The Shift from Panic to Precision: Understanding the 53% Decline
The software industry is currently navigating a period of intense volatility, recently punctuated by a phenomenon Wall Street has dubbed the “SaaSpocalypse.” This term emerged after investors, spooked by the rapid advancement of autonomous AI agents like Claude Cowork and the potential for these tools to replace traditional enterprise software, erased nearly $300 billion from SaaS market caps.
Amidst this financial turbulence, new data has emerged showing a staggering 53% drop in AI-driven discovery sessions between July and December 2025. At first glance, this figure appears to confirm the worst fears of the industry: that the honeymoon phase for AI-driven software discovery is over. However, a closer look at the data reveals a much more nuanced story. This isn’t a narrative of AI’s failure, but rather a story of how AI is maturing, how user behavior is shifting toward integrated workflows, and why the “drop” is actually a reflection of standard B2B buying cycles.
For SEO professionals and digital marketers in the tech space, the 53% decline is a distraction. The real story lies in the shifting distribution of traffic, the rise of workplace-embedded AI, and the critical technical gaps that are preventing SaaS companies from being discovered by the next generation of buyers.
The Competitive Landscape: Copilot’s Meteoric Rise
Between November 2024 and December 2025, SaaS websites recorded a total of 774,331 LLM-driven sessions. While ChatGPT remains the undisputed leader in volume, the growth rates of its competitors suggest a fundamental change in where and how users interact with artificial intelligence.
SaaS AI Traffic by Source (Nov 2024 – Dec 2025)
Source | Sessions | Share
ChatGPT | 637,551 | 82.3%
Copilot | 74,625 | 9.6%
Claude | 40,363 | 5.2%
Gemini | 15,759 | 2.0%
Perplexity | 6,033 | 0.8%
While ChatGPT captures over 82% of the traffic, its growth rate has stabilized at 1.42x. In contrast, Microsoft’s Copilot has seen an explosive 15.89x year-over-year growth. In late 2024, Copilot was a non-factor, driving a mere 148 sessions. By May 2025, that number had grown 20-fold. By the end of the year, Copilot solidified its position as the second-largest referrer of AI traffic to SaaS platforms.
This growth is driven by proximity. Unlike ChatGPT, which requires a user to navigate to a separate tab or app to conduct research, Copilot is embedded directly into the Microsoft 365 ecosystem. When a business analyst is drafting a proposal in Word or a sales manager is projecting revenue in Excel, Copilot is there to answer questions like, “What CRM integrates best with our current stack?” or “Find me a project management tool for a 20-person team.”
This “workplace-embedded AI” captures intent at the exact moment it occurs. It captures the “work” that ChatGPT never sees because the user never has to leave their primary workflow. The May 2025 surge in Copilot traffic suggests a mass realization among enterprise users that they could research and evaluate software without disrupting their current tasks.
The “Internal Search” Bottleneck: Why 41.4% of Traffic is Landing on the Wrong Page
One of the most revealing aspects of the recent data is where AI-driven users land when they finally click through to a SaaS website. The distribution is highly skewed, revealing a significant gap in how AI agents perceive and navigate software sites.
Top SaaS Landing Pages by LLM Volume
Page Type | LLM Sessions | % of AI Traffic | Penetration vs Site Avg
Search | 320,615 | 41.4% | 8.7x
Blog | 127,291 | 16.4% | 8.1x
Pricing | 40,503 | 5.2% | 3.2x
Product | 39,864 | 5.1% | 2.0x
Support | 34,599 | 4.5% | 2.1x
Internal search result pages are the dominant landing surface, capturing 41.4% of all AI traffic. This is more than the combined traffic of blog, pricing, and product pages. For a SaaS marketer, this should be a cause for concern. Users aren’t landing on search pages because search pages provide the best experience; they are landing there because the AI doesn’t know where else to send them.
This is a “safety net” effect. When an LLM like ChatGPT or Claude is asked a specific question about a software’s capabilities, it attempts to find a direct answer. If the product or pricing pages lack clear, structured data that the AI can parse, the AI defaults to the site’s internal search bar. It assumes that the search schema will generate a relevant list of results even if a specific, high-value page isn’t indexed or understood.
Internal search page penetration is 8.7x the site average. This is not a sign of optimization; it is a sign of a crawlability problem. The AI recognizes the search URL structure and trusts it as a fallback. However, internal search pages are often poorly formatted for conversion, providing paginated lists with minimal detail. If your highest-intent AI traffic is landing on a generic search result page, your conversion rates will inevitably suffer.
Debunking the Decline: Seasonality and Fiscal Cycles
The 53% drop in traffic from July to December 2025 has been used by some analysts to argue that AI discovery is a dying trend. However, when we overlay this data with traditional B2B buying behavior, the decline looks less like a crash and more like a standard seasonal rhythm.
SaaS AI traffic peaked in July 2025 with 146,512 sessions. The subsequent months showed a steady decline:
July 2025: 146,512 (Peak)
August 2025: 120,802 (-17.5%)
September 2025: 134,162 (+11.1%)
October 2025: 135,397 (+0.9%)
November 2025: 107,257 (-20.8%)
December 2025: 68,896 (-35.8%)
The drop-off in November and December was particularly sharp, mirroring the behavior across all major platforms. ChatGPT’s volume was slashed by half, and even the high-growth Copilot saw its traffic nearly halved.
The reason for this is simple: AI-driven software discovery is a workplace activity. August is the height of the summer vacation season in the Northern Hemisphere. November includes the Thanksgiving holiday in the U.S., and December is dominated by the global end-of-year holiday break. When people aren’t at their desks, they aren’t using AI to research enterprise software.
Furthermore, Q4 represents the closing of the fiscal “buying window.” For many corporations, annual budgets are either exhausted by November or frozen for the year-end audit. Even if a team is still working, they are often deferring new software evaluations until Q1 when new funding becomes available. The July peak reflects mid-year momentum when Q3 budgets are fresh and projects are in full swing. The 53% drop isn’t a sign of AI’s obsolescence; it’s a sign that AI has become fully integrated into the standard B2B work cycle.
Strategic Imperatives for SEO Teams
To thrive in this new landscape, SaaS SEO teams must move beyond tracking total session volume and start focusing on the specific metrics and page types that drive AI discovery.
Segmenting by Page Type and Penetration
The site-wide AI penetration for SaaS sits at roughly 0.41%. However, this number is misleading because it averages out high-performing sections with low-performing ones. As we’ve seen, search pages (1.22%) and blog pages (1.13%) have much higher penetration than product pages (0.28%).
If you only track site-wide averages, you might miss a decline in high-value page penetration hidden by a surge in low-value blog traffic. SEO teams should segment AI traffic in GA4 or their preferred analytics platform by page category. Tracking penetration (AI sessions divided by total sessions) on a monthly basis allows you to identify which parts of your site are “AI-friendly” and which are invisible to LLMs.
Optimizing the Internal Search Surface
With 41.4% of AI traffic hitting internal search, SaaS companies must stop treating these pages as purely navigational. They are now primary discovery surfaces. If an LLM is going to send a user to your search results, those results must be crawlable and indexable.
Check your robots.txt files and ensure your search result pages aren’t being blocked. More importantly, implement structured data—such as SoftwareApplication or Product schema—directly within the search results. If a user (or an AI agent) lands on a search page, it should see pricing, key features, and user counts immediately, not just a list of blue links.
Making Data Legible: The Death of Gated Pricing
AI agents prioritize information they can verify. If your pricing is hidden behind a “Contact Sales” button, an AI agent cannot recommend you for queries involving specific budget constraints. Pricing pages currently show 0.45% AI penetration. This is lower than it should be for such a high-intent page type.
To stay competitive, SaaS companies need to publish transparent pricing on crawlable pages. This includes seat minimums, contract terms, and clear tier definitions. Transparent, structured pricing pages get cited; gated “black boxes” get ignored.
The same logic applies to content. Generic “thought leadership” that discusses broad industry trends is rarely cited by LLMs when a user asks for a software recommendation. LLMs look for comparison-focused content: “Product A vs. Product B,” “Best CRM for Small Business,” or “Top Alternatives to Salesforce.” Use tables, bullet points, and clear data points to make your blog content legible to the models.
Conclusion: The Future Belongs to the Findable
The “SaaSpocalypse” sell-off and the 53% drop in AI traffic represent a transition period for the industry. The initial novelty of AI discovery has worn off, replaced by a more deliberate, work-integrated behavior. Users are no longer just playing with chatbots; they are using embedded tools like Copilot to make real-world business decisions.
The decline in traffic during Q4 2025 is not a warning of AI’s demise, but a confirmation that it is now subject to the same seasonal and fiscal realities as the rest of the B2B world.
For SaaS companies, the challenge is no longer just about ranking on page one of Google. It is about becoming “findable” to the AI agents that are increasingly acting as intermediaries for human buyers. The winners in this new era will be the companies that optimize their internal search, expose their data through structured schemas, and prioritize transparency over gated content. As the market stabilizes, survival will favor those who have built a digital presence that AI can not only see but understand and trust.