← Contents09 / 20

Module 9 — Making Pipeline: Outbound, Inbound & the AI SDR Swarm

"Pipeline is the only prayer the board respects. Everything else is interpretive dance performed by idiots in front of a slide deck." — overheard from a CRO at 11 p.m., third whiskey, eyes like two cigarettes someone forgot to extinguish

It's 4 a.m. and somewhere a fleet of agentic AI SDRs is firing its nine-millionth "Quick question, {{FirstName}}—" into the goddamn void, and the void is starting to answer back, and that is the thing that sits behind my eyes at this hour like a second skeleton. The machines are getting good. Not great. Not sentient. But good enough that the $65K ramp you were going to burn on a wet-behind-the-ears SDR is now a line item in a SaaS contract with a logo that looks like a smug geometric paperclip. Here is what you are going to do about it. Pour something harsh. We're going hunting, and the hunting starts with a number — your number — and works backward until you've got a plan or you've got nothing, and "nothing" is not an option the board will accept at this fucking altitude.

THE JOB

Pipeline generation is the deliberate, systematic manufacture of qualified opportunities — enough of them, early enough, that The Number has room to breathe and you have room to miss a few without your career ending on a Tuesday. That's it. That's the whole job. You are a factory whose only product is future revenue that doesn't exist yet, and the factory has to run on schedule whether inspiration shows up or not.

Two engines drive this machine, and confusing them is the first and most expensive sin most orgs commit:

  • Demand generation — you create awareness and appetite where none existed. The buyer wasn't looking. Outbound sequences, paid ads, content marketing, events, dark social, the whole screaming apparatus. You are making the market thirsty for something it didn't know it needed.
  • Demand capture — you harvest intent that already exists. Inbound forms, branded search, "request a demo" clicks, G2 review traffic. The buyer was already walking toward the door; you just have to not be such a colossal ass that you fail to open it in time.

Most failing GTM orgs pour every dollar into capture, label it "marketing," and then act stunned when growth flatlines the moment existing demand is tapped out. You cannot harvest a field you never had the balls to plant. RULE No. 9: capture is cheaper and gen is bigger, and you need both or you die slow and confused.

THE PLAYBOOK

1. Size the pipeline you actually need — coverage math before a single goddamn email

Before you write a sequence, before you argue about messaging, before you run a "pipeline blitz" that will produce nothing but false confidence and wasted hours, do the arithmetic. Hope is not a coverage ratio.

Pipeline needed  = Quota gap ÷ Win rate
Coverage ratio   = Open pipeline $ ÷ Quota $

The blood ratio — 3x coverage — gets invoked like scripture at every pipeline review from here to San Francisco. Three dollars of qualified pipeline for every dollar of quota. Fine. Except it's a fucking lie if you don't anchor it to your actual win rate:

  • Win rate ~33% → 3x is correct.
  • Win rate 20% → you need 5x. Minimum. Write that down.
  • Win rate 50% → 2x. Congratulations, you lucky bastard, and I don't fully believe you.

Layer in cycle time because time is not your friend. If your sales cycle is 90 days, the pipeline that closes in Q3 has to exist and be qualified by end of Q2. Pipeline created today pays the quarter the cycle actually lands in, not the quarter you need the money. Build a creation target backward from the close target, by month, by source channel. Write it on the wall in permanent marker. Make it uncomfortable to look at. That discomfort is the point.

2. Build the outbound motion — the generation engine that doesn't run itself

Outbound is a four-part assembly line. Run all four or the whole shit jams at the worst possible moment — usually the month your CRO gets on a plane to visit the board.

Step 1 — Lists and targeting. Pull from your ICP definition (Module 6 — if you skipped it, go back, I'll wait, I have nowhere better to be at this hour). Tier your accounts. Tier 1 = named, researched, hand-curated, worth actual human attention. Tier 3 = high-volume, automatable, feed to the swarm. A garbage list produces garbage pipeline forever, no exceptions, no recovery. Source from data providers, enrich, deduplicate, suppress current customers and open opportunities like your life depends on it — because your quarter does.

Step 2 — Sequences and cadences. A sequence is a timed, multi-touch, multi-channel series of deliberate contact attempts. A sane B2B cadence runs 8–14 touches over 2–3 weeks, mixing channels like a competent bartender, not a panicked one. Do not fire 14 emails in 14 consecutive days. That is not a cadence. That is a hostage situation, and the hostage is your sender reputation.

Step 3 — Multichannel or die. Email + phone + LinkedIn + (occasionally) video or direct mail. The phone still works precisely because every coward abandoned it. A triple-touch on a single day — call at 8 a.m., email at 9, LinkedIn view at noon — lands harder than three emails spread across three days. Be the one who still picks up the goddamn phone.

Step 4 — Messaging that doesn't make the prospect want to scrub themselves clean. Lead with their trigger, not your feature list. "I noticed you just posted 12 AE roles in the last 30 days" beats "We're the industry-leading platform for revenue intelligence" every fucking time, in every vertical, in every economic cycle, forever. Relevance is the only thing that survives the inbox apocalypse.

3. Build the inbound and PLG capture engine

  • Inbound: SEO, content marketing, review site presence (G2, Capterra, TrustRadius), webinars, community, branded search. Slowest engine to build. Cheapest at scale. Highest-converting traffic you will ever touch — because these people came looking for you. A branded-search demo request closes at multiples of a cold outbound sequence and costs almost nothing to service. Plant the seeds or explain at the QBR why you don't have any.

  • PLG (product-led growth): the product is the top of funnel. Free trial or freemium → in-product activation → a product-qualified lead (PQL) when usage crosses a defined threshold that actually predicts buying intent — three users invited, API called 100 times, whatever your data says. Route PQLs to sales like the precious ore they are. PLG without an activation metric is just a free tier hemorrhaging server costs while the CFO develops an eye twitch.

4. Signal and intent-based outbound — the shit that actually moves the needle now

Spray-and-pray is dead. Deliverability killed it. The only outbound that converts in the current environment is triggered outbound — sequences fired by signals, not by a calendar reminder that someone set in 2022.

Build plays off signals:

  • Intent data — third-party (Bombora, G2 Buyer Intent, etc.) showing accounts surging on your category keywords. First-party signals: repeat pricing-page visits, whitepaper downloads, doc center activity, anything that says "someone over there is thinking about buying something."
  • Trigger events — funding rounds, executive hires into relevant roles, job postings that reveal pain (e.g., "Head of Revenue Operations" posting means they don't have one), tech-stack changes detected via job descriptions, expansion into new markets, M&A activity that creates chaos and therefore need.

A signal-fired sequence hitting an account within hours of the triggering event converts at a categorically different rate than the same sequence sent cold on a Tuesday because the SDR had a slot open. The signal is the relevance. Without the signal, you're cold calling into the dark and wondering why everyone's annoyed.

5. The AI SDR swarm — deploy with your eyes wide the hell open

Agentic AI SDRs can research accounts, draft personalized outreach, send at scale, self-route positive replies, and run dozens of simultaneous sequences without once crying in the parking lot or asking for a $10K OTE bump. They are genuinely impressive and genuinely dangerous. Here is the actual scorecard, not the vendor pitch:

Good at: infinite-volume Tier 3 prospecting that would take a human team six weeks; real-time signal monitoring across thousands of accounts simultaneously; instant first-draft personalization that's better than what most reps write; never sandbagging, never quitting, never complaining about their territory assignment.

Bad at: genuine judgment calls. Reading a room. Knowing when not to send — the art of restraint that separates a world-class AE from a noise machine. True multithreaded relationship-building across an enterprise account. Recognizing that "AI-generated personalization" has crossed into uncanny-valley spam territory and is actively torching your brand reputation with every send.

Human-in-the-loop is not optional. This is the line between using the swarm and being consumed by it. Let the swarm draft, research, build the sequence, and queue the sends. Let a human approve Tier 1 and high-ACV sends, set guardrails, own the brand voice, and read the fucking replies. The failure mode is total automation: a bot responding to a grief email from a dead lead's colleague, in a thread no human ever touched, at 3 a.m. You saved one SDR salary. You cannot put the brand back.

My attorney advises that no vendor SLA covers "we sent ten thousand emails from your domain and got you on every major blocklist." That's yours. Own it.

6. Volume vs. deliverability — guard the domain like it's the last one on Earth

Volume is now trivially easy. Which is exactly why deliverability is the entire fucking ballgame. You can generate infinite sends with the swarm; you cannot generate infinite inbox placements if you crater your sender reputation. Lock these down before you send a single sequence:

  • SPF, DKIM, DMARC — email authentication records that prove to inbox providers that you are who you say you are. Missing any of these and Gmail, Outlook, and every other serious inbox is quietly shadow-filing you into oblivion. Post-Google/Yahoo bulk-sender rules (2024), DMARC is not optional. It is table stakes. If your domain doesn't have these set up, stop reading this and go fix that. Right now. I'll be here.

  • Domain warmup — never blast from a cold domain or your primary corporate domain. Buy secondary sending domains (yourcompany-hq.com, meet.yourcompany.com), warm them deliberately — start at 20–30 sends per day, ramp slowly over four to six weeks — and spread volume across multiple domains to keep per-domain numbers low and reputation clean.

  • List hygiene — verify email addresses before sending, suppress hard bounces immediately, kill spam trap addresses before they find you. Bounce rate above 2–3% and inbox providers mark you as a shit-show operation. Because you are one.

  • Reputation monitoring — watch your spam complaint rates (keep under 0.1% or providers blacklist you), track engagement metrics, check blocklists regularly. Your sending reputation is a credit score. One binge of garbage volume built on a dirty list tanks it for months, and you can't borrow your way out.

The swarm makes infinite volume trivially possible. That makes deliverability the only skill that matters. Sending more is easy now. Sending more without immolating the domain reputation that took you three years to build is the whole discipline.

HOW IT GOES TO HELL

The VP of Vibes announces a "pipeline blitz" — all-hands, everyone dials more, no list, no signal, no targeting criteria, no plan beyond the announcement. The blitz generates 400 disco-qualified junk opps from every VP of Whatever they could find on LinkedIn, which all die in Stage 1 within three weeks. Coverage looks fucking gorgeous on the slide. Win rate craters. The Number is unmoved. The VP of Vibes declares victory on the coverage slide and conveniently changes jobs before Q3.

The Founder Who Still Thinks He's the Best Salesperson in the Building insists outbound is "just hustle, you don't need tools for hustle." Refuses to fund data enrichment, sequence tooling, or deliverability infrastructure. Reps prospect from a two-year-old exported CSV. Bounce rate hits 14% inside a week. The primary domain gets flagged by Google's bulk-sender systems three business days before the board meeting. The domain can't be unflagged. My attorney advises you cannot sue Google for this, regardless of how deeply you want to, and regardless of how many whiskeys you've had when you call him about it.

The Guru — Lance "The Closer" DiMarVo — sells a $2,000 course on "Hyper-Personalization at Infinite Scale," which is a phrase that describes, precisely, the thing that gets you marked as a spammer by every inbox provider in the Western world. You bought it. Your SDR team bought it. The course was written by a GPT wrapper running off Lance's 2019 LinkedIn posts. Your domain is now on three blocklists. Lance has moved on to selling a masterclass on AI governance.

The bot-only org — the inevitable conclusion of "let's remove humans to cut costs" — turns the swarm to full autonomy, removes all human review, and cranks volume. Eighty thousand sends in the first week. Complaint rate spikes. Three domains blacklisted by Thursday. The brand is now searchable as "the spam company" on Reddit. You saved one junior SDR's $58K salary and spent it burning the sender infrastructure that supports your entire outbound motion. Math is hard.

FIELD RULES

  • Coverage is a function of your actual win rate, not a number you heard some bastard cite at a conference. Do the division. 3x is a default, not a law.
  • Capture what exists; generate what doesn't. Fund both or stall. There is no third option.
  • The signal is the relevance. Triggered outbound beats cold outbound by an order of magnitude. Without a signal, you're just noise with a budget.
  • Deliverability is the real constraint now. SPF/DKIM/DMARC, warmed secondary domains, clean verified lists — or your volume dies in spam folders and you've accomplished nothing.
  • Let the swarm draft. Keep a human on the trigger. Full autonomy is how you become "the spam company" in your category. That reputation does not wash off.
  • Pipeline created today pays the quarter the cycle lands in. Build the creation target backward from the close target, by month, by source. Then look at that number every morning until it stops feeling terrifying, which it never will.

From the field, 4:14 a.m. — Expense report, line 11: "secondary sending domains (4) — $48." Cheapest goddamn insurance I have ever purchased in this industry. The swarm is running in the next room and it just drafted a better opening line than the one I spent twenty minutes on, and I am sitting here genuinely unsure whether to be relieved about that or to start drinking earlier than usual. Both, probably. — your operator, still awake, still watching the bounce rate.