A deep dive into how one of the world’s largest software download platforms transformed their presence across AI-powered search
Executive summary
Softonic, a global leader in software downloads with over 100 million monthly users, faced a critical challenge: despite decades of brand authority, their visibility in AI-powered search experiences was declining rapidly. After implementing LLM Pulse, Softonic achieved remarkable improvements across all key AI visibility metrics within 6 months.
Key results at a glance
| Metric | Before LLM Pulse | After LLM Pulse | Change |
|---|---|---|---|
| Brand mentions in LLM responses | 12% | 47% | +292% |
| Citation rate | 8% | 35% | +337% |
| Net sentiment score | 0.42 | 0.78 | +86% |
| Share of voice (software downloads category) | 15% | 52% | +247% |
| Reputation score | 61/100 | 89/100 | +46% |
The challenge
A legacy brand facing the AI search shift
Softonic has been a trusted name in software downloads since 1997. With a vast library of over 400,000 programs and reviews spanning two decades, they had built an unassailable position in traditional search.
But the landscape was shifting.
By early 2025, Softonic’s analytics team noticed a troubling pattern:
- Direct traffic from AI assistants was growing 340% year-over-year – but Softonic wasn’t capturing their share
- Competitors with weaker domain authority were being recommended more frequently in ChatGPT, Claude, and Perplexity responses
- User queries like “where to download [software] safely” increasingly returned recommendations that excluded Softonic entirely
- Their brand was sometimes mentioned with outdated or negative context from years-old forum posts
The problem was clear: traditional SEO metrics no longer told the full story. Softonic needed visibility into how AI models perceived and recommended their brand.
What they couldn’t measure, they couldn’t improve
Before LLM Pulse, Softonic’s team operated blind:
- No way to track how often they were mentioned across different LLMs
- No visibility into whether mentions were positive, neutral, or negative
- No understanding of which competitors were winning the AI recommendation game
- No data on which product categories showed strength vs. weakness in AI responses
The solution: implementing LLM Pulse
Phase 1: Discovery and benchmarking (weeks 1-4)
Softonic deployed LLM Pulse to establish baseline metrics across:
- 4 major LLM platforms: ChatGPT, Claude, Perplexity, and Gemini
- 12 target markets: US, UK, Spain, Germany, France, Italy, Brazil, Mexico, Argentina, India, Japan, and Indonesia
- 847 tracked queries spanning software categories, brand terms, and competitor comparisons
- 23 direct competitors monitored for share of voice analysis
Initial findings were sobering:
The baseline audit revealed Softonic was mentioned in only 12% of relevant queries – far below their traditional search market share of 35%. Worse, when mentioned, the sentiment skewed negative (0.42 net sentiment score) due to AI models surfacing outdated complaints about bundled software from 2015-2018.
Phase 2: Strategic optimization (weeks 5-16)
Armed with LLM Pulse data, Softonic’s team implemented a multi-pronged strategy:
1. Content reformulation for AI comprehension
LLM Pulse’s query analysis revealed that AI models struggled to extract clear value propositions from Softonic’s existing content. The team:
- Restructured 3,200+ software review pages with clearer recommendation frameworks
- Added explicit safety verification language that AI models could cite
- Created 156 new comparison guides optimized for AI synthesis
2. Entity and knowledge graph optimization
Using LLM Pulse’s citation tracking, the team identified which sources AI models relied on for software download recommendations:
- Updated Wikipedia presence with current, accurate information
- Secured mentions in 47 authoritative tech publications
- Built partnerships with 12 industry review sites for co-citation
3. Reputation repair campaign
The net sentiment analysis pinpointed exactly which negative narratives were being surfaced. Softonic:
- Published a transparent “Our journey” page addressing historical bundleware concerns
- Generated 890+ fresh user reviews through an improved review collection system
- Created video content demonstrating current safety protocols
4. Competitive intelligence activation
LLM Pulse’s share of voice tracking revealed competitors’ positioning strategies:
- Identified 8 query categories where competitors dominated despite inferior offerings
- Discovered that certain long-tail queries had zero brand competition – easy wins
- Mapped competitor citation sources and developed relationships with the same publishers
Phase 3: Continuous monitoring and iteration (ongoing)
Softonic established a weekly review cadence using LLM Pulse dashboards:
- Daily: Automated alerts for sentiment drops or competitor gains
- Weekly: Query performance review and content gap analysis
- Monthly: Strategic planning based on trend data
- Quarterly: Comprehensive competitive landscape assessment
Results: the transformation
Brand mentions: from invisible to unavoidable
Within 6 months, Softonic’s presence in AI responses transformed dramatically:
Mention rate progression:
- Month 0 (baseline): 12%
- Month 2: 19%
- Month 4: 31%
- Month 6: 47%
The breakthrough came in month 3, when optimized content began being incorporated into LLM training data and RAG systems.
Citation quality: becoming the trusted source
Raw mentions mean little without context. LLM Pulse tracked not just if Softonic was mentioned, but how:
Citation improvements:
- Direct link citations increased from 8% to 35% of relevant queries
- “Recommended by” phrasing appeared in 62% of mentions (up from 23%)
- First-position recommendations grew from 4% to 28% of competitive queries
Sentiment transformation: rewriting the narrative
Perhaps the most remarkable shift was in how AI models characterized Softonic:
Net sentiment score evolution:
- Baseline: 0.42 (slightly positive, but with significant negative mentions)
- Month 6: 0.78 (strongly positive, consistent trust signals)
Sentiment composition shift:
| Sentiment | Before | After |
|---|---|---|
| Positive | 48% | 71% |
| Neutral | 29% | 21% |
| Negative | 23% | 8% |
The negative mentions that remained were primarily about regional availability issues – addressable operational concerns rather than trust problems.
Share of voice: category dominance
In the “software downloads” category, Softonic went from underdog to leader:
Share of voice by platform (month 6):
- ChatGPT: 58%
- Claude: 49%
- Perplexity: 61%
- Gemini: 42%
- Average: 52% (up from 15% baseline)
Reputation score: rebuilding trust at scale
The composite reputation score – combining mentions, citations, sentiment, and authority signals – showed the cumulative impact:
Reputation score progression:
- Baseline: 61/100
- Month 6: 89/100
- Percentile ranking: Top 3% among tracked software brands
Business impact
Traffic and conversion
The AI visibility improvements translated directly to business results:
- AI-referred traffic: +215% increase in visits from AI assistant deep links
- Conversion rate: AI-referred visitors converted at 2.3x the rate of organic search visitors
- Revenue attribution: Estimated €1.2M additional annual revenue from AI channel
Brand perception
Qualitative research showed shifting user perceptions:
- Brand trust scores increased 34% in post-visit surveys
- “Safe and reliable” associations grew from 41% to 67%
- Unprompted recall in the software downloads category improved from 28% to 51%
Competitive positioning
Softonic’s gains came partially at competitors’ expense:
- Primary competitor share of voice dropped from 31% to 19%
- Softonic became the default recommendation in 73% of “safe download” queries
- Three competitors initiated similar AI optimization programs (tracked via LLM Pulse competitor monitoring)
Implementation insights
What worked
- Data-driven prioritization: LLM Pulse’s query-level data allowed Softonic to focus on high-impact opportunities first, rather than boiling the ocean.
- Cross-functional alignment: The dashboards gave marketing, content, PR, and product teams a shared understanding of AI visibility goals.
- Patience with compound effects: Early wins were modest. The team trusted the process and saw exponential gains as optimizations compounded.
- Competitive intelligence as motivation: Watching competitor movements in real-time created urgency and informed counter-strategies.
Lessons learned
- AI visibility is a lagging indicator: Changes to content and reputation take 4-8 weeks to appear in LLM responses. Plan accordingly.
- Platform differences matter: Each LLM has different source preferences and update cycles. A single strategy won’t optimize all platforms equally.
- Negative sentiment is sticky: Old negative content requires proactive displacement, not just new positive content.
- Citation sources vary by query type: Informational queries pull from different sources than transactional queries. Optimize for both.
Looking ahead
Softonic continues to use LLM Pulse as their central AI visibility command center. Current initiatives include:
- Multilingual expansion: Extending optimization to 8 additional language markets
- Voice assistant optimization: Tracking and improving presence in Alexa, Siri, and Google Assistant responses
- Proactive reputation monitoring: Using sentiment alerts to address emerging issues before they scale
- New product launches: Pre-optimizing AI visibility for upcoming product categories
About this case study
Company: Softonic
Industry: Software downloads and reviews
Company size: 100M+ monthly users
Markets: Global (primary focus: US, Europe, Latin America)
Implementation period: 2025
LLM Pulse modules used: Brand monitoring, competitor tracking, sentiment analysis, citation tracking, share of voice, reputation scoring
Start your AI visibility transformation
Softonic’s journey from AI invisibility to category leadership demonstrates what’s possible when brands take control of their presence in the new search landscape.
LLM Pulse provides the visibility, insights, and tracking capabilities that made this transformation possible.
Ready to see where your brand stands?