Open AI, Google's Gemini and Perplexity.ai) with the following prompt: ‘I want to buy an electric car. It should have a range of at least 300 kilometres per charge, seat four people and cost no more than €55,000 as a new car. It should also have the fastest possible charging times.’
The result: the AIs recommend 4 (Bing and Perplexity) or 5 car models (Gemini and ChatGPT) respectively. Only one model, the Hyundai Kona Electric, appears on all four AI lists. The KIA e-Niro is listed three times, while two lists recommend the Skoda Enyag iV and the VW ID3 or Tesla Model 3. The VW ID 4, BMWiX, Lucid Air, Renault Zoe and Citroen eC4 are mentioned once.
Regardless of the quality of the recommendations (some models do not meet the criteria from the prompt), it is clear that only Copilot from Bing and Perplexity.ai list the sources for their recommendations. For users, it is not possible to verify the origin of the information for two of the four AIs.
What does this mean for users? The complexity of the response engines is decreasing, but at the expense of quality and transparency. This is unlikely to be significant for general and simpler questions, but for more complex searches, it will probably mean that these will continue to be solved with classic search engines in the medium term.
2. The perspective of companies/brands/advertisers
Conversely, however, it is also incomprehensible to the manufacturers in our specific example why some models are listed and others are not. The selection for the rather generic electric car prompt (see above) could have been completely different.
In the logic of search engines to date, two disciplines have established themselves with SEO (search engine optimisation) and SEA (search engine marketing), where the rules of the game were and are at least reasonably transparent. For advertisers and brands, it is not yet clear why they are or are not included in the 4 to 5 recommendations provided by AI systems.
Google recently announced at its ‘Google Marketing Live’ event that advertisers will probably be able to buy their way into the AI overviews in the future. While the purchasing mode for paid search is largely clear in the current paid search, there is no known procedure for the AI overviews yet. Speed and transparency are also required because the systems (see practical example) are already providing answers.
And here it will be very exciting to see what any embedded advertising will look like and how it will be accepted by users. After all, people often ask a clear question that they want answered precisely. What they don't want are 100 possible shopping results or answers where they don't know exactly whether they are organic or paid. They want an answer and not ‘Radio Yerevan’: ‘In principle, the answer is yes, but...’
3. Google, Microsoft & Co.
And last but not least, this is the disruption for the creators of the systems: Open AI has presented Google with a decision with ChatGPT: Will Google cannibalise its own highly profitable search engine advertising business with Gemini, or would it rather lose ground in AI? The AI overviews now presented at Gemini suggest that Google is aiming for the gentlest possible self-cannibalisation. However, the search market leader faces competition from Microsoft, among others, which is a major shareholder in Open AI on the one hand, but on the other hand can be active far beyond the Bing browser with the integration of Copilot. And from its competitor Open AI, which does not have to take existing online search into account at all. Fun fact: the lists from Bing's Copilot and ChatGPT 4o differ significantly in the practical example.
Almost half (43.4%) now have an account with an AI service, according to a representative study by Convios Consulting GmbH on behalf of GMX and WEB.DE. Fun and entertainment are the most common reasons for using AI (39.7%), followed by research at 38.9%. The figure is likely to be significantly higher at present. User acceptance of answer engines will determine how quickly market share shifts from Google's search engine advertising to Microsoft and Open AI in particular.
I believe that this current development will shape our industry discussion for at least the next two years and that the search market will change significantly.
Answer engines are an important overarching market topic. That is why our industry associations are called upon to act: the BVDW for digital agencies, service providers, publishers and also advertisers (retail media), the OWM for all advertisers and, of course, other market partners and regulators (cartel office, etc.). We urgently need more transparency about the rules by which answer engines operate. It is time for open discussions between market partners, time for framework conditions, for white papers or codes of conduct that give everyone certainty about what they can expect from AI and what limits must be observed. The associations are the right and best platform for moderating this discussion and also for accompanying it in the political arena.
Let's not allow too much time to pass. The systems are already up and running. And ask your bubble how often they have asked an AI product-related questions. You will be surprised. My colleague just planned her holiday to Portugal with ChatGPT 4o.