Abhord Product Guide (Refreshed March 2026)

This practical guide helps new Abhord users get from setup to insight, fast. It reflects the latest LLM landscape and adds updated recommendations for sampling, confidence handling, and competitor hygiene.

What’s new in this edition

• Emphasis on model variance: compare across models with consistent sampling windows and show confidence intervals when possible.
• Updated sampling guidance: start with small pilot runs, then scale using balanced or traffic-weighted allocations.
• Clearer entity hygiene: maintain canonical names plus aliases and disambiguation rules to reduce false mentions.
• Action loops: connect insights to tickets/roadmaps and schedule follow-up surveys to measure lift.

1) Initial setup and configuration

1) Create your workspace

• Set organization name and default timezone/locale.
• Add teammates with roles: Admin (settings/billing), Analyst (create surveys), Viewer (read-only).

2) Define your entities

• Brand entity (canonical): e.g., “Abhord.”
• Aliases: product names, tickers, common misspellings.
• Competitors (see section 4): each with its own alias list.
• Disambiguation rules: exclude generic terms that can inflate mentions (e.g., “flow,” “studio”).

3) Choose your Model Set

• Balanced: a representative mix of top general-purpose LLMs for broad coverage.
• Fast/Cost–optimized: for iterative testing or high-frequency tracking.
• Enterprise/High-accuracy: slower but better factuality for sensitive audits.

Tip: Keep your chosen set stable for at least one full reporting cycle to minimize model-drift noise.

4) Configure data retention and privacy

• Select retention window appropriate to your compliance needs.
• Mask PII by default; only unmask for specific diagnostic runs.

5) Add integrations

• Webhooks or API for exporting results to BI tools.
• Alert channels (email, Slack/Teams) for spikes or anomalies.
• Optional: backlog tools (Jira/Trello) to operationalize insights.

6) Establish naming/versioning

• Survey naming: purpose-audience-date (e.g., “SOV-US-Balanced-2026-03”).
• Version every change to prompts, model sets, and thresholds.

2) Running your first survey across LLMs

1) Clarify your objective

• Example: “Measure share of voice for Abhord vs. top competitors when users ask LLMs about GEO/AEO platforms.”

2) Draft the