“AI-powered” has become the default label on SEO proposals. Almost every agency and software tool now claims to use artificial intelligence, and some of them do it well. Others have added a chatbot to an old process and raised their prices. For a marketing manager comparing proposals, the label alone tells you very little.
A few questions go a long way in separating useful AI SEO services from expensive marketing copy.
Start with what AI is actually good at
AI tools are genuinely useful in a few parts of SEO work:
- Grouping keywords. Sorting thousands of search terms into topics and intent is slow by hand and fast with a model.
- Spotting technical issues. Crawling a large site and summarising patterns in errors, redirects or duplicate pages.
- Outlining. Producing a first structure for a page that a human writer then checks and rewrites.
- Comparing competitors’ content. Looking at what ranking pages cover and where a site has gaps.
Notice what is missing from that list. AI does not know your customers, your margins or what your sales team hears on calls. It cannot verify facts on its own, and it does not decide strategy. A good vendor uses AI to speed up the analysis and keeps people responsible for judgement and accuracy.
Questions worth asking
Which parts of the work use AI, and which are done by people? A clear answer is a good sign. A vague answer (“our AI handles everything”) is not.
Who checks the content before it is published? Search engines do not penalise content for being written with AI help, but they do reward pages that are accurate and useful. Unchecked AI text often contains confident mistakes.
How will you report results? Rankings and traffic are useful signals, but ask how the vendor connects the work to enquiries or sales.
What happens to our data? If you share analytics, customer lists or internal documents, ask which tools process them and where.
Do you also optimize for AI search? Search is changing. Google’s AI Overviews, ChatGPT and other assistants now answer many questions directly. Some agencies treat this as a separate discipline, often called generative engine optimisation, which focuses on being cited in those answers.
Red flags
Be careful with proposals that:
- guarantee rankings or a fixed percentage of traffic growth
- promise that AI will “predict” algorithm updates before they happen
- publish large volumes of automated content without editing
- cannot explain their methods in plain language
None of these are signs of advanced technology. They are signs of a sales pitch.
Local context still matters
For companies in markets like Thailand, where people search in both Thai and English, AI tools are only as good as the language data behind them. Many tools are strongest in English and weaker in Thai. Bangkok agencies such as Relevant Audience combine AI tooling with native-language review for this reason. Wherever you look, ask to see examples of the vendor’s non-English work.
What to judge them on
AI makes good SEO teams faster. It does not turn weak strategy into strong results. Judge vendors on their process, their honesty about what can and cannot be promised, and how clearly they connect the work to your business goals. The word “AI” on the proposal should be the least important part of the decision.
