Abhord Quickstart Guide (Refreshed — February 2026)
This practical guide walks new users from first setup to acting on insights. It reflects recent platform changes and updated recommendations based on the latest LLM behaviors.
What’s new since the last edition
- Expanded model coverage and easier multi-provider setup via Workspace Vault.
- Stronger entity resolution for cleaner “mentions” and de‑duplication across variants and misspellings.
- Share of Voice (SoV) normalization now adjusts for per‑model response volume automatically.
- Sentiment model updated for fewer neutrals and clearer polarity; confidence scores now exposed in the UI.
- Scheduled runs, webhooks, and Slack/Teams alerts for competitor spikes and sentiment shifts.
- Project-level budgets, cost caps, and audit trails to manage spend and access.
1) Initial setup and configuration
Goal: Create a reliable workspace that collects clean, comparable data at a predictable cost.
- Create your workspace
- Invite teammates and assign roles (Admin, Analyst, Viewer). Use SSO if your org has it enabled.
- Set data retention and PII redaction preferences before you ingest data.
- Connect model providers
- Add keys for the providers you plan to sample (e.g., OpenAI, Anthropic, Google, Mistral, Cohere).
- In Workspace Vault, scope keys to projects; enable per-provider rate limits and spend caps.
- Define entities and synonyms
- Add your brand, products, and competitors as entities.
- Provide synonyms and common misspellings (e.g., “Acme Pro,” “AcmePro,” “Acme Pro+”).
- Optional: add negative keywords to avoid false positives (e.g., “Apple” the fruit vs. the company).
- Create your first project
- Name, market/locale, and topic taxonomy (themes like Pricing, Setup, Reliability).
- Toggle Safe Mode if you’re in sensitive categories (health/finance/children).
Pro tips (updated):
- Use at least two providers for resilience; set per-model quotas to avoid a single-model bias.
- Start with a 30‑day retention window; increase only if compliance allows.
2) Running your first survey across LLMs
Goal: Collect comparable answers from multiple LLMs to see how they “speak” about your space.
- Define a clear objective
- Example: “Identify top‑of‑mind password managers for developers and reasons cited.”
- Translate this into a primary question and 2–3 follow‑ups.
- Author prompts with controls
- Primary: “Which password managers would you recommend to a software developer and why?”
-