The field guide

The GTM tactics menu

Every channel worth running, with a working point of view on each: what to reach for, for which kind of client, and where the leverage actually is.

This is a standalone reference, not a pitch. It’s the operating playbook behind the work: the same tactics, biases, and hard-won opinions applied across dev-tools startups, vertical SaaS, open-source projects, and our own ventures.

Who the menu is for

Four client archetypes

Channels get tagged with the codes below, so you can read the menu through the lens of the client in front of you.

Dev-tools

Technical buyer

Engineers buy or sway the buy: infra, AI tooling, developer platforms. Events, creators, and PLG carry the load.

Vertical SaaS

Non-technical buyer

A defined persona buys: in-house legal, finance, ops. Events, content, lifecycle, and classes win; the usual fight is a brand-awareness gap vs. funded competitors.

Open-source

Community-led

Community growth is the scoreboard. Engineer-authored content beats AI slop; downloads and footprint are the metric.

Founder-led

Audience-first

The founder's audience is the engine: a podcast, a community, a product with a mission. Relationship currency, warm-intro launches, crowdsourcing.

The menu — 17 channels

Channels & the stance on each

Events / Conferences / Field Marketing

The franchise play. Surgical, not blanket.

The money is in prep + follow-up, not the booth. Pre-event AE outreach, side dinners, audience-specific booth scripts, and pre-drafted follow-up that auto-sends in each rep’s name, with a same-night deadline to claim anyone they want to write personally. It raises the floor on the worst reps and cracks the whip on the best.
ROI bar: 3x pipeline per dollar with T&E included, 4 to 5x without it. Write the goal down before the event, then diagnose misses as bad event, bad audience, or bad execution; only the third earns a rerun. Events are a reward: reps who don’t generate pipeline don’t go to the next one.
Buy a charismatic logistics workhorse, not a "strategic field marketer." Most won’t make events better; what you’re buying is time back for everyone else.
Own a conference only when you have a real community and customer base to draw from; most companies should start with dinners. Then run it as a pipeline instrument, not a P&L: losing money is fine when every dollar lost returns three in pipeline.
Conferences are geography, not content. The meetings and side events around them are most of the value, and an expo pass often beats a booth. People show up to things they pay for; free events lose half the room.
Skeptical of one-shot live demos and junior-heavy hackathon booths. A long-running activation only your product could pull off beats both.

“Make sure we are everywhere we need to be. Nowhere that we don’t need to be.”

Executive Dinners / VIP Receptions

ABM in disguise. Quietly the best dollar-for-dollar play.

Find execs at an event, invite them to a private dinner, make them feel special. 10 to 30 high-ICP guests, LinkedIn pre-outreach.
Best understood as mid-funnel pipeline acceleration, not always net-new logos.
Overbook by about 20% for flakes. Dinners are often the best conversion in the whole influenced-pipeline ledger; the common failure is fumbled follow-up, not the dinner itself.
The cheapest variant is a two-company matchmaker dinner where the host plays connector. A few hundred dollars, and everyone owes you the introduction.
The trap: pitching only an unfinished product yields design partners, not revenue.

“We try and find executives at the event and invite them to the private dinner... make them feel special.”

ABD

Email / Lifecycle / CRM

The most repeatable play in the book.

Cohort segmentation + AI-drafted copy wired straight to the send API. Segments like "loyalists who attended three years running but haven’t re-registered" sold a couple hundred conference tickets on their own. Copy is resource-forward, helpful-first, self-select hooks.
Win-back to closed-lost with a free or cheap class. Overrule the "data’s too messy" objection.
Subject-line specificity beats vanity opens; labeling the content roughly doubled click-through. Open rates aren’t a good proxy.
Product telemetry is the best send trigger you own. Someone nearing a credit cap or crossing an activation threshold gets a short personal note about their situation; that one email beats the whole nurture sequence. Most PLG teams sit on this data and send newsletters instead.
Run campaigns like deploys: version-controlled sends, an automated overlap check so nobody gets hit twice, a dry run before anything ships. Boring safety habits are why you get to keep sending.
Social proof lives in the P.S. People click the postscript, and it never feels forced there. Keep it fresh with a standing routine that points it at the newest release or post.
Email is an immutable channel: protect it. An SDR blasting current customers is a brand wound.

“My two laws of email: be useful and be interesting. And the easiest way to be interesting is to be different from all the other slop in their inbox.”

ABC

Content / SEO / GEO

Priority growth channel. The bottleneck is publishing, not production.

Publishing is not an objective; increasing traffic is, and only if it serves down-funnel. Demand per-page funnel tracking.
Anti-slop rule: if you use AI, you better have 15x the quantity, or have someone who’s done the work write it. Scale only works paired with a real expert + your own tooling.
GEO/AEO is real but squishy and slow. Document a benchmark + 90-day plan instead of promising fast wins. Skeptical of measurement tools that lack LLM APIs.
Answer the questions the pricing page dodges: what it actually costs, where the limits are, what happens when you hit them, how it stacks up against the incumbent. Devs search those exact phrases and vendors are too nervous to answer plainly, so ranking is cheap. The boring candid pages outperform.
Check crawlability before spending a dollar on content. Half of dev-tool sites are React apps serving crawlers an empty div. SSR, real links in the raw HTML, structured data; indexing can turn around with zero new pages.
Technical / architecture storytelling converts: people like reading how something was done.
The CMS is dead weight now. The only reason marketers had one is that no engineer was at their beck and call, and a coding agent is exactly that. Target idea-to-live-on-site in ten minutes.
Comparison and "X vs Y" pages end internal audience debates. Publish them, let them index, and let traffic and conversion data settle who your buyer really is.

“You can build a slop cannon on your own. But you need someone reviewing the slop outputs, or it will just be slop.”

Paid Ads

A measurement instrument first. Broker the execution out.

Instant, objective read on what your audience likes; winning ad copy percolates into product marketing ("super engineer," "cleanest code" became company vocabulary).
Localize: "cleanest code" won North America and flopped abroad; "super engineer" won LatAm.
Don’t be the operator. Broker it out, mandate a weekly ad-review meeting once spend is real (~$40K/mo), and take the free SEM/social audits agencies offer; they’re surprisingly honest.
Minimalist attribution: the only attribution worth doing now is ad conversions. Twitter ads dismissed as a social play. Amplify underused offers (the "$5 free credit") as creative.
"We tried ads and they didn’t work" means they were done wrong; every giant company is built on them. Hire paid people out of B2C (they live or die by the numbers), and pick creative with the squint test: the ad still legible with blurred vision wins.

“The only approver of an ad is the audience. No one’s internal opinion matters.”

Influencer / Creator

Proven for dev tools. Don’t let it become the only bet.

Creators have gone "bottomless wallet" expensive on AI money. Theo = volume; Prime = fewer, higher-quality enterprise leads. Pull back on the ones who get too pricey.
Break out beyond dev-tool influencers: Pragmatic Engineer, Lenny’s, Fireship (universally liked; haters never see the ad).
Pass on the big $50K+ bets until the product is mature.
Match the offer to the creator’s audience, not to your own priority list. The same money on the same creator can print pipeline for one offer and produce zero leads for another. When a bet misses, go really different on the retry, not a tweak.
Distrust click attribution; rely on "how’d you hear about us?" on the first call.

“Sponsor the creators everybody likes. The haters never see the ad.”

LLM Recommendation Engineering (AEO)

The frontier play. A high-variance big bet.

To bias what models recommend, flood the training-data window (~3 months) with influencer + paid content + podcast ads all pushing one message.
Ballpark ~$1M over three months (~$500K for Theo + Prime + Pragmatic alone).
Not for the timid. Most companies’ appetite for extreme measures is low; say so out loud before anyone signs up for this.
The free tier under the big bet: llms.txt, structured data, answer pages plain enough to quote, and skills or plugins for the coding agents your buyers already sit in. (An MCP server reads as a gimmick now; the wave passed.) GEO is getting the agent to do the selling for you, and it moves in weeks, not the years SEO takes.

“A growing share of developers never see your website. They ask the model, and they take the answer.”

Video

Attention over polish.

Skip color and sound correction; reinvest the time in better hooks.
Chop one shoot into many pieces; capture footage at every event (GoPros + mics). Framework: Maintain, Elevate, Expand.
Short-form works even for lawyers; UGC dev videos at hackathons are great fodder.
A phone-at-the-desk clip routinely beats the produced launch video, but polish makes enterprise buyers feel safe. Run both.

“We’re not looking for awards here. We’re looking for attention.”

Social / X

A reactive launch playbook.

Pre-write a quote-tweet + reply timed to a partner’s announcement. Tone: a little humble, but still bragging.
Turn customer praise into anonymized social proof + a dinner-invite CTA.
At hackathons, use your account to make developers feel like stars, so they like you for promoting them rather than asking.
Low effort beats high effort: the throwaway screenshot outdraws the produced month-of-releases post, and bragging performs worst of all. Post volume is safe.

“What’s great about algorithms is that if nobody sees it, nobody saw it.”

Sales-led / Outbound / SDR

Study the outlier, not the aggregate.

Reverse-engineer your best rep and productize more of that; winning behaviors are trainable.
Speed-to-lead is the whole game: a calendar-viewer who doesn’t book gets a call within 5 minutes.
Design the SDR fail-state as a win: if you can’t qualify, still sell the self-serve plan. SDRs don’t close.
Narrow-cast over broadcast; curated outbound (a handful of qualified leads worked by hand) beats automation. Skip Clay this generation.
Run a weekly anonymized email teardown: every rep submits one send with the name stripped, the leader line-edits all of them with the reasoning attached, and a small prize goes to the best, judged blind. Five weeks of this changes how every email gets written.
Ban first-person openers. An email that starts with "I" or leans on "noticed" and "saw you" is cooked. And before building outbound at all, mine the inbound and the database you already own.

“The outlier is more valuable than the aggregate.”

PLG / Self-Serve / Pricing-as-GTM

Delay sales hires when PLG is strong.

Vercel hit $13M before the first product advocate. Self-serve to pipeline is an ops problem, not a demand problem.
Track "graduated revenue" (self-serve to enterprise) stacked on the chart so the funnel doesn’t look falsely flat.
Never lead with the unproven product; it’s always the assist, the add-on, the upgrade. Discount the proven service to bundle the new seat.
Pricing to win a marquee logo can be worth the margin; push toward variable / "agentic" pricing to capture value beyond seats.
Founders systematically under-price. Buyers will pay many times what you think a seat is worth when a closer asks with confidence. And usage top-ups are not flight risks anymore; stop forcing them onto subscriptions.

“The unproven product is never the headline. It’s the assist, the add-on, the upgrade.”

Partnerships / Co-marketing

A real pipeline source most teams under-attend.

Tie partner comp to total bookings; attribution isn’t reliable here.
Two questions in every partner conversation: what are your company goals, and what are you personally measured on? The answers tell you what to trade.
When a partner comes to you, ask for more than the integration-and-blog-post minimum. They came to you, which means they have a goal.
Ask big customers to post and backlink post-procurement.
But breaking out of your own audience just straight up costs money; coattails aren’t free.

“Breaking out of your own audience just straight up costs money.”

Community & Gamification

For OSS, community growth is the metric.

Double downloads / footprint is the real success measure; treat it as the scoreboard.
Give the community a mission bigger than the product: crowdsourced compute, shared datasets, public goods (the Folding@home model). People show up for missions, not roadmaps.
Leaderboards and public stats turn usage into marketing; a little friendly controversy around them is free distribution.
Meetup-in-a-box: pay local organizers a few dollars per checked-in attendee for pizza and beer. Goodwill money that compounds into thousands of event attendees across markets you could never staff.

“For open source, the community isn’t the audience. It’s the scoreboard.”

Interactive Content / Classes / Education-as-Wedge

Low-commitment inbound that converts in-product.

Assessments and quizzes ("how ready is your team, really?") are cheap to build and self-qualifying; the prospect tells you exactly where it hurts.
Corporate AI-training classes as a product-led wedge: ~70% win rate once a prospect is in the product. Charge for the class and comp seats freely (price is a seriousness signal), and accredit it with the buyer’s professional body so attendance earns career credits.
A cheap paid workshop is the door-opener to the seven-figure deal: the buyer who won’t start big will pay to have your best people in their building for two days.
Managed services that teach the client while doing the work is a very interesting go-to-market.
Build the two-minute version of the argument instead of the 1,500-word version. Developers extend trust to artifacts they can run, not claims; one working tool out-earns a quarter of content.

“Once a prospect is actually in the product, the win rate runs around seventy percent.”

ABC

Benchmarks / PR / Newsletters

Independent proof, yes. Your own benchmark, careful.

Independent benchmarks are a proof point to mirror on-site immediately.
Publishing your own benchmark risks the optic: you made your own because you’re bad at the standard ones? Benchmark yourself against models freely, but competitor comparisons need a neutral third party; contribute your methodology to one instead of owning it.
Low regard for old media; you only get in the news when you fundraise. Build PR backwards from "what is the tweet," and pay agencies for the ranked backlink list, not the ideas. Vertical-newsletter sponsorships at scale.
Launches need formal tiers + a story per launch; willing to delay the "launch" until there’s data. Bundle launches by default: launch weeks bleed attention day over day unless a through-line carries them. Engineering does releases; someone has to decide which ones earn a launch.

“Start every launch from the end: what is the tweet?”

Podcast (your own)

Relationship currency, not a numbers channel.

A founder-run podcast keeps you deal-adjacent: prospects who would never take a sales call say yes to an interview, and they come on because of what they do.
Fifty episodes of transcripts with your exact ICP is a content mine: an agent drafts the posts, the FAQs, and the topical cuts. When a past guest gets promoted, that’s a post, a link, and a warm message.
A podcast going dark is the clearest "channel not shipping" smell there is.

“People will say yes to a podcast invite who would never take a sales call.”

ABCD

GTM Engineering / Agentic Ops

The force multiplier under every channel above.

Skip the marketing automation platform. A database, a send API, and an agent support more sophisticated tactics than the big suites; teams over-rotate on powerful tools and undercomplicate the tactics.
The agent unlock is the play you always knew would work but could never hand-execute: twenty micro-segments with a unique email each, some lists five people deep, shipped in an afternoon.
Standing agents earn their keep: a Monday competitive-intel rollup, a weekly search-performance brief in Slack, changelog drafts straight from the repo.
Basic automation still works; agentic is the layer on top, not a replacement for it. And never tell an agent to iterate until it hits 100%, because the objective becomes lying.

“As far as an executive is concerned, a marketing ops person is someone you Slack and then an email happens. There’s no reason that can’t be an agent now.”

The operating system

Principles underneath the channels

Run marketing as a science

Line-by-line funnel spreadsheets per page, event, and email; a sheet never has an integration problem. "Call your shots": predict outcomes, then review. That’s how you train a team.

Attribution informs, it doesn’t award credit

A VP wanting sales-vs-marketing credit splits is a red flag. Tie everyone to the shared bookings goal instead.

Crypto vs. index fund

Don’t spend it all on safe copycat bets. Reserve room for high-variance big bets.

Run toward the burning building

Find the unowned, high-visibility problem and lead it to earn the right to do the work that matters.

Get good at saying no

Strategy is the documented no: write down what you’re not doing, so when it comes up (it always does) you point at the decision instead of re-arguing it. Every new tactic is paid for in one of three currencies: time from another tactic, budget, or a person.

Earn the right to scale

Isolate decisions to the very next step: not the next 10,000 customers, the first 50. Most problems are ones you haven’t earned the right to worry about yet.

Marketing extracts; it doesn’t invent

See the workshop →