AI Search Ranking Factors Since 2024: What Changed and What Didn't

Since 2024, AI assistants have moved from answering mostly from training data to also searching the live web and showing sources. That shift made traditional signals like crawlability, clear pages and trusted mentions more important, not less. What changed is how answers are assembled and how brands are shown inside them.

The big shift: from memory to live search

Early chatbot answers came mainly from what a model learned during training. If your brand was not well known before the training cut-off, the model simply did not know you. During 2024, that changed. Google rolled out AI Overviews in its search results, ChatGPT added web search, and Perplexity grew as a search tool built around cited sources. Copilot continued to draw on Bing.

For businesses, this means a new page or a fresh review can now influence an answer within days or weeks, instead of waiting for the next model to be trained. It also means that if your site cannot be crawled or is slow to load, you miss out on the live part of the answer.

What changed for brands

Citations became visible

Many assistants now show links next to their answers. Being the cited source is a new kind of visibility, separate from being named. A brand can be mentioned without a link, or linked without being recommended. Tracking both became necessary.

Comparison answers became common

Users increasingly ask for shortlists: “three best project tools for a design studio” or “Shopify or WooCommerce for a small wine shop”. Assistants answer by pulling from review sites, comparison articles and vendor pages. Brands that publish honest comparison content and appear on independent lists have more chances to be included.

Conversations replaced single searches

A buyer may ask a follow-up like “which of those has a free plan?” The brands that stay in the answer are those whose facts (pricing, features, locations) are clear and consistent across sources.

What stayed the same

• Crawlable, readable pages. Text in HTML that bots can fetch is still the starting point.
• Authority from others. Reviews, press coverage and directory listings still act as votes of confidence.
• Relevance to the question. A page that directly answers a specific question still beats a vague page.
• Accurate business data. Consistent name, address and phone details still matter, especially for local queries.

If you want a side-by-side view of where classic SEO and AI search differ, read AI search vs traditional SEO.

Habits to drop

Some tactics that were already weak became even less useful. Keyword stuffing does not help an assistant summarise you. Thin pages created for every keyword variation add noise. Paid link schemes and fake reviews carry real risk, because review platforms and search engines actively remove them, and they can damage trust in your brand.

Habits to build now

1. Write a short, factual answer at the top of each key page.
2. Keep one source of truth for facts like prices, hours and service areas, and update other listings from it.
3. Earn mentions on the sites your industry trusts, through useful data, expert comments or real partnerships.
4. Check how assistants answer your top questions every month, not once a year.

How Keboost helps

Keboost was built for this new kind of search. It tracks the prompts your buyers ask across ChatGPT, Gemini, Claude, Perplexity and Copilot, shows which brands and sources appear, and creates content designed to be found and cited. It cannot promise a ranking, but it shows what is happening and what to fix. Get a free AI visibility check, or see plans from €29 per month.