Error 404: Web not found, killed by Google’s AI search

IT’S been 24 years since Google Search became the dominant tool to find everything on the internet, but now Google is upending the system it helped engineer by injecting artificial intelligence (AI) into the equation.

Now companies everywhere, including South-east Asia, are very nervous about what this means for their business.

“There’s generally panic,” said Vahe Arabian, the founder of State of Digital Publishing (SODP) and SODP Media. But more than just panic, he said most people are interested in learning what the new norm is going to be about.

“They are trying to get ahead. They are looking for clarity,” he said.

The reason for the current level of panic and curiosity started on May 14, 2024, when Google rolled out AI Overviews, a new version of its search engine that gives users an AI-generated synopsis to their query, instead of the usual lists of web pages and links.

The response from the tech community was immediate.

Many industry researchers, who had already been monitoring AI Overviews – previously known as Google’s Search Generative Experience – warned the feature would kill web traffic, reduce revenue from advertising, and make it much harder for users to find smaller businesses online.

According to tech research firm Gartner, by 2026, traffic generated by traditional search engines will drop by 25 per cent, with the majority of traffic going to AI chatbots and other virtual agents. ChatGPT has been setting new records in web traffic on almost a monthly basis and saw over 2.5 billion visits in June.

Advertising, which is often directly correlated to web traffic, is also expected to take a hit, with many industry insiders commenting on the irony of Google’s self-sabotaging technology. The company’s ad revenue grew to US$237.8 billion in 2023.

Raptive, an Internet media consultancy, estimated AI Overviews could cost the industry US$2 billion a year in ad revenue.

What is the level of impact on businesses?

Pramod Rao, founder and CEO of Threado, a US-based developer of AI chatbots, thinks Google is putting user experience ahead of monetisation due to the current market demand.

“That may have an adverse effect on their ads business, but they are doing it primarily because the user experience is moving in that direction,” he said.

Just how soon the impact from AI Overviews and other AI-driven search engines could be felt in South-east Asia and other regions is unclear. Google has said it is focusing on the US market first, but the new feature has started to show up in search results in South-east Asia.

Microsoft Bing also uses proprietary AI in its web searches, while Meta now uses results from both Google and Microsoft.

But so far, AI Overviews has struggled in its home market. The initial launch saw multiple issues with hallucinations and even more mundane – and well-publicised – problems, like the AI not being able to tell parody from reality.

In one example, it famously quoted an article from The Onion – a well-known parody news site – that recommended using glue to get cheese to stick to pizza when baking.

But those issues are only causing Google to slow down, not stop. A number of companies and experts that Tech in Asia spoke to say web-based companies around Asia are preparing for AI Overviews’ inevitable impact here.

One of the biggest questions for companies is the format. On the current version of AI Overviews – which is still being tested – most of the links on the search results page have been moved down to create space for Google’s AI-generated answer.

Some of the links are also being pushed off to the second page of the search results, which, in the competitive world of Internet search rankings, is where “forgotten content and websites go to die”.

The lack of sources itself could be a problem for Google, Rao said, adding that results the AI generates should include citations.

He thinks that the form factor will eventually evolve to “something which benefits not just Google, but also the various players that are feeding content” to it.

SEO will take a hit

Another big question: Who gets to provide the data for the AI Overviews’ answers?

Under the old system, search results were broken down into two categories: paid advertisements, which were usually identified as sponsored ads, and a list of sites chosen by Google’s algorithm.

Convincing the algorithm to place your webpage on the top of the search results is an act of wizardry known as search engine optimisation (SEO). SEO specialists crafted, monitored, and integrated keywords onto a webpage so that it could be easily found by “web crawlers” – small programs that would take the information they found back to the algorithm.

“There will be a massive impact on SEO strategy,” said Rajiv Dingra, founder and CEO of ReBid, an advertising consultancy that offers a customer data platform for marketing. “This will almost eliminate any organic search.”

Whether the old SEO strategies are dead or just evolving is still being discussed by marketers, according to SODP’s Arabian. He points to a recent survey that shows only 17 per cent of marketers, public relations professionals and content creators believe AI search will make SEO obsolete. He said 61 per cent believe SEO will instead transform to fit the new AI model.

Organic search is the traffic sent to a website from a search engine. High organic search traffic often indicates a successful SEO strategy.

Dingra said the lack of organic traffic will force advertisers to start relying more on paid traffic, which is generated by spending on advertising. But the limited amount of space on the new results page will likely result in increased competition and higher ad costs.

Less space = higher ad rates

Not everyone can afford even the existing advertising rates charged by social media companies. Theresa Lam, a PR consultant who started Gila Goodness a few years ago as a way to raise extra money, said her company is currently fighting just to maintain its web presence.

Gila Goodness offers a unique fusion of Western and South-east Asian food: nasi lemak, ondeh-ondeh, tom yum, and pineapple tart-flavoured granolas. The business has grown steadily, but while the company has been trying to make inroads in local stores, only about 3.5 per cent of her sales each month came from Google Search.

Increasing the company’s advertising costs would quickly eat away her profits, Lam said. She currently spends about RM100 (S$28.66) a month for SEO services on what she calls a bargain plan, which takes up about 5 per cent of her total monthly revenue.

Advertising on Facebook or Google just isn’t in her budget. Currently, 100 per cent of the traffic to her site is organic, she said.

She is concerned that anyone searching for granola in Malaysia would only see an AI Overviews-generated paragraph about the health aspect of the food. But if advertising on Google could give her the ability to target a relevant audience, she might consider it.

For big techs, it’s just business

Arabian of SODP and Andrew Thompson, the head of research and content for Melbourne-based Media Collateral, are working on a report of over 250 media professionals to gauge how the industry is handling these changes. Titled “The Gen AI x Comms Industry Impact Report”, the report is due out on Jul 30.

Some of the things that they note is that publishers are still leery of allowing big tech companies to dominate their distribution channels. Thompson said the lessons learned from previous dealings are still very relevant.

He warns that the real priority of big tech companies isn’t disseminating good information, but making profits.

“If you fly in the face of that, then all of a sudden your business models could just go kaput,” Thompson said.

But not everyone thinks that moving away from big tech is feasible, despite constant discussion in the industry to do so.

“There is always talk, but the alternatives are very few,” said ReBid founder Dingra. “The big four – Google, Amazon, Meta, Microsoft – have a distribution channel that cannot be replicated overnight.”

In terms of search, Google holds a 90 per cent market share as of June 2023, according to SimilarWeb.

Despite this, Dingra agrees that the goals of the big four are not the same as content producers, who need to be careful when dealing with those companies.

“Their [the big four’s] focus is their own business and the sanctification of their searches.”

The hazards of relying on social media and other tech companies is arguably best known in digital media, which, again, is expected to take some of the biggest hits from falling web traffic related to the introduction of AI search.

Daniel Marchette, CEO of Cambodia-based news site South-east Asia Globe (SEAG), see this as another in a long line of hard-learned lessons about relying on someone’s channels for distribution.

Marchette tells Tech in Asia that the publication saw drops in traffic following editorial changes at Facebook in 2017 and again when Twitter made similar changes in 2022. He said that neither is any longer a major source of traffic for the SEAG.

“The changes really just re-emphasize the conversation that’s been happening for a while, which is: We can’t rely on these third-party platforms to be the heart and soul of our engagement or our interaction with our users,” said Marchette.

He said that Google is still the main driver of traffic for SEAG, but the function it serves is to allow people to discover the website.

“Very few people are going to find our article on Google and say, ‘Oh, I need to subscribe today,’” he said. “What is important is now that the person’s on our website, how do we get them to somehow come back? How do we put them in the funnel, essentially?”

One of the ways they are doing that, he said, is by producing more in-depth stories that are meant to engage readers with news that isn’t found elsewhere.

Both Arabian and Thompson agree that long-form stories are probably the best way for media companies to convert readers into subscribers.

“AI search will reward longer-form content, more unique content, and content that comes from original expertise and experience,” Thompson said, although he warns that this will be more difficult for small publishers. But he notes it may be a necessity as Google is de-emphasizing rehashed content.

Content that is “curated and collated from different sources” and then repackaged will be “hard to push”, Thompson added.

Finding a niche a solution

While it may require work time and development, there are opportunities for niche content, in media, advertising, or even ecommerce.

The goal for many web-based businesses will be to find a niche that they can own – one that is not already flooded with content. If that niche holds people’s interest – if it’s compelling, Thompson said – then it may be able to carve out an area for itself with the AI.

Along with finding a niche, Arabian suggests media outlets develop two to three different monetisation models, such as newsletter or events, in addition to their editorial coverage.

“It’s about picking one vertical and then maybe covering three tangents off that,” he said. “But the reality is that we can’t over rely on one thing at all as a business, otherwise you’re going to just get caught. That’s, that’s the reality of it.” TECH IN ASIA