What do our AI answer industry benchmarks show?
Broadcastwell measures and fixes whether AI search engines name B2B software companies.
Run the free checkOur AI answer industry benchmarks set fourteen B2B software categories side by side from one dated sample: 286 checked vendors, 2,095 valid answers captured on 22 and 23 September 2026, ten fixed buyer questions per category and three runs on each of five engines. Naming concentration ran from 30.8% in transportation management to 67.0% in CRM.
Updated 9 November 2026.
The five engines are ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode. Every figure below is published in the September release of the Absence Index or in the source tables of our report on it, with its count and base. The question bank and the checked vendor set are fixed, not a probability sample of buyers or vendors, and the repeats can be dependent. Read a row as what the engines answered to those questions on those dates.
The fourteen categories, September 2026
| Category | Checked vendors | Valid answers | Named most (count, share, 95% interval) | Top three share of names | Answers with no cited entry | Vendor-site share of citations |
|---|---|---|---|---|---|---|
| Construction software | 22 | 150 | Procore, 109 of 150, 72.7% (65.0% to 79.2%) | 48.5% (44.3% to 52.6%) | 2 of 150 | 46.9% |
| CRM software | 10 | 150 | HubSpot CRM, 113 of 150, 75.3% (67.9% to 81.5%) | 67.0% (62.7% to 71.1%) | 6 of 150 | 23.9% |
| Dealership software | 22 | 150 | CDK Global, 90 of 150, 60.0% (52.0% to 67.5%) | 38.7% (35.0% to 42.6%) | 2 of 150 | 21.7% |
| Dental software | 20 | 149 | Dentrix, 120 of 149, 80.5% (73.4% to 86.1%) | 47.6% (44.1% to 51.2%) | 2 of 149 | 28.0% |
| Field service software | 22 | 150 | Housecall Pro, 108 of 150, 72.0% (64.3% to 78.6%) | 46.4% (42.3% to 50.4%) | 9 of 150 | 43.9% |
| Hotel software | 25 | 149 | Cloudbeds, 115 of 149, 77.2% (69.8% to 83.2%) | 42.0% (38.0% to 46.2%) | 2 of 149 | 39.1% |
| Insurance agency software | 19 | 150 | Applied Systems, 127 of 150, 84.7% (78.0% to 89.6%) | 55.3% (51.3% to 59.2%) | 1 of 150 | 23.4% |
| Legal practice software | 21 | 150 | Clio, 126 of 150, 84.0% (77.3% to 89.0%) | 47.8% (43.9% to 51.8%) | 1 of 150 | 26.2% |
| Loan origination software | 25 | 150 | nCino, 92 of 150, 61.3% (53.3% to 68.8%) | 42.0% (38.0% to 46.2%) | 0 of 150 | 27.2% |
| Project management software | 10 | 150 | Asana, 97 of 150, 64.7% (56.7% to 71.9%) | 54.2% (49.5% to 58.8%) | 5 of 150 | 23.6% |
| Property management software | 22 | 148 | Buildium, 119 of 148, 80.4% (73.3% to 86.0%) | 48.7% (44.7% to 52.7%) | 3 of 148 | 49.0% |
| Radiology software | 24 | 150 | Sectra, 87 of 150, 58.0% (50.0% to 65.6%) | 40.5% (36.5% to 44.6%) | 1 of 150 | 22.1% |
| Transportation software | 25 | 150 | AscendTMS, 64 of 150, 42.7% (35.0% to 50.7%) | 30.8% (27.1% to 34.9%) | 4 of 150 | 22.2% |
| Veterinary software | 19 | 149 | ezyVet, 115 of 149, 77.2% (69.8% to 83.2%) | 44.9% (41.0% to 48.9%) | 1 of 149 | 30.8% |
| All fourteen | 286 | 2,095 | not pooled | not pooled | 39 of 2,095 | 30.5% |
The top three share is the three most-named vendors' names over all checked-vendor names in that category, from our vendor concentration table; an answer can name several vendors. The vendor-site share uses the post-audit source classes over every cited entry. We do not pool the most-named vendor across categories, because each category has its own checked set and its own questions.
What the spread tells a buyer
Measured. Vendor sites carried 49.0% of citations in property management software answers and 21.7% in dealership software. The most answers without any cited entry came in field service software: 9 of 150. Five target attempts were excluded across all categories, every one a Google AI Overviews attempt, and excluded attempts leave both the count and the base.
Reasoned. A category where a few vendors take most of the names is harder to enter from outside that group, and a category where vendor sites carry most citations puts more weight on a vendor's own pages. Neither pattern says why an engine answered as it did, and neither predicts the next capture. Our registered persistence study waits for comparable releases before comparing standings over time; see does low presence persist.
Where to go from a row
Each category name opens our service page for that industry, with its two quoted buyer questions, the engine-by-engine counts and what each offer would do there. The Index keeps the full category records with every question and vendor, and our industry answers explain absences and cited hosts for each category. For one engine at a time, start with our page on ChatGPT visibility.
Sources
- Broadcastwell, The Absence Index, release 2026-09, release record, DOI 10.5281/zenodo.22907695; answers captured 22 and 23 September 2026, published 23 September 2026, read 9 November 2026.
- Broadcastwell, answer citation bases, post-audit source classes, vendor concentration and excluded answers, report DOI 10.5281/zenodo.23804758, read 9 November 2026.
- Broadcastwell, Methodology, method v1.1, read 9 November 2026.
Related: our insight on concentrated visibility discusses what a high share means, and our service model lists the offers each row links to.