AI can solve your SEO problems easier. It has also made it easier to solve the wrong problems.
We can research markets, interrogate customer data and even produce project plans in a fraction of the time some of these tasks used to take.
But faster does not necessarily mean more effective.
AI systems can produce inconsistent outputs, while the issues we run into in digital work do not usually have a single cause. A visibility problem may originate in the CMS rather than the SEO strategy, with content teams depending on an under-resourced development pipeline. A drop in conversion may reflect a mismatch between proposition and customer expectations. A delayed project may be the result of competing priorities in different departments rather than a flawed project plan.
Yet we naturally focus on the most visible issue. We fix a broken URL, write a plan to increase conversion or chase a delayed implementation. The immediate problem disappears there and then, and everyone moves on - until the same problem, or a version of it, happens again.
There is a paradox here: we can become very good at solving problems while neglecting the system that keeps producing them.
AI can amplify this paradox.
We produce more data analysis, recommendations and outputs without necessarily becoming better at deciding where to address recurring issues.
This is where systems thinking is useful, and seasoned SEOs may already be doing some of it without naming it.
What systems thinking actually is
Systems thinking may sound more theoretical than it really needs to.
It is a way of understanding problems by considering the wider environment they sit in and how the parts connect. It considers patterns, relationships, dependencies and cause and effect loops, instead of components in isolation.
I like to think of this as fixing a hole in a bucket, only for new holes to appear each time you fix one. Continuing to fix the individual holes might solve the immediate and more visible problem. Systems thinking asks why the holes keep appearing after observing behaviour and before addressing any underlying issue.
Seasoned SEOs already work like this to some extent.
When rankings fall, SEOs investigate what’s happened and what changed before deciding whether changing title tags is the best move at that moment. During a migration, we don't consider redirects without thinking about URL architecture, internal linking, templates, hreflang, analytics and technical implementation.
These are elements of systems thinking.
The SEO paradox
SEO, especially now in AI times, is a good example of why systems thinking matters.
Organisations hold SEO teams accountable for organic and AI visibility, while many of the factors driving it sit elsewhere in the organisation, ie, product, engineering, brand, PR, customer experience and commercial priorities. SEO influences most of these areas without controlling all of them.
Meanwhile, people, search engines and AI systems discover an organisation through its website, products, reviews, publications, communities and other sources. Internally, these belong to different teams. Externally, they contribute to how the same organisation is discovered, understood and trusted or not.
Search engines and AI systems don't see our organisational charts and neither do customers.
Visibility is therefore a system outcome, not just an SEO outcome.
SEO increasingly sits at the intersection of search, brand, product and the wider business.
On The Search Session, Gianluca Fiorelli explores these connections through conversations with SEO leaders, looking at how they are changing the way we think about visibility.
Browse The Search Session episodes →
Three concepts worth looking at
Three concepts are particularly useful in everyday SEO work, with SEOs already using them instinctively.
Dependencies - tell us what relies on something else. An SEO recommendation requiring engineering isn't actionable until engineering capacity, priorities and dependencies are understood.
Constraints - limit what we can achieve. If several teams are waiting for one overloaded technical team, producing recommendations faster will not necessarily increase delivery.
This is particularly relevant as AI and automation increase problem identification speed. Our ability to find SEO issues can grow faster than an organisation’s ability to resolve them.
Feedback loops - show how the consequences of an action affect the system. Better website information might reduce customer-support demand, while support queries might expose content gaps. Closing those gaps might improve discovery and bring in customers whose behaviour also provides another source of insight.
The value lies not only in knowing what to do, but in knowing where to look before deciding how to address an issue.
The work is making this thinking systematic, instead of situational. In other words, deliberately looking for dependencies, constraints, feedback and consequences, instead of jumping immediately to tactics.
A simple framework to work with
SEOs don't need a complex systems map every time traffic declines or a rollout is delayed. Consider dependencies, constraints and feedback loops as what you are looking for.
The framework below offers a simple way of working through them:
Observe -> Connect -> Find the constraint -> Act -> Learn -> Repeat
Observe: What are we actually seeing?
Always start with the observation, not the explanation. “Organic traffic fell 20%” is an observation. “Google doesn't consider the new content strong enough to index it or mention it on AIO” is a hypothesis.
The distinction becomes particularly important when AI can produce plausible-sounding explanations almost instantly.
Connect: What is this related to?
Move past the most visible metric and ask which pages, templates, products, teams, markets or processes interact with the outcome.
Sometimes an SEO investigation needs to go beyond search and into the business.
Find the constraint: What is limiting improvement?
An audit may surface many technical issues. But if engineering cannot implement them for three months, another list of recommendations does not improve delivery, whereas a conversation about capacity might.
The constraint may be resourcing, ownership or budget. Finding it tells you where action may actually have an effect, rather than where the largest number of problems are.
Act: What happens if we change this?
Consider trade-offs before acting. Removing pages might simplify a website, while cutting customer journeys that matter. Preserving every URL during a migration might protect some search equity, while blocking architectural fixes the site needs.
The goal is to understand trade-offs, not to avoid them. Making an informed trade-off decision may be a sign that you have understood the system.
Learn: How did the system respond?
Look beyond the immediate metric to how visibility, conversion, customer behaviour, demand or other connected outcomes have changed. That response provides new information about the system and becomes the starting point for the next observation.
Good systems thinking starts with understanding what is actually happening before deciding why it is happening. That means having reliable data on how visibility is changing across pages, markets, devices, search engines and, increasingly, AI search.
Advanced Web Ranking helps you monitor those changes and give you a clearer view of the signals behind them before you decide what action to take.
What does this look like in practice?
An ecommerce site losing visibility across an important product category.
The obvious first place to look is SEO, which is after all what we control. You find product pages returning 404s when stock reaches zero, breaking internal linking and creating an inconsistent search experience.
When you follow the dependency further, you may realise that teams lack a clear process for deciding which products come back in stock and when. That changes how you address the issue. SEO can supply demand data to inform stock decision, ecommerce operations may need to improve the inventory process, while engineering may need to change platform or integrating a martech piece.
The ranking decline is the visible symptom, but the area to address sits somewhere else in the system.
AI can optimise the wrong part of the system.
Where AI fits
Generative AI has made it easier and less time-consuming for organisations to increase content production without increasing their team at the same rate.
That can look like an efficiency gain for a company. But, content is part of a system: more content creates editing and fact-checking, requires more subject-matter input, creates more pages to publish and maintain, and increases competition for internal links and user attention.
More localised content requires more market context, while more measurement creates more data that someone has to interpret.
However, none of this answers the question that actually matters: did the organisation need more content in the first place? What was the constraint that led to increasing content production?
If it was production capacity, AI may genuinely help. But, if the constraint was weak positioning, limited expertise, poor distribution or lack of customer demand, more content production amplifies the original problem faster.
AI has helped optimise the wrong part of the system.
This is the distinction I find matters most in AI-assisted work. It is easier for the process to be:
Prompt → Output → Action
A more useful version keeps a human decision in the loop:
Context → Observe → Connect → Find the constraint → AI assistance → Human judgement → Act → Learn
AI can create real leverage throughout every step of this loop, eg, analysing information, finding patterns, challenging hypotheses, comparing scenarios and accelerating execution. It can help identify where action may have the biggest positive effect. What requires human judgement is whether that action makes sense in the wider business context.
Speed without that judgement produces the wrong output faster. It does not make the system better
From fixing issues to improving systems
None of this means SEOs need to become systems theorists or that every issue needs organisational investigation. Sometimes a broken URL is just a broken URL to fix.
The skill lies in recognising which type of problem you're looking at. Many seasoned practitioners already investigate causes, work across teams, negotiate trade-offs during migrations and check what happens afterwards. They just don't always name it as a discipline or do it consistently.
The opportunity is to make that behaviour deliberate and consistent.
As AI helps us work faster and takes over more of the execution, knowing where in the system to act and why is now more valuable. Systems thinking helps us decide where we actually need that speed at any given time.
Article by
Montserrat Cano
Montserrat Cano is a global digital growth and delivery professional whose work combines international SEO and AI visibility, digital strategy and cross-functional programmes. She helps organisations navigate complexity across markets, teams and technologies, connecting strategy with execution to improve visibility, customer experience and business outcomes. Montserrat is an international speaker, author, industry contributor, Search Awards judge, mentor and Women Techmakers Ambassador.
