What we collect, how we resolve entities, how we score confidence, and where our claims come from. If you're evaluating Recon for procurement, this is the page to read first.
Last updated: August 16, 2026
Every source Recon draws on is publicly available. Nothing gated, nothing purchased from consumer databases, nothing intercepted.
LinkedIn, Instagram, Twitter/X, TikTok, Facebook, YouTube, and Reddit. We collect public profiles and public posts only. No private accounts, no direct messages, no gated content.
Google Maps, Crunchbase, SEC EDGAR where applicable, state business registries, and Facebook Ad Library metadata. All entries used are publicly listed or filed with public agencies.
Google Reviews, Yelp, Trustpilot, and industry-specific directories. We read what any prospective customer of the subject would read.
News, press releases, company blogs, and industry publications indexed by public search engines and RSS feeds.
On enterprise engagements only, and only with explicit written authorization from the client, we can incorporate publicly-available breach-exposure intelligence into a report. This is never used on the standard Recon SKUs.
A business surfaces on 6+ platforms with different handles, variant names, and inconsistent addresses. Getting the resolution right is where most category tools quietly fail.
Every match is scored HIGH / MEDIUM / LOW. Low-confidence matches are flagged inside the report — never silently promoted into a "confirmed" section. If we can't resolve an entity with confidence, we say so.
Recon surfaces the real market competitors — the businesses actually winning the same conversations — not the category tags a database would return.
The narrative layer answers "what's moving in this vertical right now, and does it apply to this subject?"
The killer feature. Anyone can scrape mentions; the hard part is separating "someone tweeted your keyword" from "someone is about to buy this service and asking who to hire."
Every lead is assigned a Buying Intent tier — HIGH / MEDIUM / LOW — with the reasoning inline. Analysts can see exactly why a lead was scored the way it was, and so can the client.
The single most important design decision in Recon. Buyers, journalists, and procurement teams need to know instantly what's observed fact vs analyst inference vs opinion.
Direct data extracted from a public source. Source URL provided in every case.
Derived from observable proxy signals — e.g., posting cadence estimated from a 30-day sample.
Pattern extrapolation from multiple weak signals. Used sparingly, always flagged.
Analyst opinion or next-action guidance. Clearly labeled as opinion, not fact.
Clarity beats overselling. Here are the lines we don't cross.
Automation surfaces candidates. Analysts approve. Nothing goes to a client unreviewed.
2-hour SLA. Single-analyst review across the 3-tab dashboard before delivery.
4-hour SLA. Senior-analyst review on the Sales Intelligence and Gap Analysis tabs, plus base analyst review on the rest.
Weekly refresh cycle with change-detection callouts vs the prior report. Analyst review every cycle.
Every report is reviewed by a human before delivery. Automation surfaces candidate findings, confidence tags, and lead scores. Analysts approve, adjust, and write the narrative sections.
If it's in the report, you can verify it.
Every quantitative claim in every Recon report includes a source citation — a URL, a timestamp, or "analyst inference from N observations." Clients can independently verify anything we assert.
If you receive a report where a specific claim is not attributable, that's a bug. Email info@dataflame.ai and we'll fix it.