Last year, when AI Overview first launched, we started tracking roughly 5,000 keywords associated with HR and Work Tech to measure the probability of AI Overview appearing for queries across search intent. You can read that study and the follow-up data here.
In 2025, the question is no longer “how frequently do AI-assisted search features appear” (because they almost always do). Instead, we are asking how many sources are mentioned, and if there is a relationship between traditional search results, AI Overview, and AI Mode.
This year, we refreshed our list to be more precise, cleaned out outdated queries, and included emerging tech (eg: AI), and partnered with SE Ranking to assess visibility in AI Mode (Google’s latest AI adventure).
Let’s unpack what this data means for your 2026 search strategy.
| Executive Summary 1. Your Google rankings don’t guarantee AI visibility. Treating AI visibility as a byproduct of SEO will only get you partially there. Only 25% of top-ranking URLs appeared in AI Mode. It’s essential to build a dedicated AI visibility strategy that builds from organic. 2. AI Mode and AI Overview are not the same. There was only a 15.7% domain overlap between AI Mode and AI Overview. AI Mode cited 11,000+ unique domains with sources rotating regularly. Domains with strong topical depth appear reliably across queries and systems. 3. Intent determines everything. The domains winning TOFU queries aren’t winning BOFU. AI actively selects different source types based on what the user is trying to accomplish. Map out product offerings to content to ensure you are reaching potential buyers across touchpoints. |
AI Mode Citation Patterns and Distribution for HR and Work Tech Queries
Our first objective was to determine the number of citations per response to understand whether HR and Work Tech queries trigger diverse or limited sourcing behavior.
Across all queries, AI Mode responses averaged 12.28 sources per answer, ranging from 1 to 28 citations. Only 0.12% of responses contained a single source, indicating that AI Mode consistently draws from multiple references when generating answers.
Most responses fell into the 11–15 source range (44.94%), followed by 6–10 sources (31.35%), and 16–20 sources (18.49%). Fewer than 3% of responses had fewer than six sources, and just 2.54% exceeded 20.
Search intent shapes what people search for and how AI Mode builds its answers. A user looking for “best HR software” expects a range of tools and solutions, while someone searching for “what is employee engagement” is anticipating educational context.
Queries with strong purchase intent, such as bottom-of-funnel (BOFU) searches, draw from the widest range of sources, averaging nearly 13 citations per response. Comparative and top-of-funnel (TOFU) queries follow closely, while template-style searches generate the fewest citations, averaging under 10.
In other words, AI Mode adapts its citation depth to match the user’s next step-whether they’re learning, evaluating, or deciding.
| Keyword type | Average sources |
| BOFU | 12.95 |
| Comparative | 12.21 |
| TOFU | 11.96 |
| Template | 9.68 |
| What this means for you: AI Mode cites 10-15 sources per answer on average. That’s 10-15 opportunities per query, not a winner-take-all situation like position #1 in traditional search. Stop doing this: Obsessing over whether you’re “the” answer. Focus on being “an” answer consistently. Do this instead: Measure share of voice, not rank and track how often you appear across a cluster of queries in your topic. |
Does AI Mode Use a Wide Range of Sources?
To understand how consistently AI Mode selects its sources, we analyzed domain diversity across three separate parsing attempts of the same keyword set. The results show that AI Mode draws from an unusually wide range of domains, and that this diversity shifts from one run to the next.
Dataset 1 contained 11,188 unique domains, with 2,598 domains appearing exclusively in this dataset (23.22%). Dataset 2 showed 11,134 unique domains with 2,516 exclusive domains (22.6%), while Dataset 3 presented 11,264 unique domains with 2,737 exclusive domains (24.3%).
| Dataset | Unique domains | Exclusive domains | Exclusive % |
| 1 | 11,188 | 2,598 | 23.22 |
| 2 | 11,134 | 2,516 | 22.6 |
| 3 | 11,264 | 2,737 | 24.3 |
Across all runs, 6,187 domains (35.1%) appeared consistently, and 9,774 domains (55.46%) appeared in at least two datasets.
This pattern reveals an important takeaway for marketers: AI Mode combines a stable core of high-trust domains with a long tail of rotating sources.
This means visibility is not limited to a small set of giants. AI Mode frequently brings in new, niche, or topic-specific domains depending on the query. For brands in HR and Work Tech, this creates more opportunity than traditional search:
- You don’t need to be one of the top 20 domains in the industry to be cited.
- High-quality, authoritative content on specific topics can surface even if your domain isn’t dominant overall.
- Consistency comes from credibility and depth, not sheer domain authority alone.
For marketers, this highlights the value of topical expertise, content breadth, and building trust across multiple subject areas.
| What this means for you: 35% of domains appear consistently, while 55% rotate. This means topical authority gets you into the “stable core,” while niche expertise can get you into the rotating pool. Stop doing this: Assuming you need massive domain authority to compete. Specialized expertise surfaces alongside giants. Do this instead: Identify the 10-15 domains that consistently appear for your key topics. Either become one of them, or get featured on them. |
Which Domains Are Cited the Most Often in HR and Work Tech Queries?
The two key elements of visibility in AI-assisted search are:
- Being cited as a source
- Being mentioned as a brand
AI systems tend to rely on what they perceive as authoritative domains, but what counts as “authoritative” varies heavily by industry, topic, and search intent. In HR and Work Tech, the domains AI Mode relies on show a blend of educational resources, SaaS platforms, and general business publishers, not just traditional SEO powerhouses.
A notable finding is YouTube’s dominance. Once Google-related properties are removed, YouTube emerges as the most cited domain, with 1,271 citations across 942 responses.
This is significant for two reasons:
- YouTube is the world’s second-largest search engine, and part of Google’s broader search ecosystem.
- Google has a deep understanding of YouTube’s content structure, metadata, and authority patterns, making it an ideal input for AI-generated summaries.
For HR and Work Tech topics, YouTube’s prominence suggests that video content (tutorials, product demos, explainers) is becoming a key information source for AI Mode, even in B2B contexts typically dominated by written content.
Following YouTube, Zapier, Connecteam, and Indeed all show strong and consistent visibility:
| Top 10 Domains | Number of citations | Number of answers |
| youtube.com | 1271 | 942 |
| zapier.com | 1185 | 850 |
| connecteam.com | 1112 | 746 |
| indeed.com | 1091 | 823 |
| aihr.com | 571 | 470 |
| slack.com | 482 | 398 |
| servicetitan.com | 444 | 317 |
| pumble.com | 413 | 322 |
| rippling.com | 392 | 365 |
| forbes.com | 364 | 341 |
Specialized HR education platforms like AIHR, communication tools such as Slack, and operations tools like ServiceTitan all rank highly. Forbes also shows consistent visibility as a broad business authority.
| What this means for you: YouTube is the #1 “non” Google source. Video content isn’t optional for AI visibility. It’s a primary input. Stop doing this: Treating video as a “nice to have” separate from your search strategy. Do this instead: Create video versions of your highest-value content. Tutorials, explainers, and product demos are particularly effective and commonly appear in AI Overview and AI Mode. |
Does AI Mode Change Source Domains Based on Intent?
The keyword-type breakdown shows that AI Mode adjusts its preferred sources based on what the user is trying to accomplish.
BOFU (Transactional Intent)
BOFU queries heavily feature vendor domains. Connecteam, Zapier, Rippling, Slack, and ServiceTitan rank highest, reflecting AI Mode’s preference for solution-oriented content when the user is close to making a decision.
Comparative (Evaluation Intent)
Comparative queries again prioritize Zapier (known for its comparison content and tool breakdowns), along with Pumble and PCMag. This indicates that AI Mode leans on review-style sources when users are comparing options.
TOFU (Learning / Research Intent)
TOFU queries highlight informational and educational domains:
- Indeed (840 citations)
- YouTube
- AIHR
These reflect a preference for broader, concept-oriented content early in the journey.
Template (Practical / Resource Intent)
For template-style searches, AI Mode consistently surfaces structured resource platforms like Smartsheet, GetJobber, and Housecall Pro.
| BOFU | ||
| Domain | Number of citations | Number of answers |
| connecteam.com | 639 | 413 |
| zapier.com | 502 | 401 |
| youtube.com | 477 | 355 |
| rippling.com | 178 | 162 |
| slack.com | 167 | 141 |
| nextiva.com | 163 | 107 |
| workvivo.com | 154 | 101 |
| pumble.com | 143 | 118 |
| servicetitan.com | 137 | 102 |
| zendesk.com | 127 | 102 |
| Comparative | ||
| Domain | Number of citations | Number of answers |
| zapier.com | 651 | 422 |
| connecteam.com | 359 | 242 |
| pumble.com | 257 | 191 |
| slack.com | 238 | 191 |
| youtube.com | 207 | 152 |
| pcmag.com | 172 | 149 |
| workvivo.com | 119 | 75 |
| joinblink.com | 101 | 83 |
| ringcentral.com | 94 | 76 |
| rippling.com | 88 | 83 |
| TOFU | ||
| Domain | Number of citations | Number of answers |
| indeed.com | 840 | 595 |
| youtube.com | 559 | 411 |
| aihr.com | 445 | 365 |
| betterworks.com | 334 | 292 |
| linkedin.com | 236 | 212 |
| servicetitan.com | 221 | 153 |
| deel.com | 187 | 168 |
| lattice.com | 175 | 142 |
| forbes.com | 174 | 163 |
| bamboohr.com | 164 | 151 |
| Template | ||
| Domain | Number of citations | Number of answers |
| indeed.com | 46 | 42 |
| smartsheet.com | 40 | 28 |
| servicetitan.com | 32 | 25 |
| getjobber.com | 28 | 27 |
| youtube.com | 28 | 24 |
| housecallpro.com | 27 | 21 |
| aihr.com | 19 | 18 |
| rippling.com | 19 | 18 |
| bamboohr.com | 18 | 18 |
| gofmx.com | 18 | 11 |
| What this means for you: AI Mode matches source types to user intent. Your content strategy needs to contain pages across different intents (informational, how to, comparative, templates, buying guides, and transactional). Stop doing this: Creating random acts of marketing and hoping it works. Do this instead: Map your content inventory to different search intent types and identify content gaps. Which products, features, or industries are not represented across the funnel? |
How Consistent Is AI Mode Citation Selection?
As outlined in the methodology, we used three parsing attempts per query to build a richer data set that reflects real-world search situations. When we compared the outputs from all three attempts, we observed whether AI Mode consistently returned the same sources or switched between different ones.
At the URL level, the average overlap between runs is 18.3%, meaning fewer than one in five URLs remain consistent across attempts. Most keywords (4,833) show at least some overlap, while 191 keywords (3.8%) show zero shared URLs across all three runs. Only 159 keywords (3.2%) maintain full URL overlap across every attempt.
This level of variability suggests that AI Mode often rotates between multiple valid sources rather than relying on a fixed set of URLs.
For marketers, this suggests that visibility is not strictly binary. Even if your content is eligible, it may not appear on every run. Consistency comes from building broad topical authority, not from ranking a single “lucky” page.
| Metric | URL-Level |
| Keywords with overlap | 4,833 |
| Keywords with 0% overlap | 191 |
| Keywords with 0% overlap (%) | 3.80% |
| Keywords with 100% overlap | 159 |
| Keywords with 100% overlap (%) | 3.20% |
The distribution of URL-level overlap highlights how much AI Mode varies its source selection from one parsing attempt to the next. 36.8% of keywords show an overlap of 10% or less, and another 35.6% fall within the 10–20% overlap range. Only 13.7% of keywords reach the 20–30% overlap band, and overlap levels above that become increasingly uncommon.
In other words, for most queries, only a small portion of URLs remain consistent across runs.
| Metric | Domain-Level |
| Keywords with overlap | 4,931 |
| Keywords with 0% overlap | 93 |
| Keywords with 0% overlap (%) | 1.90% |
| Keywords with 100% overlap | 160 |
| Keywords with 100% overlap (%) | 3.20% |
At the domain level, consistency improves slightly. The average domain overlap increases to 22.7%, with 4,931 keywords showing at least some shared domains across attempts. Only 93 keywords (1.9%) show no domain overlap at all. This suggests that while URLs frequently change, AI Mode tends to draw from a stable core group of trusted domains even as the exact pages shift.
URL-level visibility may fluctuate, but domain authority and topical strength offer a more stable path to recurring AI citations.
| What this means for you: Only 18.3% of URLs remain consistent across runs. But at the domain level, consistency jumps to 22.7%. AI Mode rotates through specific pages, but keeps returning to trusted domains. Stop doing this: Putting all your effort into ranking one URL. AI systems reward topical depth across your site, not a single optimized page. Do this instead: Build authority at the domain level, not the page level. |
Does Ranking in Traditional Search Help AI Mode Visibility?
AI Mode’s overlap with traditional organic search results is limited, highlighting distinct retrieval patterns between the two systems. An analysis of the top 10 organic results shows that AI Mode covers an average of 25% of URLs and 32.6% of domains appearing in Google’s traditional SERP.
At the URL level, 430 keywords (8.5%) show zero overlap with any of the top 10 organic results, while none achieve full (100%) coverage.
When aggregated at the domain level, coverage improves slightly with only 182 keywords (3.6%) showing no overlap, suggesting broader recognition of authoritative domains even if specific URLs differ.
This pattern indicates that AI Mode does not replicate traditional search rankings. Instead, it follows a query-fan-out mode that retrieves contextually relevant results from a wider range of sources, rather than prioritizing top organic listings alone.
This sourcing pattern points to an intent-driven retrieval strategy that is more exploratory than the keyword-based ranking logic used in traditional search.
| What this means for you: Organic rankings improve your odds, but don’t guarantee AI visibility. We have a new marketing channel to strategize, monitor, and measure. Stop doing this: Assuming your SEO strategies are building and/or maintaining your AI visibility. Do this instead: Continue investing in SEO fundamentals, but add AI visibility tracking as a separate KPI. |
Does Appearing in AI Overviews Help Visibility in AI Mode?
AI Overviews appeared for 72% of the analyzed keywords, indicating strong deployment across HR and Work Tech related queries.
However, appearing in an AI Overview does not necessarily translate to visibility in AI Mode. The overlap between the two features remains limited at just 11.9% at the URL level and 15.7% at the domain level.
In practical terms, this means that only a small fraction of pages or domains featured in AI Overviews are also surfaced in AI Mode responses.
A deeper look shows that 1,329 keywords (27.5%) have zero URL overlap and 1,251 keywords (25.9%) have no domain overlap between the two AI-powered experiences. This limited intersection suggests that AI Mode and AI Overviews draw from different data sources and use distinct retrieval logic.
For marketers, these findings provide evidence that visibility within AI search experiences is fragmented across different systems. To maximize exposure, marketers should focus on semantic relevance and diversified content strategies that position their brand across multiple AI-driven surfaces.
| What this means for you: AI Overview and AI Mode do not mirror each other’s results. Due to the probability nature of LLMs, sources are selected based on a range of personalized factors. Stop doing this: Treating AI search as one monolithic thing. AI Mode, AI Overview, ChatGPT, Claude, etc are not simply replicating each other’s results or pulling from the same deck of sources. Do this instead: Track your visibility across your most important AI sources to benchmark visibility and identify gaps. |
How HR and Work Tech Marketers Can Optimize for AI Visibility
Research and data make your reports look good, but let’s talk about what you should actually do with this information. How can you walk into your next planning meeting with the language and evidence to get buy-in for AI visibility efforts?
Below are three key takeaways from our analysis, along with specific actions you can take.
1. Optimize for Multiple AI Platforms, Not Just One
AI search is evolving rapidly. New LLM models, platforms, and AI browsers launch constantly, and what works today may not work in six months.
You might see SEOs on Reddit advocating for Reddit-focused strategies due to its visibility in AI Mode and AI Overviews across many industries. However, our data consistently shows that Reddit is not a primary source for HR and Work Tech queries. What dominates in retail or tech may not apply to your niche.
This is why diversification matters. Rather than chasing a single platform or tactic, focus on building a resilient content strategy across the internet:
- Create content in multiple formats: written articles, video tutorials, structured templates, and case studies
- Distribute across owned channels (your blog), earned channels (guest posts, industry publications), borrowed channels (YouTube, LinkedIn, review sites), and paid channels.
- Optimize for traditional search while also preparing for AI-assisted search
- Monitor how your content performs across different AI systems (AI Overview, AI Mode, ChatGPT, Perplexity) using tools like our LLM Dashboard
We’re in the early days of AI search. The brands that will succeed are those building for flexibility, not betting everything on one system.
2. Organic Rankings Still Matter, But They Don’t Guarantee AI Visibility
One of the most common questions we get is: “If I rank on page 1, will I appear in AI Mode?”
The answer: sometimes, but not always.
Our data shows that AI Mode covers only 25% of URLs and 32.6% of domains from the top 10 organic results. Additionally:
- 8.5% of keywords show zero URL overlap between AI Mode and top 10 organic results
- 25.9% of keywords show no domain overlap between AI Mode and AI Overview
- Only 3.2% of keywords maintain 100% URL consistency across multiple AI Mode checks
AI Mode doesn’t simply republish Google’s top 10 results. Instead, it pulls from a broader set of sources based on topical relevance and search intent. So how should we prepare?
- Continue investing in traditional SEO, as ranking well still improves your odds of AI visibility
- Build topical authority across multiple pages (and properties)
- Create content that directly answers user questions at each stage of the journey
- Track both organic rankings and AI visibility to understand your full search presence
The goal is to become a trusted source on specific topics, not just rank for keywords. AI systems reward depth and relevance across your content ecosystem.
3. Intent Determines Which Domains AI Mode Prefers
AI Mode values topical authority, not just domain authority.
Not all domains are treated equally across different types of queries. Our breakdown by search intent revealed that AI Mode actively adjusts its source preferences based on what the user is trying to accomplish.
Many AI visibility studies group all keywords together across industries, which creates a misleading picture of what “works.” Even within HR and Work Tech, the dominant sources shift dramatically based on query intent.
Map your content strategy to specific intents and identify which domains dominate each stage:
- For TOFU content, focus on educational depth, clear explanations, and diverse media formats (video, articles, infographics)
- For BOFU content, ensure your product pages are detailed, include use cases, and demonstrate clear value
- For comparative content, get featured in trusted review sites and comparison platforms (G2, Capterra, PCMag)
The simple version? Stop trying to be everywhere. Use your competitor and domain analysis to identify the 10-15 sources that matter most for your intent mix, then focus your digital PR efforts there.
AI Mode Visibility for HR and Work Tech
This research analyzed 5,085 queries. Here’s what the data says you should be asking yourself:
- AI Mode cites 12+ sources per query. Are you one of them?
- 55% of cited domains rotate regularly. Are you in the rotation?
- Only 25% of top-ranking URLs appear in AI Mode. Is your organic traffic about to erode?
- AI Overview and AI Mode have just 15.7% domain overlap. Are you visible in both?
Your competitors are getting cited. Are you?
The brands winning AI visibility aren’t doing anything mysterious. They’re building topical authority through comprehensive content, multiple formats, and presence on the platforms AI systems trust.
Where do you stand?
If you’re seeing traffic shifts you can’t explain, or planning your 2026 marketing strategy and want to account for AI search, start with visibility data. Our LLM Visibility Audit shows you:
- How much traffic you’re currently getting from ChatGPT Search, Perplexity, and Claude (and how it converts)
- Where your brand is being cited and where competitors are appearing
- Which domains are influencing AI responses in your category
- An action plan for closing your biggest visibility gaps
Want to see where you stand? Get in touch about an LLM Visibility Audit!
| 📌Methodology: SE Ranking helped us analyze 5,085 HR-related keywords and tested each one three separate times on the same day using Google’s AI Mode. This helped us measure the outputs and the consistency of AI Mode’s citations and source selection behavior. To reflect different stages of search intent and content formats, keywords were categorized into four groups: – TOFU (Top-of-Funnel) informational and research-driven queries – BOFU (Bottom-of-Funnel) evaluation and purchase-driven queries – Comparative “vs” and alternative evaluation queries – Template reusable prompt-style or format-seeking queries We evaluated results across four dimensions: – Citation volume: number of sources displayed – Source volatility: changes in source presence or order across attempts – Overlap with AI Overviews (AIO): shared or divergent citation patterns – Relationship to organic rankings: directional alignment vs. divergence All comparative metrics are based on the second crawl, selected for noise reduction. Volatility findings incorporate all three crawls to determine stability, churn, and repeatability.⚠️ Note: This methodology measures behavior at a fixed point in time and may not reflect system adjustments Google deploys after the test window. |
