The First-Principles
Funnel

This campaign didn't start as “let's run a nurse line campaign.” It started as a gap found by pulling three unrelated data sources and putting them next to each other.

• 01

  Search demand. Search demand for the core topic cluster was substantial. The company's own organic content was capturing a small fraction of it.

• 02

  Unconverted traffic. A separate set of pages was pulling real, engaged traffic with literally nothing for a visitor to convert on.

• 03

  Answer-engine visibility. How often the company got cited in AI-generated answers was strong on core questions and effectively zero on adjacent ones, marking exactly where the open ground was.

Together those three facts make a specific, arguable case: real demand exists, it's not being captured, and there's already a warm signal, existing engaged traffic, sitting there unconverted. That's the brief before there was a brief.

Not a new topic and not a new channel. A funnel built around what was already true: real demand, traffic that was already showing up unconverted, and one proof asset ready to carry it.

That's the two-sentence version. Everything from here is the audience, channel, and asset architecture built to act on it.

Every build decision traced back to one page: what we're doing, for whom, and why.

The anchor asset was an existing, cleared proof point, a real case study already approved for external use, not something built new for this campaign. Net-new logo acquisition and existing-account expansion ran as two distinct, sequenced motions instead of one blended ask, expansion staged to follow once net-new proved out. The keyword strategy stayed narrow, high-intent terms only, with adjacent topics used as audience angles rather than diluting the core theme with content nobody searching for the core problem would find relevant.

The buyer isn't one person. My default is to map the decision by role, not just by company type, and give each one a specific pain the offer actually solves, not a generic value statement reused across all of them.

Every channel got a stated job and a stated destination, not a scatter of disconnected placements.

Paid search and display carried the primary and retargeting load. Paid social carried account-based reach into named accounts. Organic content earned attention upstream of search, feeding people down into the paid funnel instead of competing with it. Email carried a short nurture sequence for engaged non-converters. Sales outreach ran alongside the ads, working from the same proof points the campaign itself was built on. Everything funneled to one of two conversion points, never a scatter of one-off landing pages nobody could account for.

I built a single map covering every content asset the campaign would touch, tagged to where it's used and whether it existed yet.

A gated lead magnet, existing proof points already cleared for external use, top-of-funnel content, sales collateral, and the destinations they all point to, one map, one rule: if it's not on the map, it's not being used. That single rule kept the campaign from quietly sprawling into ad hoc assets nobody could account for later.

How a lead moves from first touch to pipeline, and how nobody who slips through gets wasted. My default is an explicit flow with loop-backs, not a straight line.

• 01

  Multiple entry points converge on a small number of conversion points, never a scatter of destinations.

• 02

  A visitor who lands but doesn't convert isn't lost. They're routed into a retargeting pool and brought back to the same conversion point.

• 03

  A visitor who converts moves through a short, fixed nurture sequence where every touch is a new angle, never a repeat of the last one.

• 04

  No response after the nurture sequence doesn't mean the record is dead. It returns to the retargeting pool and re-enters on the next click or fresh intent.

• 05

  Once in Sales, a lead is qualified against an agreed definition. Stalled leads, engaged but not progressing, get a dedicated re-engagement path instead of being left to expire.

• 06

  One measurement layer runs the whole way through: cost per qualified lead, blended and by channel, reviewed on a fixed cadence, with a hard checkpoint at a set point for the scale, hold, or fix decision.

Cost per lead decides, every time. The same rule runs on every channel regardless of format.

Launch multiple versions at once, let cost per lead judge them, shift budget to the winner and pause the losers, then swap a fresh version in for whatever just lost, and keep going. No channel gets a single, static ad and calls it done.

I modeled three funding tiers, with one clear recommendation and a stated reason for rejecting the other two, not left implicit.

• 01

  Minimum tier. Proves the mechanism captures leads at all, at the lowest possible cost, deliberately too thin for the full channel mix to work.

• 02

  Recommended tier. The smallest budget where every channel can actually function together instead of one starving the others.

• 03

  Full-scale tier. The mix at real scale, more than needed to prove the model, appropriate once it already has.

The ramp was modeled to be honest about its own learning curve: cost per lead starts high while the channels are still learning and falls toward the target run rate over time. A fixed review sits at the end as a hard decision gate, so the budget wasn't open-ended and didn't need to be defended qualitatively every month.

This campaign didn't exist in isolation. My default is to design each new campaign with the other live campaigns in mind, rather than fighting them for the same budget or the same data.

That means reusing infrastructure that's already proven, like a shared retargeting pool, instead of building a duplicate one before there's enough volume to justify it, while keeping keyword targeting and attribution structurally separate wherever two campaigns could plausibly compete for the same lead. A lead's source stays legible, and one campaign's performance never gets credited to another.

The campaign launched. Not a pilot channel, not a soft test, the whole plan, every channel live at once, built from the strategy, the audience work, the channel plan, the asset map, and the testing plan.

My role ended shortly after launch, before enough time had passed to pull real performance data. What's here is the strategy and the finished creative, not a results readout, because the results were never mine to see.

• 01

  Finding an opportunity by cross-referencing data sources, not starting from a channel or a format I wanted to use.

• 02

  Sequencing discipline: proving one motion before layering on a second, even when the second is tempting to start immediately.

• 03

  A complete, mapped channel plan and asset map with an explicit rule instead of a loose list of ideas.

• 04

  A lead-journey design with built-in recovery loops instead of a funnel that treats a non-conversion as a dead end.

• 05

  A budget recommendation that states why the other options were rejected, not just why the chosen one was picked.

• 06

  Designing one campaign in the context of the others already running, reusing what should be reused and separating what needs to stay separate.