The SEO industry is undergoing a profound transformation in 2025.
As large language models (LLMs) increasingly power search experiences, success now depends on withstanding traditional algorithm fluctuations and strategically positioning brands within AI knowledge systems.
This article explores key insights and practical implementation steps to navigate this evolving landscape.
Withstanding algorithm updates in 2025
Traditional algorithm updates remain a reality, but our approach to handling them must evolve beyond reactive tactics.
The typical SEO response to traffic fluctuations follows a familiar pattern:
- Identify the drop date.
- Cross-check with known updates.
- Audit on-site changes.
- Analyze content.
- Review backlinks.
- Check competitors.
- Look for manual actions.
This reactive methodology is no longer sufficient.
Instead, we need data-driven approaches to identify patterns and predict impacts before they devastate traffic.
Let me share three key strategies.
Breaking down the problem with granular analysis
The first step is drilling down to understand what changed after an update.
- Was the entire website affected, or just certain pages?
- Did the drop affect specific queries or query groups?
- Are particular sections or content types (like product pages vs. blog posts) impacted?
Using filtering and segmentation, you can pinpoint issues with precision.
For example, you might discover that a traffic drop:
- Primarily affected product pages rather than blog content.
- Or specifically impacted a single category despite maintaining rankings, potentially due to a SERP feature drawing clicks away from organic listings.
Leveraging time series forecasting
One of the most powerful approaches to algorithm analysis is using time series forecasting to establish a baseline of expected performance.
Meta’s Prophet algorithm is particularly effective for this purpose, as it can account for:
- Daily and weekly traffic patterns.
- Seasonal fluctuations.
- Overall growth or decline trends.
- Holiday effects
By establishing what your traffic “should” look like based on historical patterns, you can clearly identify when algorithm updates cause deviations from expected performance.
The key metric here is the difference between actual and forecasted values.
By calculating these deviations and correlating them with Google’s update timeline, you can quantify the impact of specific updates and distinguish true algorithm effects from normal fluctuations.
SERP intent classification
As search engines’ understanding of user intent evolves, tracking intent shifts becomes crucial.
By analyzing how Google categorizes and responds to queries over time, you can identify when the search engine’s perception of user intent changes for your target keywords.
This approach involves:
- Classifying search queries by intent (informational, commercial, navigational, etc.).
- Monitoring how SERP layouts change for each intent type.
- Identifying shifts in how Google interprets specific queries.
When you notice declining visibility despite stable rankings, intent shifts are often the culprit.
The search engine hasn’t necessarily penalized your content. It’s simply changed its understanding of what users want when they search those terms.
The rise of AI-driven search and entity representation
While traditional algorithm analysis remains important, a new frontier has emerged: optimizing for representation within AI models themselves.
This shift from ranking pages to influencing AI responses requires entirely new measurement and optimization approaches.
Measuring brand representation in AI models
Traditional rank tracking tools don’t measure how your brand is represented within AI models.
To fill this gap, we’ve developed AI Rank, a free tool that directly probes LLMs to understand brand associations and positioning.
Brand AI visibility tracking
Here, I’ll illustrate the approach to measuring and interpreting AI visibility for one participating brand.
We utilize two prompt modes and collect this data on a daily basis:
- Brand-to-Entity (B→E): “List ten things that you associate with Owayo.”
- Entity-to-Brand (E→B): “List ten brands that you associate with custom sports jerseys.”
This bidirectional analysis creates a structured approach to AI model brand perception.
The analysis performed after two weeks of data collection revealed that this brand is strongly associated with:
- “Custom sportswear” (weighted score 0.735).
- “Team uniforms” (0.626).
This shows strong alignment with their core business.
However, when looking at which brands AI models associate with their key product categories, dominant players like Nike (0.835), Adidas (0.733), and Under Armour (0.556) consistently outrank them.
Tracking association strength over time
In addition to an aggregate overview, tracking how these associations evolve daily is important, revealing trends and shifts in AI models’ understanding.
For this brand, we observed that terms like “Custom Sports Apparel” maintained strong associations, while others fluctuated significantly.
This time-series analysis helps identify stable brand associations and those that may be influenced by recent content or model updates.
Competitive landscape analysis
When analyzing which brands AI models associate with specific product categories, clear hierarchies emerge.
For “Custom Basketball Jerseys,” Nike consistently holds Position 1, with Adidas and Under Armour firmly in Position 2 and Position 3, but where is Owayo?
This visualization exposes the competitive landscape from an AI perspective, showing how challenging it will be to displace these established associations.
Grounded vs. ungrounded responses
A particularly valuable insight comes from comparing “grounded” responses (influenced by current search results) with “ungrounded” responses (from the model’s internal knowledge).
This comparison reveals gaps between current online visibility and the AI’s inherent understanding.
Ungrounded responses show stronger associations with cycling and esports jerseys, while grounded responses emphasize general custom sportswear.
This highlights potential areas where their online content might be misaligned with their desired positioning.
Strategic implications: Influencing AI representation
These measurements aren’t just academic; they’re actionable.
For this particular brand, the analysis revealed several strategic opportunities:
- Targeted content creation: Developing more content around high-value associations where they weren’t strongly represented
- Entity relationship strengthening: Creating explicit content that reinforces the connection between their brand and key product categories
- Competitive gap analysis: Identifying niches where competitors weren’t strongly represented
- Dataset contribution: Publishing structured datasets on Hugging Face that establish their expertise in specific sportswear categories
Implementing a proactive AI strategy
Based on these insights, here’s how forward-thinking brands can adapt to the AI-driven search landscape.
Direct dataset contributions
The most direct path to influence AI responses is contributing datasets for model training:
- Create a Hugging Face account (huggingface.co).
- Prepare structured datasets that prominently feature your brand.
- Upload these datasets for use in model fine-tuning.
When models are trained using your datasets, they develop stronger associations with your brand entities.
Creating RAG-optimized content
Retrieval-augmented generation (RAG) enhances LLM responses by pulling in external information. To optimize for these systems:
- Structure content for easy retrieval: Use clear, factual statements about your products/services.
- Provide comprehensive product information: Include detailed specifications and use cases.
- Craft content for direct quotability: Create concise, authoritative statements that RAG systems can extract verbatim.
Building brand associations through entity relationships
LLMs understand the world through entities and their relationships. To strengthen your brand’s position:
- Define clear entity relationships: “Owayo is a leading provider of custom cycling jerseys.”
- Create content that reinforces these relationships: Expert articles, case studies, authoritative guides.
- Publish in formats that LLMs frequently index: Technical documentation, structured knowledge bases.
Measure, optimize, repeat
Implement continuous measurement of your brand’s representation in AI systems:
- Regularly probe LLMs to track brand and entity associations.
- Monitor both grounded and ungrounded responses to identify gaps.
- Analyze competitor positioning to identify opportunities.
- Use insights to guide content strategy and optimization efforts.
From SEO to AI influence
The shift from traditional search to AI-driven information discovery requires a fundamental strategic revision.
Rather than focusing solely on ranking individual pages, forward-thinking marketers must now:
- Use advanced forecasting to better understand algorithm impacts.
- Monitor SERP intent shifts to adapt content strategy accordingly.
- Measure brand representation within AI models.
- Strategically influence training data to shape AI understanding.
- Create content optimized for both traditional search and AI systems.
By combining these approaches, brands can thrive in both current and emerging search paradigms.
The future belongs to those who understand how to shape AI responses, not just how to rank pages.
Future work
Savvy data scientists will notice that some data tidying is in order, starting with normalizing terms by removing capitalization and various artifacts (e.g., numbers before entities).
In the coming weeks, we’ll also work on better concept merging/canonicalization, which can further reduce noise and perhaps even add a named entity recognition model to aid the process.
Overall, we feel that much more can be derived from the collected raw data and invite anyone with ideas to contribute to the conversation.
Disclosure and acknowledgments: AI visibility data was collected via AI Rank with written permission from Owayo brand representatives for exclusive use in this article. For other uses, please contact the author.
[Watch] How to withstand Google algorithm updates in 2025
Watch my SMX Next session for more insights on how to improve your site and withstand future algorithm updates.
source https://searchengineland.com/withstand-algorithm-updates-optimize-ai-search-453182
0 Comments