Google, ChatGPT, Perplexity force agencies to rethink websites

Ad agencies have begun redesigning their websites to improve their chances of being discovered by potential clients using AI-powered search platforms such as ChatGPT and Perplexity. Creative agencies with flashy homepages could risk being de-emphasized by AI search tools given the systems’ preference for simple, text-based information, experts told Ad Age. Now, agencies such as Austin, Texas-based Diko are exploring ways to reconfigure their image- and video-heavy sites, including creating a separate, text-only layer that AI can more easily crawl. Brands are now using AI search tools to help them prepare requests for proposals, said Heather Physioc, chief discoverability officer for VML. Agencies, from VML and Diko to Gus and others, are having to respond by highlighting particular areas of expertise, such as customer experience and growth marketing, in order to show up in prospective clients’ AI-powered search results. “We are testing and learning where the AI search engines source information, how they present it, and what factors influence our ability to show up,” Physioc said. The growth of AI search has prompted agencies, as well as brands, to rethink their search strategies. AI-powered search engine Perplexity is serving more than 230 million queries a month, according to the startup. ChatGPT’s search tool serves nearly 40 million queries a day, according to SparkToro. Google, meanwhile, is delivering AI-generated results in overviews on its main search page, and AI Overview now reaches 1.5 billion users, according to the search giant. To increase their chances of showing up in consumer searches, brands are learning how they are portrayed by different AI tools and making adjustments to improve their visibility. Agencies are now doing the same to cater to prospective clients. VML has updated its website to reflect authority in certain areas of work, such as customer experience, in order to align with results returned by AI search, Physioc said. The agency has essentially conducted simulations in which it used an AI search platform as a prospective client would, asking questions about important areas of work and the top agencies that offered them. Agencies P3 Media and NoGood have leveraged a platform called Goodie to assess how different AI models surface their companies in search results, and then make adjustments based on Goodie’s recommendations. P3, for instance, determined that a lack of academic citations in its blog posts and whitepapers was hindering its sense of credibility. By increasing the frequency with which it cited expert material, the agency has incrementally grown AI-search traffic to its website over the past three months, according to co-founder David Wagoner. Data-driven Marketing Playbook: Join industry leaders on May 22 to explore how data can drive marketing strategies, optimize media buying and improve customer engagement. Learn more here. Agencies rethink flashy websites to stay competitive As agencies tweak their online presences, some are finding that their own creativity is hurting their chances of being discovered by AI search. Creative and production-focused agencies tend to use their websites to show off their work with clients, often opting for flashy arrays of images and videos over descriptive paragraphs in order to sell the mood of the campaigns. The problem is that AI search relies on text-based information more than anything visual, potentially hurting the visibility of these sites, VML’s Physioc said. Indeed, creative agencies are starting to grapple with this reality. Diko, the agency based in Texas, hosts a website that is full of zany and over-stimulating content—including a fly as a cursor—that is meant to give clients an idea of what the shop is all about. “We wanted it to be emotional—to stir up excitement and confusion,” said founder and creative director Diko Daghlian. “But these are very non-machine things.” Diko has recently held discussions on how to make its website more amenable to AI search tools. The agency is considering translating the website into a text file format and implementing that as a layer so AI tools can read its information far more easily. In this scenario, the creative-heavy layer of the website would still be accessible to non-AI visitors. Also read: 5 tips for crafting the ideal social presence Many organizations that have concerns over their website’s crawlability are starting to consider implementing a text-based layer readable by a large language model (LLM), 829’s Giblin said. The proposed protocol that would provide the reformatting, “llms.text,” is already available via an application programming interface (API), which companies can integrate directly into their sites. But outside of Anthropic’s models, LLMs do not yet fully support the protocol, according to Giblin. Design agency Gus, based in New York, has a website that could be unwieldy to an AI search tool, but the shop has attempted to counter this by including loads of text in the creative layer itself. “We’re not shy about telling a story with words on our website, and that helps the bots find Gus better than anything,” said Graham Douglas, Gus’ co-founder and creative director. As AI systems advance, search tools are improving in their abilities to crawl visual information when sourcing. But for now, text remains critical to how LLMs consume information, and agencies can easily help themselves by providing supplementary text for whatever visual media they have on their sites, such as up-to-date file names and descriptive alt-text, said VML’s Physioc. And the more accessible that agencies make this textual support, the better. “If it’s 20 layers deep, that tells AI engines that the information is not that important,” Physioc said. Agencies turn to answer engine optimization to improve AI search rankings P3 Media knew AI search was worth looking into when some of its own clients mentioned that they used ChatGPT to learn more about the e-commerce agency, said Wagoner, who declined to share the clients’ names. “Six months ago, there was literally zero mention of P3 in the context of LLMs,” Wagoner said. “Now it’s trickling into the conversation.” P3 wants to understand the kinds of questions prospective clients are asking AI search platforms to increase its visibility in search results, and to put the agency in the best light when prospective clients ask the large language models specifically about its offerings. For the past three months, the agency has worked with Goodie, a platform specializing in AEO, or “answer engine optimization.” AEO, like search engine optimization (SEO), aims to improve a company’s online discoverability, but with a focus on catering to AI bots doing searches on behalf of human users. This distinction poses many new considerations for organizations looking to influence how their brands are surfaced. Indeed, roughly 30% of factors influencing AEO may have had little impact on SEO and could still be uncovered, said Jack Giblin, senior search engine optimization manager for digital agency 829 Studios. This “is where the real value is going to be,” Giblin said. For example, each AI system—from ChatGPT to Gemini to Anthropic’s Claude—views brands differently as authoritative sources of information, and therefore, brands are having to optimize their web presence in the face of potentially competing search platforms. AI search engines also have predictable ways of deeming content as accurate and inaccurate. They mostly favor that which they consider to be accurate, so brands must seek to construct their content as such. This sense of credibility was largely lacking in P3’s online presence, Goodie found. But the platform recommended as a solution that the agency include more academic research and citations in the work that it publishes on its website, mostly whitepapers and blog posts. P3 took the advice. “If we make our own analysis, that’s one thing,” Wagoner said. “But if we cite a meta-analysis from Harvard Business Review, that’s a whole different thing.” Goodie works by first conducting a thorough examination of how a company is surfaced in AI search results, said Mostafa ElBermawy, founder and chief executive of Goodie. This phase hinges on asking various large language models (LLMs) thousands of queries that prospective clients might ask, such as “Give me a list of the top five growth marketing agencies that work with startups.” Goodie then identifies useful patterns such as internet sources that are most frequently mentioning the agency, factors negatively affecting the agency’s visibility and opportunities to improve. Growth marketing agency NoGood also used Goodie to improve its visibility in AI-powered search (ElBermawy founded both NoGood and Goodie, but said the two organizations operate independently.) The platform discovered a substantial amount of outdated information about NoGood on the internet, such as the size of its team and the products it offered. Based on this insight, NoGood was able to reach out to these sources—which were mainly industry thought leaders and outlets publishing listicles—and update the information. Goodie is one of just a handful of companies now offering AEO services, which includes marketing agencies Marcel Digital and Elatre, and automation agency Ace Workflow. Ace Workflow itself has benefited from AEO-related efforts; last month, the company fielded a call from a prospective healthcare client that said they had discovered the agency via a recommendation from ChatGPT. And just last week, another prospective client reached out to Ace Workflow after asking ChatGPT for a list of agencies that had experience in form application Fillout and software services Zapier and Hubspot. “There’s currently a significant rush for companies to revamp their content strategies to ensure their content is more easily understood by LLMs,” said Tim Rodgers, founder of Ace Workflow.

How AI search is forcing agencies to rethink their websites to attract brands

Featured Stories