With over 20 years of experience, we transform your digital presence. We specialize in website and E-Shop development, SEO and Digital Marketing, ERP software and smart automation that take your business to the next level.
Address:
154 Egnatia Street, Thessaloniki 546 36
Phone:
+30 231 231 4095
E-mail:
hello@twodots.gr
Ready for the next step?;
AI marketing is evolving towards "agent-to-agent marketing", where AI agents take over the search and evaluation of products before the customer sees them. This requires e-commerce brands to improve their technical infrastructure and data quality to be readable and reliable by machines. Proper preparation can provide an advantage in visibility and conversions.
The article summarizes the most important points and turns them into practical steps for businesses that want better organic visibility, a cleaner user experience and more reliable content.
AI marketing is entering a new phase: from marketing to humans using AI tools, to marketing to AI agents who search, compare, filter and recommend options before the end customer even sees a product list. Ahrefs“ article on ”agent-to-agent marketing" and Moltbook serves as a particularly useful signal for e-commerce owners: the future of product discovery will not be limited to Google, marketplaces and social feeds. It will increasingly include environments where autonomous or semi-autonomous AI agents talk to each other, evaluate brands, read product data and make decisions based on the quality of information they can understand. See also: Digital Marketing & SEO, business automation & AI, website construction, e-shop construction.
For an online store, this doesn't mean that it has to abandon SEO, performance marketing or email automation. But it does mean that it has to evolve them. AI marketing is not just “writing content with AI”. It's about systematically preparing the brand, content, product feeds, structured data, reviews and technical infrastructure to be readable, trusted and preferred by engines acting as intermediate buyers. In this environment, trust is built not just with eye-catching visuals or clever slogans, but with clear data, consistency, documentation and machine-readable experiences.
The discussion is not theoretical. Gartner predicts that by 2028 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, while 15% of daily work decisions will be made autonomously by agentic AI. Meanwhile, McKinsey recorded that regular use of generative AI by organisations rose to 65% in 2024, almost double the 33% of the previous survey. For e-commerce owners, these numbers point to something practical: customers, marketing teams, search tools and comparison platforms will increasingly operate with AI agents. Those who prepare early will have an advantage in visibility and conversions.
Agent-to-agent marketing is the approach in which a brand communicates not only with humans, but also with AI agents that act as researchers, consultants, marketers or automated decision-makers. An agent may be tasked with finding the best pair of sneakers for a user with a specific budget, comparing material sustainability, return policy, delivery time, reviews, availability and final price. In a more mature scenario, the agent would not just present results. He or she will negotiate, exclude brands that don't have clean data, select products based on personalized preferences, and execute the purchase.
Moltbook, as presented by Ahrefs, is interesting because it shows what a social or information ecosystem might look like where agents have their own “social” behavior: they create profiles, exchange information, recognize credibility, and interact with content not like humans, but like machines seeking clarity and documentation. The bottom line for a brand is not Moltbook itself as a platform, but the shift it symbolizes: machines are no longer just distribution channels. They become active participants in the commercial decision.
In this context, AI marketing must combine four levels. First, classic brand positioning so that the store has a clear value proposition. Second, technical SEO and AI SEO so that the engines can read the information correctly. Third, generative engine optimization and LLM optimization so that content is more likely to be cited or utilized by answering engines. Fourth, clean commercial infrastructure: product feed optimization, structured data, shipping policies, returns, warranties and reviews in a format that can be audited.
As shown in the chart below, Gartner links the rise of agentic AI to two critical business changes by 2028: more applications will incorporate AI agents and a measurable percentage of daily decisions will be automated.
Gartner predictions for Agentic AI through 2028
Source: Gartner, 2024 - Agentic AI predictions
Until recently, product discovery was based on a relatively familiar sequence: a user searched on Google, entered a marketplace, saw an ad on social media or received an email. The brand was trying to show up at the right time with SEO, paid ads, social media and remarketing. Today, the chain is becoming more complex. A user can ask an AI assistant “find me the best laptop for remote work up to 900 euros with a good battery and reliable support in Greece”. The assistant will not necessarily display ten blue links. It will synthesize an answer, compare options and perhaps suggest two or three products.
This is where generative engine optimization appears, i.e. optimizing content so that it can be used by generative engines and LLMs. It does not replace SEO. It extends it. An e-commerce site still needs fast pages, clean architecture, proper categories, internal linking and quality content. But now it also needs answers that are specific, documented, comparable and easy to extract. For example, a product page that says “high quality and modern design” gives little value to an agent. A page that includes materials, provenance, measurements, certifications, instructions for use, warranty, compatibility and actual reviews gives many more brands.
AI search optimization requires that we think about the information as an analyst would read it. If an AI agent has to compare ten brands, he or she will prefer the one that provides clean data: price, availability, shipping costs, return policy, features, reviews, FAQs, schema markup, and consistent description across channels. This changes content marketing as well. “10 reasons to buy” articles are not enough if they don't contain real experience, comparisons, data and answers to objections. E-E-A-T becomes even more important because LLMs and agents need evidence of experience, expertise, authority and credibility.
The transition does not happen in a vacuum. McKinsey shows that organizations are adopting generative AI at a rapid pace, which increases the likelihood of purchasing, marketing and customer service teams using AI agents in their day-to-day operations.
Regular use of Generative AI by organizations
Source: McKinsey Global Survey, 2023-2024
The first practical change is to move from “pages that convince people” to “pages that convince people and machines”. This does not mean cold or colourless content. It means accurate content. If you sell cosmetics, an AI agent needs to be able to understand ingredients, skin type, contraindications, certifications, country of manufacture, cruelty-free items, FAQs and real results. If you sell electronics, it must be able to compare technical specifications without ambiguity. If you sell fashion, it must have access to size guides, fabric, fit, care instructions and returns data.
The second change concerns structured data. The schema markup for Product, Offer, AggregateRating, Review, FAQPage, Organization and BreadcrumbList is not just an “SEO detail”. It's the language in which engines understand what's on the page. The more critical data that is left hidden in images, tabs that don't load properly or vague descriptions, the harder it becomes for an agent to trust the store. Consistency between the site, merchant center, marketplaces and social shops also becomes critical. If the price, availability or description differs by channel, the system may consider the information less trustworthy.
The third change is product feed optimization. In traditional performance marketing, a feed influences Google Shopping, Meta catalogs and marketplaces. In AI marketing, the feed becomes even more central because it contains the commercial data that can be used by recommendation engines, shopping assistants and future machine customers. A feed with incomplete titles, poor categories, wrong attributes and generic descriptions loses value. Conversely, a feed with detailed attributes, clean identifiers, GTIN where available, correct images, availability and consistent taxonomy becomes an asset.
The fourth change is the proof of trust. AI agents don't have an emotional reaction to a nice banner, but can incorporate ratings, policies, delivery times, return rates and third-party references into their proposal. This means that reviews, responses to negative reviews, return policy, “About Us” page, certifications and customer support content should be easily accessible. Conversational commerce will increasingly rely on such signals because the agent will be trying to reduce risk on behalf of the user.
The need to reduce risk is clearly seen in the classic e-commerce funnel. According to the Baymard Institute, the average recorded cart abandonment rate is 70.19%. If people are abandoning so often due to friction, uncertainty or cost, AI agents will become even more stringent in filtering out stores that don't provide clean answers before checkout.
Average cart abandonment rate
Source: Baymard Institute, Cart Abandonment Rate
The first step is to map out the decisions a customer needs to make before they buy. Don't start with channels, start with questions. For each key product category, list what a serious buyer would ask: which product is right for me, why should I choose it, what's the difference between it and the cheapest, what happens if it doesn't fit, when is it delivered, what's the warranty, are there hidden costs, what do other customers say? These questions are the material that AI agents will also use to evaluate your proposal.
The second step is an audit of the product pages. Select the 20 products with the highest commercial value and check if each page includes a full title, unique description, technical features, benefits, advantages, limitations, instructions, FAQs, reviews, shipping and returns policy, and structured data. If something is only in an image, convert it to HTML text as well. If something is unclear, make it measurable. If you write “fast delivery”, replace it with “delivery in 1-3 business days for available products”, if applicable.
The third step is to improve the feed. Check titles, categories, attributes, GTIN, MPN, brand, color, size, material, gender, age group, availability, price, sale price and images. For each product, the title should be descriptive without being spammy. A good title for agent-to-agent marketing not only chases clicks, but allows for accurate classification. For example, “Waterproof Men's 10K Hooded Jacket - Black” is more useful than “Super Offer Jacket”.
The fourth step is creating comparison content. AI agents love comparisons because they reduce uncertainty. Create “how to choose” guides, comparison tables, articles by use case, and pages explaining which options fit different needs. If you sell office equipment, write a guide for “best office chair for 8 hours of work”, but back it up with criteria: ergonomics, settings, hardware, warranty, user weight, dimensions and after-sales support. This is simultaneously content marketing, LLM optimization and market experience.
The fifth step is to strengthen E-E-A-T. Add authorship where there is advisory content, mention experts, explain the methodology of your proposals, show company details, support, physical presence if available, ways to contact and actual policies. AI marketing does not reward the vague. It rewards the demonstrable. If an article suggests products, explain why. If a product has limitations, mention them. Honesty is a commercial advantage when agents are trying to protect the user from making the wrong choices.
The sixth step is measurement. Create a dashboard with organic traffic, impressions in rich results, conversion rate per landing page, cart abandonment, in-site searches, customer support inquiries and product feed performance. Also track whether your brand appears in AI search tool responses for important queries, not as an absolute KPI, but as an indication of visibility. AI SEO and generative engine optimization are still in development, so systematic testing is needed, not hasty conclusions.
An AI agent acting as a shopping assistant is not “charmed” in the same way that a human is charmed. It will process signals. The most critical ones are clarity of the offer, reliability of information, proof of quality, availability, overall price and post-purchase experience. For this reason, e-commerce brands need to see their content as a trust database. Any ambiguity is a potential barrier. Any well-structured information is a potential asset.
In practice, an agent could reject a product because they don't find a clear returns policy, because the reviews are few or unclear, because the feed doesn't mention critical attributes, because the page loads slowly or because there are conflicting prices. This makes marketing automation more challenging. It's not enough to send a personalized email. You need the entire ecosystem to be consistent: site, CRM, feeds, ads, customer support, reviews and content.
A useful prioritization framework for e-commerce owners is this: first fix what hinders understanding, then what hinders trust, and finally what hinders persuasion. Understanding is about product data and technical accessibility. Trust is about reviews, policies, security and brand credibility. Persuasion is about positioning, offers, bundles, storytelling and ecommerce personalization. If you reverse the order, you risk investing in nice campaigns for products that the engines can't properly evaluate.
The graph below shows no arbitrary scoring of factors. It organizes, as a practical checklist, the fields that should be checked in an e-commerce audit for agent-ready experience, without assigning false weighting percentages. For this reason it is not used as a statistical chart, but as a functional illustration of the key audit areas resulting from structured data, product feeds and AI search requirements.
To remain consistent with the actual data available, the graphs in the article are limited to metrics from Gartner, McKinsey and Baymard. The operational checklist should be implemented as an internal audit and scored with real data from your own store.
The key message from the discussion around Moltbook and agent-to-agent marketing is that the next big change in digital commerce will not just be a new advertising channel. It will be a new mediation logic. Between the brand and the customer there will be more and more systems that analyze, compare and decide. These systems will favour brands that have clear information, technical consistency, real credibility and content that answers specific needs.
For an e-commerce owner, AI marketing should be treated as a growth infrastructure. It starts with SEO, but it doesn't stop there. It includes AI search optimization, generative engine optimization, structured data, product feed optimization, reliable customer experience and content that can be used by both humans and AI agents. The earlier this infrastructure is organized, the easier it will be for the brand to adapt to shopping assistants, conversational commerce, machine customers and new search environments.
The practical advice is simple: don't wait until the agents are fully mature to fix the foundation. Start with the highest value products, clean data, enrich pages, implement schema, organize feeds and build content with real experience. AI marketing won't necessarily reward the biggest brand. It will reward the brand that can be understood, proven trustworthy and safely recommended by an agent at the moment the customer is ready to buy.
Sources
Do you want an e-shop that sells?;
We set up e-shop fast, secure and ready for online sales, with proper tracking and SEO basis.
Agent-to-agent marketing is the approach where brands communicate not only with humans, but also with AI agents that act as researchers or buying assistants. The agents search, compare and recommend products based on the quality of information.
AI marketing requires online stores to tailor their content to be understood by AI agents. This includes optimizing product feeds, structured data and providing clear and reliable information.
GEO is optimizing content to be useful for generative engines and LLMs. It extends traditional SEO by requiring specific, documented and comparable answers.
An e-commerce site must offer accurate and detailed content, improve structured data and optimise product feeds. It must also focus on the reliability and clarity of information.
AI agents process signals such as clarity of offer, reliability of information and overall post-purchase experience. Clear and accurate information increases the likelihood of choosing a product.
Structured data allows engines to understand the content of a page. It is critical to the reliability and correct presentation of information to AI agents, directly affecting the visibility and performance of an e-commerce site.
The key elements include AI search optimization, generative engine optimization, information reliability and providing content that can be used by humans and AI agents. These help to achieve better visibility and conversions.
