Lookelo audience-question sample
Research completed September 12, 2026; published September 13, 2026.

Window: August 14-September 12, 2026, inclusive. Reddit timestamps were converted to America/New_York dates.

Discovery: 120 candidates from capped Reddit searches in GoogleMyBusiness, smallbusiness, SEO and localseo (30 newest results per community, service-related terms, month filter), plus 12 web-discovered discussions verified through the Reddit API. Search caps and query choices affect which questions enter the sample; searches were not exhaustively paginated.

Screening: 132 candidates; 76 eligible observations retained. Promotions, rhetorical marketing posts, general business questions outside the selected services, and unverified questions were excluded. LinkedIn and X public discovery was attempted but produced no usable verified dated questions for counting. All counted evidence is Reddit. Participants span countries, industries, owners and practitioners; this is not a representative survey of New England businesses.

Counting: assign each post one primary question cluster. Count a known asker once within that cluster, even across threads. Rank by independent ask count, then platform breadth, then most recent question. No likes or engagement metrics enter the ranking. Broadly related problems, such as appearing in different AI engines versus measuring AI visibility, remain separate clusters.

Dataset: /research/2026-09-question-sample.csv contains all 76 eligible observations and source permalinks. Count distinct independent_ask_unit values within each question_cluster to reproduce frequencies. Unit identifiers are anonymous and unique only within each cluster. They do not connect participants across clusters.

Top five: verification 8 independent asks from 9 posts; Maps ranking 8; Google traffic drop 6; AI measurement 5; SEO service value 4. Verification wins the tied count on recency. The five topics contain 32 posts representing 31 independent asks within clusters.

Limitations: a bounded convenience sample, not internet-wide question frequency, search volume, a causal study, or an estimate of customer prevalence. Posts establish what people asked, not the truth of their reported results or suspected causes. Editorial clustering involves judgment. Numerical business and tracking scenarios in the accompanying articles are explicitly hypothetical, not measured client outcomes.