Zbyněk Fridrich is a contract website positioning specialist and final yr’s winner of the Greatest website positioning Mission award within the Czech Republic.
With 17 years within the subject, he has one arduous rule for each new shopper: “If you wish to work with me, you want Semrush.”
To place his purchasers for fulfillment, he analyzes how AI programs suggest manufacturers—and the place his purchasers are lacking from these suggestions.
On this information, I’ll share his actual workflow and what it produced for WorkLounge, a co-working model that doubled its natural visibility and AI visitors in 5 months.
The Two-Section Method to Search Visibility
Zbyněk’s workflow covers two important steps that leverage AI sentiment and AI immediate insights from Semrush’s AI Visibility Toolkit:
- Section 1: Get management of your sentiment—perceive what AI is saying about your model proper now, and repair it.
- Section 2: Discover new alternatives—discover prompts the place try to be showing however aren’t and shut the hole.
“The purpose is to have most management over what AI says about us in the present day. Solely within the second step can we search for new content material alternatives.”
Skipping section 1 is the error most manufacturers make. If AI has the improper image of you, mindlessly creating extra content material doesn’t repair the issue.
Step 1: Analyze Sentiment in Model Efficiency
Zbyněk begins by opening the Model Efficiency instrument, including the shopper’s key rivals and goal location, and checking how engines like ChatGPT, Google AI, and Gemini are describing the model.
That is essential as a result of visibility in AI search isn’t pushed by “rankings” alone. It’s decided by whether or not AI fashions think about your model authoritative and related sufficient to suggest.
“It is not essential to be in each reply. What issues most is sentiment and total visibility.”
First, Zbyněk examines the model’s total sentiment in AI solutions and checks the way it compares to different gamers within the area of interest.
Then, he scrolls right down to the Enterprise Drivers report and explores the precise attributes, associations, and subjects that maintain developing in AI outputs.
Subsequent, he opens the Notion report to investigate particular perceptions—constructive and destructive—that affect AI positioning.
Lastly, Zbyněk critiques the AI-generated strategic suggestions within the toolkit—actionable ideas primarily based on his shopper’s AI visibility information.
“For WorkLounge, the preliminary image was clear—and damaging. AI persistently described the area as loud. Telephone cubicles weren’t talked about wherever. Member entry was described as 9–5, regardless that members can come and go 24/7. None of this was precisely correct, however it’s what AI needed to work with primarily based on the prevailing web site content material.”
Step 2: Rewrite Current Web site Content material
From right here, Zbyněk identifies the important thing areas of focus—notion change factors—and updates the shopper’s web site to right no matter AI is getting improper.
If destructive or inaccurate critiques or third-party sources are shaping AI’s notion of your model, your web site content material must straight tackle and counter these narratives.
“I by no means work on new content material if I have not fastened the content material already on the web site.”
For WorkLounge, that meant going by means of 90 pages of product and repair content material and rewriting them to provide AI correct data. For instance, Zbyněk coated:
- Opening hours: Structured content material now clearly distinguishes 24/7 member entry from 9-5 public hours
- Quiet zones: Telephone cubicles and quiet areas—already bodily there—lastly documented on the positioning
- Membership advantages: Product pages rewritten so the member vs. non-member expertise is unmistakable
Step 3: Repair Technical Points and Implement Structured Knowledge
Subsequent, Zbyněk makes use of the Website Audit instrument to establish and repair points that stop engines like google and LLMs from accurately studying the shopper’s web site.
These fixes embody:
- Including structured information
- Bettering web page construction
- Enhancing content material formatting
- Resolving points with inner hyperlinks
He additionally experiments with LLM.txt—a file that provides AI crawlers specific directions about the best way to interpret your web site’s content material.
It is nonetheless an rising observe with no definitive proof of affect, however Zbyněk has seen constructive outcomes with it for his purchasers.
Step 3: Use AI Immediate Knowledge to Plan Content material
As soon as sentiment is in fine condition, Zbyněk pulls the prompts from the Narrative Drivers instrument and makes use of them to plan new content material.
These are the precise questions individuals ask AI instruments about subjects within the shopper’s area.
He then downloads the info to construct a complete audit and reporting file for his purchasers.
Lastly, Zbyněk picks the 20-30 most related prompts per venture. Every immediate turns into an FAQ block on the related services or products web page or a brand new content material piece.
For WorkLounge, this meant constructing FAQ sections round questions like what the entry coverage is for members, what quiet work choices can be found, and the way the area compares to conventional places of work.
To sum all of it up, prompts additionally inform him the place the gaps are. If a immediate is not surfacing his shopper in a robust place, that is both a chunk of content material that should exist or one which must be fastened.
Step 4: Distribute Throughout Channels on the Proper Time
From right here, Zbyněk works on pushing this new content material throughout all channels—weblog, social, publication, hyperlink constructing—timed to when search demand for that subject peaks.
“AI instruments learn your complete digital footprint, not simply your web site. The extra persistently and precisely a model seems throughout trusted sources, the stronger the sign.”
For WorkLounge, content material about quiet zones and telephone cubicles went out in every single place without delay, together with an up to date Google My Enterprise profile. Social posts bolstered the identical messaging. Publication content material tied it collectively.
The timing issues too.
Zbyněk aligns content material pushes with seasonal demand peaks for every subject—so the content material lands when individuals are actively trying to find it, not each time the editorial calendar occurs to release.
Step 5: Monitor Efficiency in Parallel
To trace the effectiveness of his efforts and iterate, Zbyněk screens goal prompts and key phrases within the Place Monitoring instrument.
This enables him to see the natural affect throughout key search surfaces—from Google AI to ChatGPT.
For WorkLounge, AI Overview visibility went from 17% to 35% over 5 months—straight tied to the content material modifications made in phases 1 and a pair of.
And guess what? Natural visitors and rankings adopted, too.
Site visitors attributed to ChatGPT has additionally grown practically 20x versus the prior interval and continues to development upward.
“This complete cocktail is aimed toward bettering visibility in LLM programs, growing natural visitors, getting extra mentions in AI overviews, and boosting the general enterprise efficiency of the web site.”
Step 6. Generate Shopper- and Exec-Prepared Experiences
Lastly, Zbyněk makes use of Semrush’s My Experiences characteristic to export the AI Visibility and website positioning information into clear, shareable PDF stories.
These go straight into shopper displays—displaying sentiment scores, immediate visibility, strategic ideas, and progress over time with out requiring purchasers to log into the instrument themselves.
It is also how he will get buy-in. When purchasers see AI describing their model negatively in a sophisticated report, the case for fixing it makes itself.
The Outcomes: What 5 Months of AI Visibility Work Look Like
WorkLounge began this course of in September. By January, each metric Zbyněk tracks had moved:
- Sentiment rating: 67 → 82. Essentially the most elementary shift. AI instruments went from describing WorkLounge as a loud, 9-to-5 workplace area to precisely representing what the model truly gives—quiet zones, 24/7 member entry, and a spread of workspace choices.
- AI Overview visibility: 17% → 34%. The share of related key phrases the place WorkLounge seems in Google AI Overviews doubled. It is a direct consequence of the content material modifications—AI now has correct, well-structured data to tug from.
- Natural visitors: up. Higher AI sentiment results in extra correct AI suggestions, which results in extra branded searches, which results in extra natural visitors. The chain is constant throughout all of Zbyněk’s purchasers operating this workflow.
“I by no means do one thing solely due to AI. I deal with AI visibility as a part of website positioning.”
The New Natural Visibility Playbook
Zbyněk’s strategy is not sophisticated. He makes use of AI sentiment information to seek out what’s damaged, fixes the content material, then builds on prime of a clear basis. The outcomes observe.
What makes it work is treating AI visibility and website positioning as one technique—not two separate workstreams.
Repair the narrative AI tells about your model, and all the pieces else improves with it: natural rankings, AI Overview appearances, and referral visitors.
That is precisely what Semrush One is constructed for. It combines conventional website positioning instruments with AI visibility information in a single related workflow—so you may monitor immediate visibility and key phrase rankings, run web site audits for each engines like google and AI crawlers, and report on all of it with out switching instruments.
If you wish to run the identical workflow Zbyněk makes use of, that is the place to start out.
144 Comments