Spearleaf · Position Zero Playbook v10 · 2026-06-16
Start here Changelog
Quickstart

Run a digital-PR sprint for AI and brand visibility

Turn one original-data story into many journalist-ready, locally-angled pitches so LLMs name your brand alongside the problem it solves.

On this page

Run this when a brand ranks fine on Google but is invisible inside LLMs, or when you need earned media that teaches AI to associate the brand with the problem it solves. The logic: Google ranks pages, AI ranks entities and builds consensus from what the web repeatedly says about you. The method here is Dawood Bukhari's NARC plus digital-PR engine: build one original-data story, score it before you spend a dollar, then snowball it into dozens of local pickups while running HARO with co-occurrence discipline. Earned media is the lever because (per the talk) about 85% of AI brand mentions come from earned media, not brand-owned content.

Use it for a sprint (one dataset, one wave) and repeat in waves on the same data with new narratives. Do not start here for an entity that has no consistent positioning yet. Lock that first (see step 1).

Do this now

  1. Lock the entity (NARC, the N). Answer three questions the same way on every surface: who you are, who you serve, what you solve. Same positioning, same expertise signals, same brand story everywhere. Inconsistent positioning poisons every mention that follows.

  2. Set the co-occurrence target (NARC, the C). Write down the one problem the brand solves. Every story, quote, and pitch in this sprint must name the brand alongside that problem. If a pickup names the brand but not the problem, it is wasted training signal.

  3. Map the competitor mention gap (lever 1). Count each competitor's mentions and treat it like link velocity: you need MORE mentions than the top competitor, all on the same narrative and same problem. Set the wave's target above the highest competitor count.

  4. Pick a topic with universal emotion plus credible data. High-arousal topics win (job security, holiday travel, dangerous intersections). Bad: "Employment statistics by state." Good: a topic people actually care about and would reshare to signal identity. If nobody cares, stop.

  5. Engineer the dataset. Use a credible public source (examples from the talk: BLS, Bureau of Transportation Statistics, STSI). Define a clear methodology and normalize per capita (for example per 100,000 residents) so small populations do not distort the rankings. Keep the full dataset and method on hand: journalists (and sometimes government offices) will ask you to defend it.

  6. Prepare three angles up front. Data angle (the core, used in every story), human angle, economic angle. Having a second hook ready means a story still lands if the first angle misses.

  7. Write a scroll-stopping headline. Stop-the-scroll logic. Good: "1 in 4 Americans Now Fear Layoffs More Than Recession." Weak: "Study Examines Layoff Concerns." If you can predict the whole story from the title, it fails the novelty test (movie-spoiler rule).

  8. Build one visual that explains the whole story in about 5 seconds on mobile. One map or graph. Adding the visual lifts pickups a lot. (The speaker mentioned a Map Graphic Creator automation to make maps input-driven instead of hours of manual work. Availability unknown to us. Mark as a tool to ask about.)

  9. Check timing. Confirm search demand is stable or rising and there is no competing major news cycle, and aim to be first to report. Example from the talk: publishing a layoff study during earnings season (good) versus during election week (bad).

  10. Assemble the journalist package. Target: a journalist can write the story in 15 minutes. Include the clean methodology, downloadable visuals, 3 to 5 quotable stats with specific numbers (not ranges), two expert quotes, and a pre-written, copy-paste-ready narrative. QC by having your own team draft the article as a test.

  11. Score it on the Viral Readiness Scorecard before distribution. Ten sections, each Pass / Partial / Fail: Emotion, Sharing, Novelty, Credibility, Hook, Timing, Visuals, Remix, Journalist Ease, Snowball. Rule: minimum 8 passes and no more than 1 fail. If it does not clear the bar, it does not go out. Apply this before research, design, AND outreach.

  12. Localize the dataset (snowball). Lead with the national angle, then break the data into per-state or per-city rankings (top 10 states becomes 10 local stories). Send each localized story to journalists in that specific location. Local angles get picked up the most.

  13. Push for video. Target at least one local-TV video mention per 10 links by leaning on the local angles.

  14. Maximize spread and authority distribution (levers 2 and 3). Aim for many unique, diverse domains, not just repeats from the same few, and include some high-DR placements in the mix so quality accelerates trust velocity. Confirm every pickup credits the brand by name and keeps the brand-plus-problem co-occurrence.

  15. Run new waves on the same data. As coverage compounds, relaunch the dataset with a new narrative (example from the talk: "most stable" after "most unstable"). Pre-plan these remix angles before the first launch (industry, regional, productivity, and similar).

  16. Run HARO as narrative injection, in-niche only. On onboarding, filter the client's relevant niches. Answer HARO / journalist queries only within those niches plus a few general categories (entrepreneurship, marketing). Hold co-occurrence with the core problem on every answer. Do not chase unrelated questions: scattering across many niches hurts AI visibility.

  17. Measure AI visibility, then iterate, reinforce, repeat. Each placement teaches LLMs "this brand = authority on [the problem]." Re-check AI mentions, reinforce the gaps, and loop back to the next wave.

Pitfalls

Source

Distilled from Dawood Bukhari (CEO, Digital Web Solutions), "Beyond Rankings: Building Brand Visibility in the Age of AI," Day 3 of the 2026 SEO Spring Training conference. Conference-derived from session notes; figures and tool availability are as stated in the talk and not independently verified.

Connect it