The number we keep encountering in international studies is 78%: that's the share of businesses invisible in AI responses when their prospects search for a provider. But international studies are dominated by English-speaking, often tech-sector companies.
What does the picture look like for French SMEs across diverse B2B sectors?
We tested 50 of them in ChatGPT and Perplexity, systematically, with a reproducible protocol. Here's what we found.
Methodology
Company selection
The 50 companies were selected to represent the diversity of French B2B SMEs:
- 10 marketing and communications agencies
- 10 management and strategy consulting firms
- 10 HR and recruitment agencies
- 10 IT and digital service companies
- 10 legal and accounting firms
All companies had:
- An active website
- A LinkedIn company page
- At least 5 years of activity
- 10 to 200 employees
No company was selected based on known AI visibility — this was a blind sample from publicly listed B2B companies.
Test protocol
For each company, we designed 20 sector-specific prompts covering:
- Generic sector discovery ("which [type] in France do you recommend?")
- Regional discovery ("best [type] in [region/city]?")
- Problem-based queries ("I need help with [specific problem], who should I contact?")
- Comparison queries ("which are the recognised [type] in France?")
- Direct verification ("what is [company name]?")
Each prompt was tested on ChatGPT (GPT-4o with web search) and Perplexity Pro. Tests were conducted between March 15 and April 10, 2026, in private browsing mode.
Scoring
Each company received:
- A presence score: % of prompts where they appeared at least once
- A position score: average citation position (1st, 2nd, 3rd+)
- A cross-platform score: whether they appeared on both ChatGPT and Perplexity
- A sentiment score: % of citations that were clearly positive vs neutral/negative
Overall results
The headline number: 76% invisible
38 out of 50 companies (76%) had a presence score of 0% on discovery and comparison prompts. They appeared only when their company name was typed directly — the verification prompt.
This 76% figure is consistent with international data (78% in the Omni Eclipse study, 95.7% invisible on discovery queries in the 2X AI Visibility Index). The French B2B SME market isn't an exception.
The visible 24%: 12 companies
12 companies appeared on at least one discovery or comparison prompt. Of these:
- 4 appeared consistently (5+ prompts, both platforms)
- 5 appeared occasionally (1-4 prompts, usually one platform)
- 3 appeared on verification-adjacent prompts only (their name mentioned in comparative context)
ChatGPT vs Perplexity: divergent results
Consistent with the 11% cross-platform overlap data from the 680-million-citation analysis:
- 7 companies appeared on ChatGPT but not Perplexity
- 3 companies appeared on Perplexity but not ChatGPT
- 2 companies appeared on both
No company with zero ChatGPT presence had strong Perplexity presence, or vice versa — but the converse wasn't true either. Optimising for one platform doesn't automatically deliver the other.
Sector variation
Significant differences emerged by sector:
| Sector | Visible companies | Notes | |---|---|---| | IT / Digital services | 5/10 | Highest visibility — strong review platform presence | | Marketing / Comms agencies | 3/10 | Moderate — LinkedIn-driven | | Management consulting | 2/10 | Low — weak external sources | | HR / Recruitment | 1/10 | Very low — sector under-indexed in LLMs | | Legal / Accounting | 1/10 | Very low — regulated sector, limited content marketing |
The IT and digital services sector's outperformance is explained by natural presence on G2, Capterra and specialised tech forums — the source types LLMs most reliably cite for service recommendations.
What the 12 visible companies do differently
Analysing the 12 visible companies against the 38 invisible ones, five patterns distinguish them with high consistency.
Pattern 1 — Third-party review platform presence (100% of consistently visible companies)
Every company that appeared consistently had a complete profile on at least one of: G2, Capterra, Clutch, or Malt. All four consistently visible companies had profiles on 2+ platforms.
Of the 38 invisible companies, only 6 had any review platform presence — and those 6 had profiles with fewer than 5 reviews, all more than 18 months old.
The correlation is near-total. This is the single most predictive signal in our dataset.
Pattern 2 — Specific positioning language (visible in 9/12 visible companies)
Visible companies described their offer using specific, functional language. Not "we support companies in their transformation" but "we specialise in ERP implementation for manufacturing companies with 50-500 employees".
LLMs need to match a company to a specific query. Generic positioning matches nothing specifically — specific positioning matches precisely.
Pattern 3 — Recent LinkedIn activity (visible in 10/12 visible companies)
10 of the 12 visible companies had published substantive LinkedIn content in the last 30 days. By contrast, 31 of the 38 invisible companies had either no LinkedIn company page or no post in the last 90 days.
LinkedIn's rise to the #1 B2B citation source in ChatGPT Search (14.3% of responses, Semrush data) is visible in our dataset — companies with active LinkedIn pages systematically outperformed inactive ones, even controlling for other factors.
Pattern 4 — External editorial presence (visible in 8/12 visible companies)
8 visible companies had at least one article in a sector publication, newsletter, or recognised media outlet that mentioned them in context (not a press release, an earned mention). For Perplexity specifically, this was the most differentiating factor — all 5 companies with Perplexity presence had this external editorial anchor.
Pattern 5 — Schema.org structured data (visible in 7/12 visible companies)
7 of the 12 visible companies had at least Organization Schema.org implemented on their homepage. Of the 38 invisible companies, only 4 had any structured data beyond basic meta tags.
Sector deep-dive: why HR and legal firms underperform
The near-total invisibility of HR/recruitment and legal/accounting firms in our dataset deserves specific attention — these are large and commercially significant B2B sectors.
Why HR and recruitment firms are invisible: The sector is fundamentally relationship-driven and has historically underinvested in content marketing. Most HR firms' digital presence is a listing-focused website and a job board — neither of which generates the editorial or community mentions that LLMs cite. The exception would be firms that have published salary surveys or sector reports — content that gets cited externally.
Why legal and accounting firms are invisible: Regulatory constraints on advertising and professional communications create a structural inhibitor. Many firms can't or won't publish commercial content. However, firms that publish substantive analysis (legal commentary, tax reform implications, compliance guides) are present in LLM training data through bar association publications and legal media — they just choose not to create this content actively.
For both sectors, the opportunity is real but requires a content strategy shift: away from service promotion and toward genuinely informational content that gets picked up externally.
What the best-performing company did
The single highest-scoring company in our panel — a Paris-based IT services firm specialising in cloud migration for mid-market companies — exemplifies what works:
- Complete G2 and Capterra profiles with 23 and 17 reviews respectively, average dates within 8 months
- LinkedIn company page with 3.2k followers, 3-4 posts per week including original data and client results
- Listed in two sector comparison articles from the previous 6 months
- Clear, specific positioning: "cloud migration specialists for 200-1,000 employee French companies"
- Organization + FAQPage + Article Schema.org across all pages
- Bing Webmaster Tools set up with IndexNow active
This isn't a company that did something extraordinary. It's a company that did all the right things consistently — and its AI visibility reflects exactly that.
What this means if you're invisible today
If you're in the 76% — which is statistically likely — the data gives clear direction.
Start with the two highest-correlation signals from our dataset:
1. Create or complete a review platform profile. G2 if you're in software or digital services, Clutch if you're an agency, Malt if you're a consultant. Complete it fully, add your key specialisation, and ask 5 recent clients for reviews in the next 30 days.
2. Publish substantive LinkedIn content for 30 consecutive days. Not promotional — one insight, one data point, one client result per post. This is the fastest way to build the entity signal that LLMs are now explicitly using.
These two actions, done consistently, are what separated visible from invisible companies in our dataset more than any other factor.
Methodology: 50 French B2B SMEs across 5 sectors, 20 prompts per company, tested on ChatGPT GPT-4o with web search and Perplexity Pro, March-April 2026. Companies are anonymised. Full methodology available on request.
Our free scoring tool lets you test your own visibility across all 5 major AI engines. For a complete competitive benchmark, see our AI Diagnostic.