Updated September 28, 2026 | 13 min read

I Ran the Deep Company Analyser on 12 Prospects (Here's What Converted)

Last quarter, I sent 200 cold emails using standard firmographic personalization. Company name, industry, headcount. The reply rate was 2.1%. Then I switched to pain-point-first messaging built from deep company analysis. Same list segment. Reply rate jumped to 8.4%.
That gap isn’t a fluke. It’s the difference between knowing about a company and knowing what keeps their buyers up at night. This isn’t about generic “deep analysis” in the BI or consulting sense. If you’re looking for unstructured data automation or competitive intelligence dashboards, wrong article. This is deep company analysis built specifically for B2B sales outreach.
In this post, I’ll walk you through exactly how I use the Deep Company Analyser (a Claude Skill built by lemlist) to extract pain points, buying triggers, competitive gaps, and real customer language from public sources, then turn all of it into sequences that get replies. I’ll cover the setup, the five outputs, where most people get company research wrong, and how to connect the whole loop to live campaigns inside lemlist (rated 4.6/5 on G2 from thousands of sales teams).

What is a deep company analyser

A Claude Skill is a reusable .md instruction file that gives Claude a specific research or writing workflow, more structured than a one-off prompt. Skills are folders of instructions, scripts, and resources that Claude loads dynamically to improve performance on specialized tasks. Skills teach Claude how to complete specific tasks in a repeatable way. (GitHub - anthropics/skills: Public repository for Agent Skills · GitHub)
The Deep Company Analyser is one of these skills. Feed it a company’s website, reviews, and case studies, and it returns ranked pain points, buying triggers, competitive positioning gaps, and the exact language buyers use to describe their problems. Deep company analysis is the practice of pulling a prospect’s real problems, priorities, and language from public sources before you write outreach.
Technically, it’s a Markdown (.md) file packed with instructions and examples, built by lemlist and available through the Claude Skills hub or GitHub. Load it into a Claude conversation once, and it stays available for every company you research after that.

Why surface-level prospect research kills reply rates

Most outbound starts with who to contact: company size, industry, title, revenue band. That tells you where to aim, but not what to say once you get there.
You’ve probably seen the result land in your own inbox. A message that could apply to any company in the same vertical gets treated like every other generic email, ignored. Firmographic data answers “is this account a fit?” It doesn’t answer “what does this person actually care about today?”
That’s the gap deep company analysis closes. When I tested this on a batch of 12 accounts last quarter, the ones that got pain-point-first openers replied at 4x the rate of the ones that got standard intros. Once you know a prospect’s specific pain points and the words their buyers actually use, you can open with a line that reads like it was written just for them, which is usually the difference between a skim and a reply.

Other ways to research a company (and where they fall short)

Before I get into the Deep Company Analyser setup, here’s the honest landscape. There’s more than one way to do company research for sales. I’ve tried most of them.
Manual research. Open the prospect’s website, read their blog, scan G2 reviews, check LinkedIn. This works well for a handful of high-value accounts. It falls apart at 20+ companies a week because the time cost is brutal, and the output isn’t structured, so you can’t reuse the framework across accounts.
Generic ChatGPT or Claude prompting. Paste a URL into a chat window and ask “what are this company’s pain points?” You’ll get something. But without a structured skill file guiding the output, the results tend to be surface-level summaries, not ranked pain points with verbatim customer language. I’ve found the output needs heavy editing to be usable in outreach.
Dedicated enrichment tools (Clay, Clearbit, Apollo). Great at firmographic and contact data. Most of them can also pull technographics and funding info. But they don’t extract pain points from reviews, map competitive positioning, or give you the buyer’s own words. They answer “who” and “where,” not “why would they reply.”
The Claude Skill approach. A structured .md file tells Claude exactly what to extract, in what order, and in what format. The output is consistent across companies, ranked by urgency, and ready to drop into a sequence. That’s why I use it. But I want to be clear: if you only prospect five accounts a month, manual research might be all you need.

Five outputs from a deep company analysis

The skill returns five structured sections, and each one gives you something concrete to work with in your next message.

Pain points ranked by frequency and intensity

Instead of a flat list, you get pain points grouped by how often they show up across sources and how urgently buyers describe them. That ranking tells you which problem earns the first line, and which ones wait for a follow-up.
Here’s what an anonymized output looks like in practice:
Quote Icon
Pain Point #1 (frequency: high | intensity: high)
“Onboarding new reps takes 6+ weeks and we lose pipeline every time”
Sources: G2 reviews (x4), careers page (x2), case study
Pain Point #2 (frequency: high | intensity: medium)
“Reporting is manual, so leadership doesn’t trust the forecast”
Sources: G2 reviews (x3), blog post, pricing page objection handler
That structure makes it obvious what to lead with. Pain point #1 gets the opening line. #2 goes in the follow-up.

Buying triggers and timing signals

These are events like funding rounds, leadership hires, hiring surges, or tech migrations that hint a company might be ready to buy. A pain point tells you what to say; a trigger tells you when to say it.

Verbatim customer language

The skill pulls exact phrases buyers use in reviews and testimonials to describe their own problems. Mirror that language in a subject line, and a prospect recognizes the problem faster because you’ve described it the way they would.

Competitive positioning map

This shows how the target company differentiates from alternatives, and where the gaps sit. A gap in a prospect’s current vendor is often your best entry point for a conversation.

Messaging raw material

Everything above feeds directly into copywriting. Pass the output into a skill like lemlist’s Copywriting First Touch, and you get sequences built on real research instead of guesses.

How to set up and run the deep company analyser in Claude

Setup takes a few minutes and breaks into four steps.
  1. Download the skill file. Grab the .md file from the lemlist Claude Skills hub or the GitHub repo. It’s a single file with no dependencies.
  2. Load it into Claude. Drag it into a conversation, or add it to a Claude Project on a Pro, Max, or Team plan so it stays loaded without re-uploading each time. Pro gives you at least 5x more usage per 5-hour session than Free. Max gives you 5x or 20x more usage per 5-hour session than Pro. (Plans & Pricing | Claude by Anthropic) If you’re analyzing several companies in one session, Max’s higher ceiling will save you from hitting limits mid-workflow.
  3. Provide the target company and source URLs. Give Claude the company’s website, G2 or Capterra page, and any case study links, then ask it to surface pain points, triggers, language, and positioning.
  4. Review and refine. If the output feels thin, usually because public sources are limited, feed in more URLs or ask Claude to dig deeper on one section.

How to extract pain points from websites, reviews, and case studies

The quality of your analysis depends heavily on where you point it. Different sources reveal different things.

Where to find pain points on a website

Not every page is equally useful. The pricing page often preempts objections buyers raise before purchase, which points straight at friction. Product pages list problems in an order that usually reflects what the company prioritizes. “Why us” or comparison pages show what they position against, hinting at perceived weaknesses. And blog content with recurring topics often traces back to recurring buyer concerns.

What reviews reveal about real buyer problems

G2, Capterra, and Trustpilot reviews are some of the richest sources available. Negative reviews surface pain points directly, and positive reviews show the language buyers use once a problem gets solved. I’ve found that G2 alone, which hosts millions of verified B2B software reviews, often contains enough signal to build a full pain-point ranking for a single account.

How to rank pain points by urgency

A simple gut check helps decide what to lead with: how often a pain point appears across sources, how urgently buyers describe it, and how closely it maps to what you actually sell. A pain point that scores high on all three earns your opening line; something frequent but low-intensity often fits a follow-up better.

How to identify buying triggers that signal when to reach out

Pain points tell you what to say. Triggers tell you when saying it actually lands.
New capital from a funding round often unlocks budget for new tools. A leadership change means new decision-makers tend to re-evaluate existing vendors. Hiring surges signal fast-scaling teams that usually need new infrastructure to support growth. And tech stack changes create migration windows where openness to alternatives is highest.
Timing matters here too. A funding round gives you a window of weeks or months, while a leadership change closes much faster, so frame urgency around the trigger itself rather than inventing urgency that isn’t there.
Tip: if you’re on lemlist, Intent Signal Agents can catch these events automatically and drop matching leads into a campaign the moment one fires. Once deployed, your signal agents continuously monitor buying intent across your ICP. Whether it’s hiring activity, fundraising, website visits, tech changes, or custom signals, they’ll automatically detect the right moments to reach out. (lemlist Intent Signals: Spot buying clues & outreach at the best time)

How to use customer language in cold outreach

Prospects recognize their own problem faster when you describe it the way they would, not the way a marketing team would. That’s why verbatim language from the analysis belongs in your highest-visibility copy.
Element
Generic version
Customer-language version
Subject line
“Improve your sales process”
“Tired of reps losing half their day to data entry?”
Opening line
“We help companies grow revenue”
“Your G2 reviewers keep flagging slow onboarding for new AEs”
The difference comes down to specificity. Anyone could have written the generic version. The customer-language version proves you did the research.

How to map competitive positioning for sharper outreach angles

The positioning section shows how a target company differentiates from alternatives, and where the gaps sit. Your job is finding the angle their current vendor doesn’t cover.
Here’s how I think about it. Say the analysis surfaces that a prospect’s G2 reviews repeatedly mention clunky integrations with their current CRM tool. I don’t name the competitor in my outreach. Instead, I reference the gap itself: “I noticed your team keeps running into integration friction between your outreach tool and HubSpot.” That steps into a conversation the prospect is already having, rather than starting a new one.
This works best when you reference a specific, documented gap instead of a generic claim. The Deep Company Analyser gives you the raw material; your job is to frame it as a question or observation, not an attack on their current vendor.

How to connect deep company research to live campaigns

Research only pays off once it reaches the prospect. Here’s how the pieces link together.

Feed research into copywriting skills

Pass the analysis output into lemlist’s Copywriting First Touch, VP Sequence, or Follow-Up skills. Every message that comes out reflects real company intelligence instead of a generic template.

Build sequences around ranked pain points

Structure multichannel sequences (email, LinkedIn, calls) with your highest-frequency, highest-intensity pain point leading the first touch, and save secondary points for follow-ups.

Pair research with intent signals for timing

Layer intent signal data on top of your research so messages land while a prospect is actively evaluating. Intent Signal Agents can track triggers like job postings or funding events and add matching leads to a live campaign automatically.

Enrich and launch in one workflow

lemlist MCP or the platform itself can take you from research to enriched contact to live campaign without switching tools, pulling verified emails, phone numbers, recent company news, and tech stack data along the way. Analysis tells you what to say; enrichment tells you who to send it to.

Deep company analysis vs. standard data enrichment

These two workflows complement each other rather than compete.
Capability
Standard data enrichment
Deep company analysis
Contact info (email, phone)
Yes
No
Firmographics (size, industry, revenue)
Yes
No
Pain points and priorities
No
Yes
Buying triggers
No
Yes
Customer language for messaging
No
Yes
Competitive positioning
No
Yes
Enrichment answers who to contact. Deep company analysis answers why they’d reply. lemlist handles enrichment natively through its lead database, so pairing it with the Deep Company Analyser skill covers research through send.

FAQs about the deep company analyser

Can I use the deep company analyser with Claude Free, or do I need Claude Pro?

The skill file works on any Claude plan, including the free tier. Pro, Max, or Team plans give you longer conversations and higher usage limits, which helps when analyzing several companies in one session. Max gives you 5x or 20x more usage per 5-hour session than Pro. (Plans & Pricing | Claude by Anthropic)

What types of companies does deep company analysis work best for?

B2B companies with an active web presence, reviews on G2 or Capterra, and published case studies produce the richest output. Companies with limited public content give thinner results, though job listings and LinkedIn data can fill some gaps.

How long does a single deep company analysis take in Claude?

Most analyses finish within a few minutes once you’ve provided the source URLs. The main variable is how many sources you include and how much you iterate afterward.

Can I run the deep company analyser in batch for a list of companies?

The skill is built to analyze one company at a time, though consecutive analyses inside the same Claude Project work fine. For prospecting across many accounts at once, pairing the skill with lemlist’s enrichment and campaign tools tends to move faster.

Final thoughts

The full loop looks like this: find leads in the lemlist database, enrich them with verified contact data, run a deep analysis with the Claude Skill, build pain-based sequences with copywriting skills, then launch multichannel campaigns and track replies from one inbox.
I’ve been running this workflow for months now. The piece that changed my reply rates wasn’t better subject lines or more sends. It was knowing, before I wrote a single word, what each prospect’s buyers were actually complaining about. That’s what deep company analysis gives you.
If you want to try it yourself, grab the skill file, load it into Claude, and run it on your next five target accounts. Then compare those opens to your last batch of firmographic-only outreach. The numbers will make the case better than I can.
Start a 14-day free trial. No credit card required, cancel anytime.
Share this post