Orchestrating the Full Funnel: Developer Marketing in 2026

Introduction

Cockroach Labs sells mission-critical database infrastructure, the kind of product a single engineer can quietly test on their own, and a CTO still has to sign off on. In this conversation, James Sarris walks through how Cockroach Labs actually maps who needs to say yes on a deal, why they stopped filtering target accounts by revenue, and a campaign result that ran against everything conventional lead-gen wisdom would predict: an ungated asset that outperformed a gated lead-gen form for the conversions that actually mattered.

Balancing self-serve trials with top-down buy-in

• Developer tools run two motions at once: a self-serve trial any engineer can start, and a formal buying process that needs executive sign-off for something mission-critical.
• Cockroach Labs' term for this: the "shadow evaluator," the senior developer or architect running their own proof of concept before any formal evaluation begins. By the time the buying committee is in the room, that person has already formed a strong opinion.
• The fix isn't picking one motion over the other. It's making sure messaging resonates with both the engineer evaluating day-to-day fit and the executive evaluating operational risk, without diluting either into generic middle-ground content.
• Best-performing content leads with the technical truth engineers respect, then surfaces the business implications executives care about, rather than starting from the lowest common denominator.

Why 3,500 job titles broke the old approach to targeting

• Factors' own research found over 3,500 unique job titles on LinkedIn within a typical engineering org, roughly 5 to 10 times more fragmented than a function like marketing.
• Instead of guessing five or ten titles up front, Cockroach Labs mines its own CRM: which deals closed, who was in the buying committee, what were their actual titles.
• The same logic extends to product usage data. Who's logging in, setting things up, running the proof of concept? Those are the shadow evaluators, identified by behavior, not by title.
• The output isn't always a title, sometimes it's a seniority band instead, since exact titles vary too much to target reliably on their own.

ICP for companies growing 0 to $100M in a year or two

• Traditional ICP filters (industry, employee count, revenue) assume steady, predictable growth. That breaks down for AI-native companies that can scale from a few million to well over $100M in ARR within a couple of years, sometimes a couple of quarters.
• What Cockroach Labs tracks instead: hiring patterns (especially senior engineering and platform hires), funding announcements, and founder background, whether the founders came from companies with a track record of scaling fast.
• The goal is catching companies at the inflection point, before their current revenue would ever qualify them.

How AI and LLMs are reshaping the buyer journey

• Organic and direct traffic have plateaued or declined across much of Factors' customer base as buyers do more research before ever visiting a company's website.
• The marketing funnel itself hasn't changed shape, buyers still need to move from problem awareness to solution evaluation. What's changed is how much of that happens before a company ever sees the visitor.
• Being a genuinely helpful, authoritative source at each funnel stage builds the trust needed to be considered later, especially since AI-generated content is now easy to spot ("sales breath" is real, and so is content that reads like it came out of Claude with no editing).
• Sales needs the same discipline: knowing what a prospect has already seen from marketing, so the conversation doesn't repeat information or feel disconnected from what they've already learned.

Mental availability, and the campaign that shouldn't have worked

• The concept: when a buyer starts thinking about a problem, they typically think of three vendors. Being in that top three means inbound demos come to you. Being outside it means fighting for every deal.
• This is pushing more active, always-on brand advertising across channels like LinkedIn, Reddit, YouTube, and programmatic, not just relying on inbound content the way an older SEO-led playbook would.
• The contrarian result: Cockroach Labs ran an ungated version of an asset that would normally sit behind a lead-gen form. Lead-gen forms usually convert better in-platform, but the ungated version ended up driving more bottom-of-funnel conversions further downstream, visible only because of proper attribution tracking the full path.
• The takeaway James kept coming back to: marketers assume they know their persona, and the data regularly proves that assumption wrong. Treat channel mix and content gating as things to test, not things to decide in a planning meeting.

Community, DevRel, and documentation as credibility, not just content

• Docs and blogs serve different intents. Docs map to technical queries and increasingly show up as direct LLM citations; blogs map to earlier-stage problem awareness. Internal linking between the two (blog surfaces a problem, doc proves the technical capability) matters more as AI search tools prioritize on-site content over third-party opinion sources.
• Community and DevRel functions, even informal ones, give marketing teams direct, ungated insight into what their audience actually cares about, insight that's harder to get secondhand through a sales call recap.
• This kind of function is common in developer-focused companies and rare almost everywhere else, even though the underlying value (staying grounded in what your actual audience thinks, not what you assume) applies to any B2B company.

Tracking the full account journey with Factors

• Cockroach Labs uses Factors to correlate touchpoints across channels, connecting an early top-of-funnel visit to a later mid-funnel action to see the real path an account takes, not just isolated conversion events.
• Paid spend gets distributed across the full target account list instead of concentrating on whichever large accounts already generate the most impressions.
• Pre-conversion attribution, seeing which accounts are engaging before they've converted at all, was one of the actual reasons Cockroach Labs adopted Factors in the first place.

Do This Today

1. Pull your last 5 to 10 closed-won deals and list everyone who was actually involved, not just the person who signed.
2. Cross-reference those names against product usage data: who logged in, who ran the trial, who raised support tickets.
3. Group titles by seniority instead of exact title match if your org (or your buyer's) has too much title fragmentation to target reliably.
4. Build a lightweight ICP layer around hiring activity and funding announcements instead of filtering only by current revenue.
5. Run one campaign with an ungated version of a normally gated asset, and measure downstream conversion, not just top-of-funnel form fills.
6. Map your cross-channel touchpoints to see which combinations (a LinkedIn impression followed by a Google search, for instance) actually correlate with conversion.

‍