AI Integration Model for SEO
Four levels that measure how deeply AI is built into SEO work, scored across audit, strategy, implementation and reporting.
The Four Levels
Each level describes where the know-how lives. It starts in one person’s head and ends in the system itself.
- Assisted
- Standardized
- Productized
- Adaptive
-
Assisted
Individuals use AI ad hoc, beside the work. Each person prompts their own way, and what they learn stays with them.
- Know-how lives in
- A person
- Test
- Does anyone use AI for this work?
-
Standardized
Tasks run on shared, reusable assets such as Skills, saved prompts and tools. What gets standardized is the method. The output can still vary from run to run.
- Know-how lives in
- A shared asset
- Test
- Do people run a shared asset, or write the prompt fresh?
-
Productized
Whole workflows are packaged. Inputs go in and a finished deliverable comes out, whoever runs it. People review the output, not each step. Productized describes how the work runs, not how it is sold.
- Know-how lives in
- A packaged workflow
- Test
- Could someone who didn’t build it run it on a new site or client with only the inputs?
-
Adaptive
Workflows are connected, and results change what the system does next. What shipped, what worked and what was rejected all feed the next cycle.
- Know-how lives in
- The system
- Test
- Did last cycle’s results change this cycle’s behavior without a manual update?
Work with no AI in it is unrated. There is no level for it, because every level describes a way of using AI.
Four Dimensions
A team gets one level for each dimension of the work. Most teams sit at different levels in different places, so the result is a profile. The four levels are never averaged into one.
- Audit
- Establishing what is true about a site.
- Strategy
- Deciding what to do about what is true.
- Implementation
- Performing the work.
- Reporting
- Showing what changed and what it did.
-
Audit Level 3, Productized
-
Strategy Level 1, Assisted
-
Implementation Level 2, Standardized
-
Reporting Level 2, Standardized
Score Your Team
Pick a dimension and answer one question per level. Your level is the highest yes in an unbroken run, so a no closes the levels above it.
Audit
Not scoredAnswer the first question to start.
-
Level 1, Assisted
Does anyone use AI when doing audit work?
-
Level 2, Standardized
For audit tasks, do people run a shared, reusable asset (a Skill, saved prompt or tool) instead of writing the prompt fresh?
Opens when Assisted is a yes.
-
Level 3, Productized
Could someone who didn’t build the audit workflow run it on a new site or client by supplying only the inputs?
Opens when Standardized is a yes.
-
Level 4, Adaptive
Did last cycle’s results change what the audit workflow did this cycle, without a manual update?
Opens when Productized is a yes.
Answers stay in this browser. Nothing is sent anywhere.
What Each Level Looks Like
Two examples of SEO work at each level.
-
Assisted
- Someone pastes a crawl export into a chatbot, asks for the biggest issues, and pastes the answer into the audit document.
- A strategist asks a chatbot for content brief ideas, then rewrites them into the team’s template.
-
Standardized
- Keyword clustering runs from a shared Skill that anyone on the team can use, in place of each person’s own prompt.
- Alt text for audit images is drafted by a saved tool, and a person approves or edits each one.
-
Productized
- A technical audit runs start to finish: AI pulls the crawl and search data, classifies and scores the findings, and produces the deliverable for review.
- Monthly reporting runs on a schedule: AI pulls the numbers, writes the analysis, flags anomalies, and sends a draft to the account lead.
-
Adaptive
- The measured impact of each shipped fix changes how the next audit scores similar findings.
- Recommendations a client rejected, and the reasons, are logged and shape what the system proposes for that client next time.
Moving Up A Level
The kind of resource changes at each step. The first costs attention, the second costs build skill, and the third costs data and time.
Level 1 to 2
Assisted To
Standardized
The main cost is attention.
- People
- One owner per dimension
- Tools
- A shared asset library
- Inputs
- Examples of good output
- Work
- Package the repeated tasks
Level 2 to 3
Standardized To
Productized
The main cost is build skill.
- People
- A builder, a workflow owner
- Tools
- Data access, step chaining
- Inputs
- Input list and output spec
- Work
- Chain the assets, then test
Level 3 to 4
Productized To
Adaptive
The main cost is data and time.
- People
- Measurement owner, engineer
- Tools
- A shared results record
- Inputs
- Outcomes across cycles
- Work
- Capture results, feed back
Jumping Levels
- 1 to 3 works. A strong builder can go straight to a packaged workflow and create the shared assets along the way. The risk is that only the builder understands it.
- Skipping to 4 does not. Adaptive needs a running productized workflow that produces results to learn from.
- Design for 4 while building 3. Log every finding, decision and outcome from day one, and the last step gets much shorter.