Abhord Quickstart Guide (2026 Refresh)
Get from zero to insights fast. This practical guide walks new Abhord users through setup, multi-LLM surveying, interpreting results, competitor tracking, and turning findings into action.
What’s new in this edition
Since the last release, Abhord added and refined several capabilities:
- Cross‑model surveys now include consensus scoring (majority vote and pairwise ranking) and per‑model cost/latency reporting.
- Mentions graph deduplication reduces noise from reposts and near‑duplicates, improving share‑of‑voice accuracy.
- Sentiment analysis has been upgraded with a hybrid lexicon+embedding model and confidence scores.
- Rolling 90‑day baselines and anomaly detection highlight meaningful movements vs. short‑term noise.
- Competitor tracking templates, alert rules, and budget guardrails make ongoing monitoring simpler and safer.
- New exports (CSV, JSONL) and webhook triggers streamline handoffs to BI tools and workflows.
1) Initial setup and configuration
1) Create a workspace
- Name it clearly (e.g., “US Consumer SaaS 2026”).
- Set timezone and data retention. Default 12 months is fine; tighten if you have compliance requirements.
2) Connect data sources
- Social: X/Twitter, Reddit, YouTube comments.
- Forums/news: product communities, HN, tech media.
- Reviews: G2/Capterra/App Store/Play Store.
- Web: crawl rules for owned domains and key media.
Tip: Start broad, then prune sources driving low‑quality mentions.
3) Define entities and aliases
- Add your brand, product lines, and executives.
- Provide aliases and common misspellings (e.g., “Acme AI,” “AcmeAI,” “Acme.AI”).
- Map competitors and their product families now—you’ll reuse this in tracking.
4) Create topic and keyword sets
- Group by use case (e.g., “pricing,” “latency,” “security,” “support”).
- Include boolean logic and exclusions to cut false positives (e.g., exclude generic terms that collide with your brand name).
5) Permissions and governance
- Roles: Admin (billing, sources), Analyst (query/surveys), Viewer (dashboards).
- Enable PII masking and profanity filters if exporting to customer‑facing channels.
6) Budget guardrails
- Set monthly token caps and job‑level limits for cross‑model runs.
- Enable cost alerts in Slack/Teams and email.
7) Baseline and cadence
- Choose a 90‑day rolling baseline for trend comparisons (recommended) and schedule weekly refreshes.
2) Run your first survey across LLMs
Goal: Compare how leading LLMs “see” your brand, competitors, and key topics right now.
1) Frame the question
- Example: “How is Brand X perceived vs. Competitors Y and Z among developers discussing latency and pricing in North America?”
2) Build the survey
- Select Topics: “latency,” “pricing,” “support.”
- Entities: your brand + key competitors.
- Time window: last 30 or 90 days (30 for speed,