Notas del episodio
This episode of the B2B Orchestration Podcast is a reality check for anyone still pitching marketing plans on vibes. The host sits down with Raja of GNW Consulting to unpack why the old playbook — "trust me, we'll estimate this much and hit it" — no longer survives contact with today's buying committees.
The core tension: a generational shift in how leadership evaluates marketing spend. Raja argues that newer executives, having grown up inside the "marketing bullshit" of inflated PowerPoint wins and vague Q-over-Q narratives, simply won't accept forecasts without hard justification. A 5-10% miss against forecast is tolerable; a 30-40% miss demands a real explanation. The days of throwing out a number and moving on are over — every dollar now needs a defensible line back to outcomes.
That shift reshapes how Raja's team engages clients. Rather than accepting a stated goal at face value, they reverse-engineer it: pulling apart what's actually driving current results, what's working, what isn't, and building a realistic target from that evidence — instead of inheriting an inflated number a leader is already boxed into defending.
The conversation's sharpest turn is on AI. Raja is blunt that most of what passes for "using AI" in marketing — email copywriting, subject lines, LinkedIn content — is low-value noise. The real unlock is AI applied to data analytics and reverse engineering: querying account and revenue data conversationally (he describes running ad hoc analysis in Slack instead of building reports in Salesforce) to get faster, cleaner answers about what's actually driving pipeline. That's the "left brain" use case he trusts.
The "right brain" example is more novel: Leo, a GEO (generative engine optimization) agent his team built for clients. It doesn't write content — Raja is firm that AI-written content is still weak — but it audits and edits existing content from a technical/schema standpoint so LLM crawlers can actually parse and cite it correctly. He describes clients going from zero AI-driven citations to several within a fairly short window, crediting the win to how the content is structured for machine readability, not to what was actually written.
The practical takeaway: stop treating AI as a blanket solution and start piecemealing it into the specific parts of the workflow where it measurably compounds — data interrogation and technical content structuring — rather than the parts (strategy, writing, judgment) that still need a human running the reverse-engineering.