Kenny Damian

@itsalexvacca sorting replies on a conceptual level, and establishing the intent first is highly underrated. you don't want to just burn your credits to enrich a huge list of engagers who might not even be interested in your solution in the first place.

Cold email off a bought list replies at 3.43%. Signal-triggered outbound replies at 15 to 25%. A comment on your own post is the cheapest signal in that second bucket. The entire flow: Export everyone who engaged with Apify, enrich in Clay, run a waterfall for the verified email, then send one message that names the post they commented on. Content is 65% of our inbound and this is most of how it converts. But... There's a filter on top of it that you must get right for this to work. Most people score the person (role, company size, seniority, funding, the usual firmographics) and ignore what they wrote. The comment itself is the other half of the score. You see, under any LinkedIn post, three types of comments tend to show up: → The compliment. "Great breakdown." A like with extra steps. → The agree-and-add. Usually a peer building their own audience. Good for reach. → The one where someone describes their own situation. A perfect ICP match who wrote "great post" and a mediocre fit who described their broken process are not the same lead, and Clay cannot tell you which is which. The third type goes first, every time. That person sat down and wrote their own discovery notes in public, so the email back writes itself. We run outbound for 400+ B2B companies and the rule doesn't change here: the list is the strategy. Nobody would send a raw database export straight into a sequence without filtering it first, and yet people do exactly that with their own comment section because it feels warm. But all of that is sorting a pool you already have. How many of the third type land in it gets decided before you publish. The more specific the example inside the post, the more often someone answers with their own version of it. Which means the quality of the post decides the quality of the list. It's why you cannot fully separate the content work from the outbound work.

pov: you just found out 1 person in 1000 hitting spam is enough to choke your entire domain

@itsalexvacca every team should perhaps measure meetings books vs what happened those booked meetings track only the former, and you're ultimately left with an incomplete picture..

We run outbound across 70+ clients. Here are the 5 plays we use to turn a signal into pipeline: 1. Awareness scoring ↳ Not every account deserves the same effort. ↳ Score each one by how many signals it's firing and how strong they are. ↳ Then work the top 10% that are actually in-market, and let the rest keep warming. ↳ (Clay + your CRM + a scoring agent) ↳ Result: reps only spend time on accounts ready to buy. 2. Trigger automated outreach ↳ The moment a signal fires, the account auto-enrolls into a sequence. ↳ No manual list-building. No weekly pull. ↳ The copy references the exact trigger, so it lands relevant, not random. ↳ (Clay to Instantly / lemlist) ↳ Result: you reach them in hours, not weeks, while intent is still hot. 3. Slack and CRM alerts for manual outreach ↳ For your highest-intent signals, skip the automation. ↳ Fire a Slack alert straight to the rep who owns the account, with full context. ↳ They send one personal, well-timed message by hand. ↳ (n8n to Slack + Attio) ↳ Result: one human touch on a warm account beats 100 cold sends. 4. Customer expansion monitoring ↳ Signals aren't just for net-new. ↳ Point them at your existing base: new hires on the buying team, a champion changing jobs, a usage spike. ↳ Each one is an expansion or a save waiting to happen. ↳ (Common Room + UserGems) ↳ Result: expansion revenue and fewer churned logos, no new accounts needed. 5. Targeted ABM lists ↳ Turn a signal into a list. ↳ Everyone who just raised, opened the role, or added a competing tool becomes a tight ABM tier. ↳ Your account list is built on intent, not a guess. ↳ (Clay + Crunchbase + Ocean) ↳ Result: every account already has a reason to talk.

@JoshEdmondson_ @itsalexvacca

@itsalexvacca @Kenny_GTM Kenny just found out he was the lead magnet

pov: @Kenny_GTM finding out i just gave away 600 hours of his work for free

@itsalexvacca 🤯

Stop being the 47th identical cold email in your prospect's inbox. Everyone pulls from LinkedIn, ZoomInfo and Apollo, so everyone emails the same people on the same day. I mapped 60+ sources that surface buyers before your competitors ever see them. 1. Product Hunt founders the day they launch 2. G2 reviewers calling out a competitor by name 3. GitHub users publishing the exact stack you sell into. Every one is a live buying signal you pull with an Apify scraper and Clay. Check all the data sources in the map below (bookmark for later)

@itsalexvacca the most slept-on money hack in outbound.

you are paying thousands for data someone scraped off linkedin and sold back to you. plug apify into claude code and scrape it yourself: 1/ create an apify account. free tier comes with $5 in credits 2/ connect the apify mcp to claude code. one-time setup 3/ paste your website url into claude 4/ claude reads your offer and works out the icp on its own 5/ ask for the list. claude picks the scraper, runs it, hands back the rows 6/ swap the personal emails for work emails with an email api (we use leadmagic) 7/ qualify in clay, send our run: 100 leads, 70 on company domains, 10 cents of compute. same play works on your post engagers too. one of my posts had 479 last week, claude pulled every one of them and ranked them against the icp. full walkthrough in the video below (bookmark for later):

honestly the best money hack in outbound right now

@itsivanfalco you could probably find like 100 best fit leads (< 10 cents)

not enough people talk about how cheap lead gen got > plug Apify into Claude Code > point it at your website > ask for your best-fit leads and watch what happens

Claude Code can run your entire cold email operation from a few prompts. You tell it who to target. It sources the accounts, enriches them, finds the decision-makers, verifies the emails, and hands you a campaign ready to send. It runs on 6 APIs. Here is where each one sits: 1. AI Ark (@_AI_Ark). A 70M+ company database that builds your raw account universe. 2. CompanyEnrich. 30M+ companies for firmographics: size, industry, revenue, headcount. 3. PredictLeads (@predictleads). The signal layer that flags a fit account as in-market: hiring surges, funding rounds, tech changes. 4. BlitzAPI. Finds the decision-makers inside the account, so you reach someone who can sign. 5. Limadata. The email layer that turns a name and a company into a deliverable address. 6. FullEnrich (@FullEnrich). The fallback layer that cascades 20+ providers and hits a 95% match rate on the contacts you found. @InstantlyAI sits at the end and sends.

@itsalexvacca 💯

@Kenny_GTM I think it's also worth calling out the "reverse" trap in this case... using Claude Code for just about everything. Even when the Desktop app could have perhaps given you better results at a lower token consumption rate.

There are three versions of Claude, and most people pick one and use it for everything. Each is built for a different job. Bookmark this and open the right one every time. For thinking, open Claude Desktop. Reasoning through a problem or drafting a strategy in conversation. For building, open Claude Code. Shipping an n8n workflow, a Railway service, a Python script. Code touches your files, runs commands, and holds a whole project in context. For delegating, open Claude Co-Work. The home for a task you want running while you step away. I have spent 600+ hours in Claude Code. Almost everything I run lives there: n8n workflows, Railway services, Python scripts, and six AI agents on autopilot. Most people build inside a chat window, then hit the wall when it cannot touch their files. Match the surface to the job and Claude can run your stack.

@itsalexvacca sauce 🔥

I have sent 23M+ cold emails across 70+ clients. If I were building a cold email machine from zero in 2026, this is the infrastructure I would stand up before writing one line of copy. 1. Secondary domains only. Keep your primary domain out of sending entirely. If one gets burned, it should never be the domain your company runs on. I would register 20. 2. 40 mailboxes, each on its own IP. Your volume comes from the number of mailboxes, so you spread sends across 40 inboxes instead of overloading a few. 3. 20-25 sends per mailbox per day. Forty mailboxes at that rate is real volume and keeps every inbox under the spam threshold. 4. A 2-week warm-up before the first cold send. A new domain has no reputation, so you build it for two weeks or it lands in spam. 5. SPF, DKIM, and DMARC on every domain. Miss one authentication record and the inbox providers flag you on arrival. 6. Replace domains every 9-12 months, because sending reputation decays. Do this and you land 95%+ in the primary inbox. I have watched teams rewrite subject lines for a month while the real problem was a domain with no warm-up.

You can build an entire outbound campaign from scratch in one terminal window. I run this across 70+ clients. The last campaign I built: 153 contacts in a few minutes, 140 with verified emails, live in Instantly. Every step: 1. Define the ICP in plain English. Open Claude Code and describe who you want. Titles, company size, industry, region, accounts to exclude. No filters to click. 2. Pull the list through the Apollo API. Claude hits Apollo and returns 153 contacts in 90 seconds. Most teams still export those by hand, one CSV page at a time. 3. Run an enrichment waterfall. One provider always leaves gaps that show up as "no email found." So you cascade. Wiza, then FullEnrich, then Prospeo. Coverage past 90%, and the list closed at 149 verified. 4. Write the first line from the enriched data. Claude already has the firmographics and the signal in context, so the opening line references a real detail it pulled, like a recent funding round or a tech change. 5. Push into Instantly. The list loads and the campaign goes live. No manual import. You add your API keys and debug the occasional failed step. Every other step runs unattended. That afternoon of CSV exports, tab-switching, and single-provider gaps is gone.