The Shortlist Gap: who AI engines name when buyers ask, and who they never name

Broadcastwell Research · Study 1 · 26 Sep 2026 · CC BY 4.0 · DOI 10.5281/zenodo.22982426

2,095 AI answers to 140 buyer questions, five engines, three runs each. One vendor in six was never named. The pages behind the shortlists are mostly written by vendors.

Key finding

16.1% of the 286 checked vendors were never named in any of the 2,095 answers (46 vendors). 138 (48.3%) appeared in fewer than one answer in ten. Only 35 (12.2%) appeared in at least half.

The September Index figure (16.1% of 286 vendors, five engines, three runs) uses a different sample from Volume I's 35% (85 companies, one engine). These are different samples, not a trend.

What we measured

The Absence Index release 2026-09 (DOI 10.5281/zenodo.22907695): 14 categories of business software, 286 checked vendors from buyer directories, ten buyer questions per category, five engines (ChatGPT, Claude, Perplexity, Google AI Overviews, Google AI Mode), three scheduled runs per engine and question, captured 22 to 23 Sep 2026 (UTC). Every figure is recomputed from the raw answers with the published matching rule; the recompute reproduces the published release exactly.

Each engine is narrower than the five together

ChatGPT never named 115 of the 286 vendors (40.2%); Google AI Overviews 103 (36.0%); Google AI Mode 101 (35.3%); Perplexity 85 (29.7%); Claude 66 (23.1%). 132 vendors were named by all five; 19 by exactly one.

Shortlists are short

An answer names 3.86 checked vendors on average. The top vendor takes 12.0% to 23.7% of all namings in its category. Alternative questions name 6.30 vendors per answer; buyer-evaluation questions name 1.67 and leave 65.7% of vendors unnamed.

The sources are mostly vendor-written

Of the 140 top-ten source slots (ten per category, 109 distinct domains), 86 (61.4%) are vendor-owned and 77 (55.0%) are vendor-written list or comparison pages. Review sites and directories hold 30 (21.4%). 34 slots belong to vendors outside the checked list. The most-cited domain is a checked vendor in only 6 of 14 categories.

Per category

• Construction project management software: 4 of 22 checked vendors were never named in 150 answers; 3 appeared in at least half; Procore led with 72.7% of answers.
• Dealership management systems: 2 of 22 checked vendors were never named in 150 answers; 3 appeared in at least half; CDK Global led with 60.0% of answers.
• Dental practice management software: 4 of 20 checked vendors were never named in 149 answers; 4 appeared in at least half; Dentrix led with 80.5% of answers.
• Field service management software: 7 of 22 checked vendors were never named in 150 answers; 2 appeared in at least half; Housecall Pro led with 72.0% of answers.
• Hotel property management software: 5 of 25 checked vendors were never named in 149 answers; 1 appeared in at least half; Cloudbeds led with 77.2% of answers.
• Insurance agency management systems: 5 of 19 checked vendors were never named in 150 answers; 4 appeared in at least half; Applied Systems led with 84.7% of answers.
• Legal practice management software: 3 of 21 checked vendors were never named in 150 answers; 4 appeared in at least half; Clio led with 84.0% of answers.
• Loan origination and lending software: 4 of 25 checked vendors were never named in 150 answers; 2 appeared in at least half; nCino led with 61.3% of answers.
• Radiology and medical imaging software: 2 of 24 checked vendors were never named in 150 answers; 2 appeared in at least half; Sectra led with 58.0% of answers.
• Residential property management software: 3 of 22 checked vendors were never named in 148 answers; 3 appeared in at least half; Buildium led with 80.4% of answers.
• Transportation management systems: 3 of 25 checked vendors were never named in 150 answers; 0 appeared in at least half; AscendTMS led with 42.7% of answers.
• Veterinary practice management software: 3 of 19 checked vendors were never named in 149 answers; 2 appeared in at least half; ezyVet led with 77.2% of answers.
• CRM: 1 of 10 checked vendors were never named in 150 answers; 3 appeared in at least half; HubSpot CRM led with 75.3% of answers.
• Project management: 0 of 10 checked vendors were never named in 150 answers; 2 appeared in at least half; Asana led with 64.7% of answers.

Method

Matching rule as published (method v1.1): names and listed aliases match case-insensitive word boundaries; plain-word vendors need a listed qualifier within five words in the same sentence or their canonical domain in the cited URLs; Wilson 95% intervals. Broadcastwell's published method (broadcastwell.com/methodology, version 1.1, 8 September 2026) states: "The measurement is single turn, API based, logged out, US English, gl=us and hl=en" and "It is a controlled instrument, not a replica of any one buyer's screen." Scripts: recompute_index.py with vendor_rules_2026-09.json, study1_shortlist_gap.py, study1_sources_classified.py (in the dataset).

Limitations

One release on two UTC dates; fixed question set; vendor lists from buyer directories (unchecked vendors are not measured); word-boundary matching with a qualifier rule for plain-word names; three runs per cell; hand classification of 138 source domains; no causal claim; engines change without notice; the publisher sells in this market and ranks no vendor.

Dataset and citation

Dataset (CSV, README, scripts, CC BY 4.0): https://doi.org/10.5281/zenodo.22982426 PDF: https://broadcastwell.com/files/research/study1/study1_the_shortlist_gap_2026-09.pdf Cite: Broadcastwell Research. (2026). The Shortlist Gap: which of 286 software vendors AI engines name when buyers ask, across 14 categories and five engines (Version 1.0) [Data set and report]. Broadcastwell. https://doi.org/10.5281/zenodo.22982426