Notes for George

I grew Blackbaud's performance marketing program from $0.5M to $20M+ annually across paid search, display, paid social, YouTube, and SEO. At Refine Labs I ran or oversaw performance marketing strategy and execution across 100+ B2B funnels - including paid search, paid social, YouTube, Reddit, webinars - for companies ranging from seed to publicly traded.

Most clients came to us with one of four problems: they didn't know how to execute, they were spending in the wrong places, they couldn't scale without losing efficiency, or they needed full outsourcing of strategy and execution. The outcomes I helped them achieve were consistently record quarters in pipeline and revenue.

The bigger learning from those 100+ funnels was the pattern recognition I couldn't have developed any other way. Funnels break down for a handful of consistent reasons + the fixes are more repeatable than most teams think. That diagnostic framework became the foundation for Affect - and it shaped how I approach every performance problem since. I wrote about what I found here.

When it comes to performance marketing specifically, one thing I've come to learn is that content consumption is one of the most mismeasured metrics in performance marketing. CTR became the default metric, but it's the wrong one. For example, static ads require a click to consume the content on the site, but the right metric to understand if the content is consumed is on-page scroll depth. Video ads only require hitting play + the right metric there is the % of video viewed. I've watched companies kill campaigns that were working because they were measuring the wrong thing. More on that here.

This is where I'm most at home, geeking out in this world + is where I want to return my focus. The breadth I've accumulated since while running a full GTM motion, managing a team of 11, + operating as the sole marketing leader has made me a sharper performance marketer. This is because I came to understand how performance channels interact with the rest of the GTM ecosystem in ways that someone who only lived inside ad platforms doesn't. That context is what separates optimizing a channel from optimizing a GTM system.

Related · I stopped trusting LinkedIn's ad recommendations · measuring paid search correctly · making paid search efficient · channels don't die, playbooks do

When I joined Loxo in early 2023 as the first full-time marketing hire, there was no strategy, no playbook, + no systems as I walked into a blank slate. Every decision was mine to make from first principles.

One of the first things I noticed: BDRs weren't calling on PLG signups until day 15. When I asked why, the answer was "we don't call on these until after 14 days so we get credit." Goodhart's Law in real time - the incentive structure was actively working against the outcome we were trying to drive. That was one of several early signals that I needed to align the GTM team around the actual goal before worrying about tactics.

I built the strategy around two principles: revenue-first planning and ecosystem thinking over channel optimization. (If you read any of the articles I share in this, the one on ecosystem thinking is THE one to read)

On planning: I run both a top-down and bottoms-up model every year (honestly every quarter). Top-down works backward from the revenue target through win rate, ACV, and funnel conversion rates to tell me what leading indicators need to look like. Bottoms-up pressure-tests whether our resources can actually produce those numbers. When they don't line up, that becomes a leadership conversation where we lay out the goal, what hitting it requires, + the gap. In real time I watch the model closely enough to diagnose early when something's off and why, which is what lets you adjust without overreacting. I wrote about annual planning and what it actually looks like in practice.

On channels: I killed paid search early. At ~$100 CPCs with a 6% demo conversion rate, that's $1,667 CPL. At a healthy 25% win rate from demo to close, ad CAC was $6.7k on a $3.6k ACV. The unit economics didn't make sense, so I moved that budget into channels that compound. The harder conversation was changing the BDR payout structure on PLG signups. Nobody loves a comp change, but leaving an incentive in place that was actively suppressing pipeline wasn't an option. Once we rolled out the updated plan, PLG pipeline doubled in the first week.

Related · when common practice buries common sense · 5 questions to GTM executive alignment · the unsexiest growth lever

I think about growth as a portfolio - some bets are incremental improvements on things already working, while others are longer-term experiments where the hypothesis is strong + the upside is high, but the timeline is TBD. The job is resourcing both simultaneously.

SEO/AEO: Our ICP was actively searching for answers to problems our platform solves but we weren't showing up. I partnered with an agency and did something counterintuitive early: cut our highest-traffic keywords because they were broad, mostly irrelevant to what we sold, and not driving pipeline. Most agencies would have doubled down on the traffic numbers. I moved them toward lower-volume, long-tail, higher-intent searches that matched what our actual ICP searched for. Clicks stayed flat and in some months declined, but demo requests climbed, pipeline from SEO/AEO grew at 2x the rate of clicks, and brand search volume went up 23% in three months. I documented the whole thing across a six-part public series in real time.

Brand and entertainment as performance: We built a campaign around a "NOW That's What I Call Recruiting" spoof - fake 80s/90s infomercials, recruiting-themed song renames ("Loxanne", "Boolean Jean"), a quiz, and tour merch. Two TV commercials with an entertaining angle rather than a product-feature approach. A "Recruiters Against Humanity" direct mail package with a custom card game, drink recipes, and kettle corn - pattern-breaking in a space where most direct mail is a one-pager with a QR code. None of this ever runs cleanly through an attribution model, but the downstream signals told the story: deal velocity, win rates, and prospects arriving to demos already knowing who we are.

How I structure experiments: For bigger bets I use the methodology built into Affect - isolate the lever, benchmark before the experiment starts, track what actually moves against the baseline. For faster tests I stay anchored to the real-time models I've built. I wrote about the cost of waiting for perfect clarity and what it means to actually be data-informed vs. data-driven.

Related · back in the weeds · market the problem not the solution · standing out · the brand coefficient

The way buyers research has changed more in the last two years than in all the prior years of my career. When AI becomes the primary research tool for your ICP, the question shifts from "how do we rank" to "how do we show up in the answers AI gives our buyers."

Competitor counter-play pages: We started seeing prospects and customers citing factually incorrect statements about Loxo due to libelous claims from two newer competitors spreading through LLMs as part of the answers buyers were getting when researching the space. The challenge was responding credibly without making counter-libelous statements + doing it in a way LLMs would recognize and cite. I worked through the approach with our CEO and we built dedicated fact-check pages: loxo.co/spott-fact-check and loxo.co/atlas-fact-check. After launching, the incorrect statements decreased and LLMs started surfacing our responses above the competitors' claims. We flipped the channel from costing us deals to driving handraisers.

How we track AI visibility:

I've been putting less weight on unbranded keyword CTRs, click traffic from SEO, + click-to-conversion attribution

I've been putting more weight on AI visibility + citations, brand search volume trends, brand lift over time, + self-reported attribution citing "web research" or "ChatGPT"

Deprioritizing clicks still feels counterintuitive as it goes against decades of SEO muscle memory, but our best SEO work now might never generate a click + that's fine. Like social, we're optimizing for memorability. When the need comes to a head, they come straight to us. I wrote about getting CEO buy-in on this shift and what I call the alligator effect which is the gap between visibility and clicks that most teams are misreading as decline.

We use Profound to track AI visibility. Our content strategy is now explicitly built around citability and findability. I track brand search volume as a leading indicator of AI-driven recall.

When I read about what you're doing with Glass, AirOps, and the AEO strategy at Ramp, I recognized the thinking as I've been building toward the same place from a different angle.

I work directly with our CRO on annual planning and budgeting. We share the revenue number + rather than dividing credit by lead source, we account for how marketing investment enables sales effectiveness and vice versa, then resource both teams based on what it actually takes to hit the shared goal. Each week I report to leadership on top and middle of funnel across marketing, sales, and BDR. All three leaders are aligned on what's expected, so there are no surprises.

At Refine Labs I consistently built stronger relationships with CROs than CMOs - we cared about the same thing. My standard alignment question early in any engagement: "Would you rather hit your MQL target but miss your revenue target, or miss your MQL target but hit your revenue target?" It forces leaders to say out loud what actually matters, and the conversation flows from there. I wrote about the five questions that get GTM teams aligned and what measuring success actually looks like.

With finance, the top-down and bottoms-up models I build are anchored in historical performance data, realistic benchmarks, and a clear view of what the goal requires versus what resources can produce. When there's a gap, I surface it early as a trade-off conversation rather than showing up to planning asking for more without context.

I report directly to our CEO who's very product-oriented + hires leaders to lead. I've built trust by operating with enough autonomy that he doesn't have to wonder what's happening and enough transparency that he always knows. I send him a weekly EOW update on team priorities and progress. Monday I send a leadership update on top and middle of funnel relative to goals. He knows I'm watching what he cares about and that problems won't become surprises.

I joined Loxo in January 2023 as the only marketer. Hired three in September 2023, two more in 2024, and five in H2 2025. The team is now eleven people.

I look for two things above everything else: genuine curiosity and a growth mindset, or as a former colleague of mine called it - "figure-it-out-ability", AKA people drawn to hard problems + don't need the answer handed to them before they're willing to try. I've also found a strong pattern with people who have athletic backgrounds: strong team players, tenacious when they commit, + wired to see things through.

Beyond that I look for deep knowledge in one specific area paired with a "superpower" that creates leverage across the whole team. Our demand marketer is deep on paid social and paid search, but his superpower is HubSpot and marketing operations. When the team was four people, that meant he could orchestrate the systems behind the scenes that made everyone else more effective.

My Head of Brand and Content is the hire I'm most proud of. Came in as a copywriter. I saw early she understood brand and content fundamentals at a level most people with the title for years don't. She now manages a team of three, is my "number two", + her output bar for quality (she's still doing the work) is unparalleled. I never have to follow up on something I've asked her to handle.

I know you + Karim share the spiky over generalist view strongly, so there's full alignment here.

In January 2026 I was sitting with an ambitious growth goal and a familiar problem: I couldn't just keep pushing ad dollars to hit it, but I didn't have a clean way to diagnose where our real constraint was, model the impact of fixing it, or track whether what I changed actually moved anything. So I built it. (I heard you share on a podcast that you've also read The Goal, so this is pulling from that methodology + I wrote an essay about that here)

Affect (affect.so) is a marketing diagnostics and experimentation platform built around the same systems thinking I'd been applying since auditing 100+ funnels at Refine Labs. The diagnostic framework is hand-curated domain expertise - 13 named failure patterns across SLG and PLG, each with a specific root cause, prioritized fix list, and industry benchmarks. The app consists of four components:

• Diagnose - identifies which of 13 failure patterns your funnel is showing, benchmarked against peers
• Forecast - models ARR impact before you commit to a play, so you're running experiments with a hypothesis rather than a hope
• Affect - GTM plays and outcome tracking; you log what changed, it measures what actually moved against your pre-change baseline
• Bayes - the AI layer (SLM) across all of the above, answering questions in plain language using your own data as context

Stack: React, Supabase, Vercel, Stripe, PostHog, Recharts, Claude API powering Bayes. Built almost entirely in Claude Code. I'm not an engineer - I built it anyway because I've been obsessed with this problem for years and finally had the tools to do something about it. It's live. It has paying customers. It started as scratching my own itch.