How to Build a Prospect Research Workflow (Without Losing Your Mind)

Let's do the maths that no one in outbound wants to do out loud.

A reasonably thorough research job on a B2B prospect — LinkedIn check, recent company news, a skim of their job postings, a poke around their tech stack — takes 25 to 35 minutes. Call it 30 on average. You're targeting 50 prospects per week. That's 25 hours. A full work week. Spent on research before a single email gets written.

This is why most teams don't do proper research. Not because they're lazy. Because the maths don't work. And so they send generic emails, wonder why reply rates are 0.5%, and conclude that cold outreach "doesn't work anymore."

Cold outreach works fine. Sending the same email to 500 people who you know nothing about doesn't work. These are different problems.

The 5 Signals That Actually Matter

Most B2B prospect research is unfocused. People open LinkedIn, start reading, and 20 minutes later they've learned interesting but irrelevant things about someone's career arc. The fix is to decide in advance what you're looking for.

There are five signals that consistently produce usable email hooks:

1. Role changes. A new VP of Sales who started 3 months ago is under pressure to prove themselves. A new CTO inheriting technical debt wants to show they're modernising things. Fresh hires in decision-making roles have mandate and motivation. This is the highest-signal trigger in B2B outbound — and LinkedIn makes it free to find.

2. Funding events. Series A, B, a strategic raise, a debt facility. New money means new headcount, new tools, new urgency. Companies that just raised are in growth mode and buying. Companies that raised 18 months ago and haven't made noise since are in a different spot. Both are useful to know.

3. Product launches and announcements. A new product line, a market expansion, a rebrand. These create natural conversation starters and often signal strategic priorities. "Saw you launched X last month — what's driving the push into [market]?" is a better opener than "Hope this finds you well."

4. Hiring patterns. Job postings are underrated as a research signal. A company hiring 5 SDRs is scaling outbound. A company hiring a Head of Data is making analytics a priority. Hiring patterns tell you what a company is trying to build — often before they've told anyone publicly. Check their careers page. It takes two minutes and gives you something real to work with.

5. Tech stack. Tools like BuiltWith and G2 reviews surface what software a company is running. If they're on a legacy CRM you integrate with, that's relevant. If they're using a competitor's product, that's also relevant — just differently. Tech stack tells you who they're already paying and what problems they've already decided to solve.

Manual Workflow vs. AI-Augmented Workflow

Here's what the same 5-signal research job looks like with a human doing it alone versus with AI doing the data collection:

Total manual: 36–56 minutes per prospect. Total with AI-augmented research: 8–12 minutes. The time savings aren't in the creative work — they're in the data collection that precedes it.

Where AI Genuinely Helps

AI is good at one thing in a sales research workflow: finding and surfacing data at speed. Recent press releases, LinkedIn activity, job postings, funding announcements — these are structured data sources with clear patterns. AI can scrape, parse, and summarise them in seconds. The human equivalent is opening twelve tabs and reading twelve pages. Same information. Twenty minutes more.

AI is also good at first-draft generation once the research is done. Take 5 signals, apply a persona, produce a draft. Not the final email — the working draft that the human then edits, adjusts, or throws away entirely. The draft generation saves 8–10 minutes. The human judgment that reviews it is still non-negotiable.

Where Humans Still Earn Their Salary

Two things that AI consistently gets wrong:

Relevance filtering. AI can tell you that a company just raised £4M Series A. It cannot tell you whether that's a reason to reach out right now or whether the founding team is going to be head-down for six months integrating new hires and has zero bandwidth to evaluate new software. That judgment comes from context, experience, and a read on the specific company. Humans have it. Models approximate it badly.

Tone calibration. AI drafts emails. It does not know that the VP of Engineering you're emailing has been visibly frustrated on Twitter about a specific category of vendor. It doesn't know that her company has had two failed implementations of similar tools and she's allergic to the category. These are things you learn from actually reading. AI pattern-matches. Humans contextualise.

The correct workflow treats AI as a research associate and first-draft engine — not as a replacement for the judgment call at the end. You still decide whether the signal is compelling. You still decide whether the draft is right. You're just not spending 35 minutes gathering data you could have had in 30 seconds.

That's the whole thing. That's the workflow. Drumroll automates the data collection layer — role changes, funding, product news, hiring patterns, tech stack — and produces a reviewed draft ready for your approval. The human spends 10 minutes on what used to take an hour. Fewer emails, more relevant, better results.

The goal was never to send more. It was always to send better.