Research prospects before outreach: find the one wedge that earns a reply.
Good research produces a first line only that prospect could receive. Two or three verified facts, not a dossier.
What to research (and what to skip)
You need the facts that make outreach relevant, not a dossier:
- The role: what they own and are measured on; the wedge differs per role.
- Account signals: job postings beat intent scores: 11 open SDR roles is spend committed.
- Timing: funding, launch, leadership change; freshest in its first days.
- The wedge: one verifiable link between their reality and your offer.
Turn research into the message
The test: could this exact message have gone to anyone else? Then rewrite:
- Open with their trigger: two sentences about them before one about you.
- Trace every claim: fake research ('loved your post!' with no post) converts worse than none.
- One specific ask: tied to the trigger, with a credible reason to talk this week.
Five accounts by hand, five hundred by machine
At 15-20 minutes per contact, manual research collapses and outreach regresses to templates:
- The research layer: reads 100+ sources per contact and hands the rep a ready brief.
- Drafts included: Salesgear pre-fills every sequence step from the findings, sources attached.
- The data underneath: 800M+ owned contacts at 95% direct-dial accuracy, so the brief comes with a person you can actually reach.
Why an LLM alone falls short
A model without live sources describes the company as it was, and fills gaps with confident fiction:
- Ground in live sources: fresh pages, filings, and posts in; training-data memory out.
- Demand traceability: every fact carries its source, checkable in seconds before send.
- Synthesize, not know: LLMs surface the angle from 50 sources; they are dangerous as the source.
- The other half: no LLM ships verified emails or direct dials; research needs a data layer.
Researching prospects, answered
How do I research prospects before outreach?
Work four layers: the person's role and what they are measured on, account signals (funding, job postings, leadership changes, launches), the timing trigger that makes now relevant, and the one verifiable wedge connecting their situation to your offer. Done by hand this takes 15-20 minutes per contact; an AI research layer does it in seconds across 100+ sources with citations.
What should I research about a prospect?
The highest-yield signals in order: open job postings (hiring is spend commitment), funding events, what the person has posted or been quoted on, recent product or leadership changes, and their tech stack. You need the two or three facts that produce a first line only this prospect could receive, not a dossier.
How do I research prospects at scale?
Manual research costs 15-20 minutes per contact, which collapses past a handful of accounts and is why most outreach regresses to templates. A research layer that reads 100+ sources per contact automatically, drafts the message from the findings, and attaches sources gives every contact in a 500-person sequence the same depth the top five used to get.
Can I just use an LLM to research prospects?
Not on its own. A model working from training data is months stale and invents plausible details, and one fabricated fact costs the reply. Reliable research pipelines feed the model live sources and require every claim to be traceable. The model synthesizes; grounded data supplies the truth, and a data layer supplies the verified email and phone number.
How much research is enough before reaching out?
Enough for two things: a first line only this prospect could receive, and one credible reason you are reaching out this week. That is typically two or three verified facts, not a full profile. Depth past that point delays the send without raising the reply rate.
Research every prospect, not just the top five
Deep research across 100+ sources per contact gives every rep a ready brief, drawn from 800M+ owned contacts at 95% mobile direct-dial accuracy.
Explore deep research