Can ChatGPT Replace an SEO Agency? A Look at Two Real Projects
Is ChatGPT enough to handle SEO without an agency? Some business owners are already trying exactly that.
The idea no longer seems far-fetched. ChatGPT can look over a landing page, suggest subjects for the blog, write metadata and point out possible technical problems. It can even produce a fairly detailed action plan. Compared with the cost of hiring an SEO team, the subscription is cheap.
But SEO is not judged by the number of recommendations produced. The real test comes later: which pages gained visibility, what kind of traffic they attracted and whether that traffic created enquiries or sales.
I had an opportunity to examine this on two live projects. Both websites had previously been handled by our SEO team. After our work ended, their owners continued independently and told us that they were using ChatGPT for SEO advice and content planning.
We cannot see their conversations with the platform. It would therefore be dishonest to claim that ChatGPT chose a particular topic, changed a specific page or set the priorities for either website. What we can examine is the outcome: how the sites developed, where their organic traffic now lands and how much of that performance still comes from work completed during the agency period.
Both websites continued to grow after leaving the agency.
That might seem to settle the argument. If traffic increased, perhaps ChatGPT did replace the SEO company. Once the traffic is broken down by page type and search intent, however, the picture becomes less straightforward.
Case One: Using ChatGPT to Promote an Online Flower Shop
The first project is an online flower shop serving customers in Los Angeles. We worked with the site for about two months.
Our work began with keyword research and audits of the technical setup and existing content. We then wrote the metadata, prepared briefs for new articles and marked the pages that needed rewriting. Backlink work had also started by the time the project left our team.
The client later moved the site to another domain. From that point on, the owner handled SEO independently and used ChatGPT for at least part of the work.
The headline result looks impressive. Ahrefs shows that the website grew from almost no organic visibility to ranking for thousands of queries and attracting tens of thousands of estimated monthly visits.
On that graph alone, there would be an obvious question: why pay an SEO agency if a business owner working with ChatGPT can achieve this kind of growth?
The answer appears when we look at the pages generating the traffic.
Around 86% of the website’s estimated organic traffic now goes to blog articles. The homepage accounts for a further 11.5%. The pages directly involved in selling flowers receive very little:
- Category pages attract approximately 0.4% of the traffic.
- Individual bouquet pages account for roughly 0.2%.
Together, product and category pages receive less than 1%.
For a publisher, that distribution could be a success. A flower shop has a different job. It needs people in and around Los Angeles to find a suitable bouquet, choose a delivery option and place an order. Traffic to articles about birth-month flowers, flower meanings or tattoo ideas may increase reach, but it does not automatically bring the website closer to a sale.
There is another part of the result that matters.
Before our work ended, the team had prepared nine new articles. They were first published on the old domain and later transferred by the client to the replacement website. Those nine pages now generate about 3,900 estimated organic visits per month, close to 20% of the site’s total organic traffic. They include articles about May and August flowers, flower tattoos and different flower varieties.
In other words, the independent campaign did not begin with an empty website. The keyword research had been completed, the site had already been audited, metadata and content had been prepared, and backlink work was underway. Several pages produced during that initial period remain among the site’s strongest traffic sources.
ChatGPT and content were not the only things changing during this period. The site’s referring domains rose from about 30 to roughly 750. An increase of that size almost certainly contributed to the improvement in rankings, although we cannot separate its effect from the rest of the work.
The project has clearly grown. The limitation is not a lack of traffic but the type of growth achieved. Most of the visibility belongs to informational content, while the pages responsible for generating revenue remain almost absent from organic search.
From a distance, the Ahrefs graph suggests a highly successful DIY SEO campaign. At page level, it shows an ecommerce website attracting readers far more effectively than buyers.
Case Two: Using ChatGPT for a Psychotherapy Practice Website
The second website belongs to a psychotherapy practice in the United States. It offers individual and couples therapy to clients in Florida, including online appointments.
When we took over the project, Ahrefs estimated its organic traffic at about 15 visits per month. By the end of the active work, that figure had reached 73. During the period managed by our team, estimated traffic therefore increased almost 5x.
The client then continued working on the website without us and began using ChatGPT for SEO advice. Traffic did not stop growing. It reached approximately 194 visits in June 2026 and currently sits at around 160.
At first glance, the independent approach appears to be working. The website now attracts more organic traffic than it did when our involvement ended. Again, though, the total does not show where those visits come from.
Four articles prepared by our team currently bring in an estimated 77 visits per month:
- An article about daughters of narcissistic parents: 29 visits
- A guide to open relationships: 25 visits
- An article about signs of a narcissistic husband: 15 visits
- A piece about rumination: 8 visits
Together, these four pages account for about 46% of the website’s present organic traffic.
The homepage attracts another 76 estimated visits. That leaves more than 90% of the site’s visibility concentrated on just five URLs: the homepage and four articles written during the agency period.
Even the homepage traffic needs some context. According to the current Ahrefs data, its leading query is the informational phrase “open relationship means,” with most of the estimated visits for that query coming from the Philippines. For a psychotherapy practice serving clients in Florida, this is unlikely to be a major source of relevant enquiries.
The service pages tell a different story. Pages covering family therapy, recovery from narcissistic abuse, online therapy in Florida and anxiety treatment receive little or no estimated organic traffic. Among the commercial pages, only two therapist profiles show any noticeable activity, with five visits to one and three to the other.
There has been one useful result from the client’s independent work. A newer article about online therapy in Fort Lauderdale receives approximately six visits per month. The subject is local and closely connected to the practice’s services, which makes this traffic more relevant than visits from a general informational query. At the moment, however, the page is too small to change the overall traffic distribution.
We should be careful about attributing these results directly to ChatGPT. We do not know which recommendations the client received or which of them were implemented. The only thing we can assess is the website after the agency period.
What we see is similar to the flower shop. Organic traffic has increased, but much of the current visibility still comes from content created by the SEO team. The newer work has produced some movement, yet the service pages that should generate therapy enquiries remain weak.
The graph shows a growing website. From a business perspective, the unresolved issue is that very little of this growth reaches the pages where a prospective client can choose a service and book an appointment.
How an SEO Agency’s Strategy Differs from ChatGPT Advice
The difference is not that an SEO agency can write articles while ChatGPT cannot. Content is one of the areas where AI is already useful. SEO professionals use it too, particularly for research, early drafts and sorting large amounts of information.
The difference is who directs the work.
An SEO team starts with the commercial purpose of the website. A flower shop in Los Angeles needs local orders. A psychotherapy practice in Florida needs enquiries from people who live in the area and are looking for the services it provides. If traffic rises while the pages closest to a purchase or booking remain invisible, the higher total can be misleading.
That is why the first decisions usually concern priorities rather than content volume. The team needs to establish:
- Which pages could realistically bring in sales or enquiries?
- What searches should bring potential customers to those pages?
- Why are the pages not ranking now?
- Which changes deserve priority?
- How should blog traffic lead readers towards a product or service?
- Which results will show that the work is helping the business?
ChatGPT works from the request it receives. Ask it for a content plan and it will return article ideas. Ask how to build backlinks and it will suggest methods and possible sources. Give it a service page and it may rewrite the structure, headings and metadata.
Any one of those answers might be useful. The harder part is deciding how the pieces fit together.
Suppose ChatGPT produces a list of twenty improvements. Should the company begin with ten new articles or repair its category pages? Does it need another landing page, or would that page compete with an existing URL? Is more informational traffic worth pursuing before visitors have a clear route towards booking or purchasing?
AI can respond to those questions as well, but someone needs to recognise that they should be asked.
It also needs the right project data. A useful recommendation may depend on information from Google Search Console, analytics platforms and backlink tools, together with the site’s history, target locations, budget and work already completed. Without that context, ChatGPT is filling in gaps rather than analysing the actual project.
Receiving the recommendations is only the beginning. They still need to be reviewed and converted into instructions that a writer or developer can follow. After publication or implementation, somebody must check the site again and decide whether the change worked.
In practice, this leaves the owner doing the project management. ChatGPT may help with a page, report or individual decision, but it does not take responsibility for how those separate actions affect the rest of the website.
A managed SEO process already has that structure:
- Review the website, market and competitors.
- Identify the most valuable problems and opportunities.
- Prepare tasks covering content, technical SEO and backlinks.
- Check the implementation.
- Measure the result and adjust the next stage of work.
ChatGPT can support almost every step on this list. What it does not do by itself is supervise the whole cycle.
The two projects above make that distinction visible. Both owners kept working on their websites and both achieved further growth. In each case, however, the easiest area to expand was informational content. The commercial pages remained much weaker.
That does not prove that ChatGPT told either client to ignore products, categories or services. We have not seen those conversations and cannot know what advice was given.
It does show that several individually sensible actions can still take a website in the wrong direction. SEO strategy is not simply a collection of tasks. Timing matters, as does the effect one change has on the next. A plan that made sense three months ago may need to be dropped when rankings, traffic or enquiries tell a different story.
ChatGPT can remain part of that process from beginning to end. At present, though, a person still has to read the results, notice when the plan is drifting and decide what should happen next.
Could the Websites Have Grown Faster with an SEO Team?
There is no way to calculate the alternative result precisely. The same website cannot be promoted independently and by an agency at the same time under identical conditions.
Still, the current traffic distribution tells us where time was lost. In both cases, commercial pages remained weak while informational content continued to expand.
In my view, the websites could have developed at least twice as quickly if the systematic SEO work had continued. This does not mean that their Ahrefs traffic graphs would necessarily show twice as many visits. One broad informational article can attract more traffic than several service or category pages combined. Those smaller commercial pages may nevertheless be far more valuable to the business.
SEO work normally develops through repeated short cycles:
- Collect current data.
- Find a problem or an opportunity.
- Make a change.
- Check the result.
- Adjust the next set of tasks.
The length of this cycle matters. When a business owner chooses the wrong priority, it may remain in the plan for several months. General traffic can continue rising during that period, making the decision look successful even if the website is attracting few potential customers.
A team reviews the results regularly. If new articles are gaining visibility but product or service pages remain static, the plan can be changed during the next working period. Content, technical improvements, backlink work and analysis can also move forward at the same time rather than waiting for one person to finish each task in turn.
This is where much of the possible difference in speed comes from. An agency does not necessarily type faster or publish more articles. It is more likely to notice earlier that effort is being spent in the wrong place.
The two independent campaigns were not failures. Both websites grew. The same period could simply have been used more effectively, with commercial visibility treated as a priority before the imbalance became so pronounced.
Why ChatGPT SEO Mistakes Can Be Expensive
ChatGPT is improving quickly. Work that once required hours of manual research can now be completed in minutes, and its SEO capabilities will almost certainly continue to expand.
The problem is that an incorrect answer can sound just as convincing as a correct one.
With content, the damage is often limited to a single page. An article may include an incorrect date, refer to a study that does not exist or make a claim that cannot be supported. These errors weaken trust, but they can usually be found and corrected without affecting the rest of the website.
The risk is higher in sensitive sectors. A psychotherapy website, for example, cannot publish unchecked claims about diagnoses, treatment methods, professional qualifications or likely outcomes. A factual error in this setting is not only an SEO issue. It can also damage the reputation of the practice.
Technical recommendations can create much wider problems.
Advice to edit robots.txt, apply a noindex directive, change canonical tags, merge pages or introduce redirects may sound reasonable when read in isolation. Whether it is safe depends on the structure and history of the particular website.
A poor sentence affects one page. A poor technical change can affect thousands.
Domain migrations and URL restructuring are especially unforgiving. Missing redirects, conflicting canonical tags or an important section accidentally blocked from crawling can cause a sharp loss of search visibility. Search engines may temporarily lose access to a large part of the website, and recovery can take months.
The effect is not always immediate. Traffic may begin falling weeks after the change, by which point the connection between the implementation and the decline is harder to trace.
SEO specialists can make mistakes too. The difference lies in the checks around the decision. Before a major change, a technical SEO and GEO audit examines the current configuration, indexing signals, analytics data and dependencies between pages. After implementation, crawling, rankings and traffic are monitored to catch unexpected effects.
ChatGPT cannot take responsibility for that process. It may describe the risks, prepare a checklist or help diagnose a problem, but a person still decides whether the recommendation is appropriate and whether it should go live.
The greater the potential site-wide effect, the less suitable the task is for unreviewed AI advice. Rewriting a title takes minutes to reverse. Repairing a failed migration or an indexing mistake may take an entire season.
So, Can ChatGPT Replace an SEO Agency?
The two projects show that a website owner can use ChatGPT to continue SEO work and achieve further growth. Denying that would ignore what happened on both sites.
Growth alone, however, does not prove that the agency was fully replaced. Neither owner started from zero. Keyword research, audits, metadata, content and initial backlink work had already created a foundation. A considerable share of the current traffic still comes from pages prepared by the SEO team.
The results are also concentrated in a small number of largely informational URLs. The flower shop attracts readers while its category and product pages receive less than 1% of estimated traffic. On the psychotherapy website, the homepage and four agency-written articles account for more than 90% of current visibility, while most service pages receive almost nothing.
ChatGPT can replace some SEO tasks. It can speed up research, organise data, suggest content and produce useful first drafts. For an owner who understands SEO and has time to manage the work, it can reduce both cost and routine.
What it does not replace is ownership of the strategy. Someone must still decide which pages matter, choose between competing priorities, provide the right data, check technical changes and recognise when traffic is growing without helping the business.
That is the part an SEO team provides. Its value is not access to secret information unavailable to ChatGPT. It is the ability to choose an appropriate action for a particular website, follow the result and change direction before months of work are spent on the wrong target.
AI tools will become better at retaining project history, analysing live data and coordinating longer sequences of tasks. For now, they still need an experienced person to set the context and check the decisions.
ChatGPT is already a useful SEO tool. It is not yet an SEO department.
If your website is growing with the help of ChatGPT but you are unsure whether that growth is producing enquiries or sales, book a consultation. We can review which pages are driving the result, where effort is being lost and what deserves attention next.
Good luck with your SEO!