Marketing Advisor
How NexusPoint Built a 234-Source Research-Backed Marketing Brain That Answers Any Channel Question in Seconds
Client
Challenge
Most agency founders and founders running their own marketing face the same impossible task: they need to stay current on what actually works across 6+ channels, but the advice changes every quarter and most of it comes from people selling something. The default approach is to follow a handful of marketing gurus, skim their newsletters, and apply their frameworks as gospel. The problem is that what worked in 2024 is often penalized in 2026. LinkedIn changed its algorithm. Gmail started filtering AI-generated copy. Instagram's ranking signals shifted from likes to sends. And the frameworks that built careers, "spray and pray" volume, open rates as a KPI, "raise your prices" as universal advice, are now actively harmful. Four specific problems: No single source of truth. Marketing advice lives across newsletters, podcasts, Twitter threads, blog posts, LinkedIn comments, and paid courses. There's no unified, cited repository of what actually works in 2026, just conflicting signals from people with different incentives. Frameworks age without warning. Hormozi's "Rule of 100" (send 100 DMs/day until something works) was sensible in 2022. In 2026, it gets your domain burned and your LinkedIn account restricted. But most advice still repeats these frameworks without dating them. Channel-specific knowledge is fragmented. The person who knows LinkedIn's 360Brew algorithm inside out doesn't know Instagram's sends-based ranking. The cold email expert doesn't know Reels retention curves. Getting a complete marketing picture means consulting 3-5 different specialists or sources. No decision framework for "should I do this?". When a founder asks "should I run LinkedIn ads?" or "what's a good cold reply rate?", the honest answer requires current benchmarks, positioning context, and channel trade-offs. Most advice gives a generic yes/no without the numbers behind it. NexusPoint built the Marketing Advisor to solve all four, a single research-backed skill where every recommendation traces to a cited 2026 source and every benchmark has a number behind it.
Goal
The Marketing Advisor is a Claude skill backed by a NotebookLM-synthesized research corpus of 234 unique 2026 sources. It is not a framework library. It is not a prompt. It is a cited, grounded marketing intelligence system where every recommendation leads with the 2026 number, then the tactic. One command triggers any mode. Seconds later: A channel strategy grounded in current benchmarks ("cold email reply averages 3.4% in 2026; top decile is 8-12%. To get there, cut the list and personalize line one.") A content calendar with platform-native format specs, hooks, and CTAs An ICP definition with 5-layer scoring and validation steps An offer positioning using the Value Equation with 2026 pricing data An automation blueprint with tool names, data flow, and measurement plan The clean split: The Marketing Advisor owns strategy, campaigns, content calendars, ICP, channel selection, offer positioning, and benchmarks. The actual 1:1 outreach copy, connection notes, cold email body, DM sequences, is owned by the Sales Playbook skill. The Marketing Advisor frames the strategy, then hands off the copy. This keeps each skill focused and testable.
Result
The Marketing Advisor is a research-backed marketing intelligence system where every recommendation traces to a cited 2026 source. It is not a chatbot that "knows marketing", it is a Claude skill with a 234-source NotebookLM corpus, 12 distilled reference files, and a live query fallback that appends its findings to the corpus for reuse. The architecture, deep web-research into a NotebookLM notebook, structured extraction via ask --json into playbooks with inline citations, a channel benchmarks scoreboard, a kill list of retired practices, and a clean strategy/copy split with a sibling skill, is not specific to digital agency marketing. The same system works for: Any service business that needs to stay current on channel-specific benchmarks and adjust their marketing as platforms change their algorithms A SaaS company researching ICP, positioning, and channel strategy for a new product launch A consultancy building a research-backed go-to-market library that accumulates knowledge across engagements An agency that needs to give current, defensible marketing advice to clients without relying on aging frameworks Any business that's tired of conflicting marketing advice and wants a single, cited source of truth for "what actually works right now" The key innovation is that the corpus is not static. The live NotebookLM fallback means the system can answer questions the original research didn't anticipate, and those answers are automatically saved so the next identical question is answered from disk. The corpus compounds.