Abhord Quickstart Guide (March 2026 Refresh)
This refreshed edition adds clearer workflows, stronger model coverage guidance, and practical playbooks based on what’s changed in generative search behavior over the last year.
What’s new in this edition
- Expanded model coverage guidance: design surveys to compare multiple leading closed and open LLMs, plus regional/language variants. Weight results by model popularity where relevant.
- Mentions intelligence upgrades: improved brand disambiguation (e.g., Apple vs. “apple”), URL/domain canonicalization, and deduping across near-identical responses.
- Sentiment and stance: go beyond “positive/neutral/negative” with aspect-level tags (pricing, support, performance) and “stance” for vs. queries.
- Share of Voice (SOV) normalization: recommended SOV uses weighted sampling and intent clustering for more stable trend lines.
- Action layer: stronger handoffs to content, SEO, comms, and product marketing via exports, alerts, and issue workflows.
1) Initial setup and configuration
- Create your workspace
- Add your organization and invite collaborators. Set roles (Admin, Analyst, Stakeholder) to control who can edit surveys, entity rules, and alerts.
- Define entities
- Brand entities: your company, products, and key URLs (homepage, docs, pricing, comparison pages).
- Variants and synonyms: add common misspellings, ticker symbols, and shorthand (e.g., Abhord, Abhord.ai).
- Competitors: list primary and secondary competitors with variants.
- Disambiguation rules
- Positive matches: domains, canonical names, product SKU codes.
- Negative matches: words that should exclude a match (e.g., “apple fruit,” “sage herb”).
- Tip: Start strict, then relax as you validate early results.
- Locales and personas
- Choose markets (country/language) and, if relevant, audience personas (e.g., SMB buyer vs. enterprise architect). Locales matter: different LLMs rank/recall brands differently