Hey {{first_name}} ,
A board member asked me last month what our marketing campaign cost in tokens. I had a number for him. He didn't believe me. He was right not to.
The number was honest in the way historical numbers are honest — it was what we'd actually spent. But he was asking the real question: could I tell him what the next campaign would cost? Could I commit to a planning number? The answer was no, and the reason was structural, not personal. I want to walk you through that reason, because I think most CMOs and CFOs reading this are circling the same problem and most don't have language for it yet.
This issue is about that gap. The cost honesty problem in the agentic era. What it actually breaks. And the worldview shift that makes it manageable.
Sreedhar Peddineni,
CEO, GTM Buddy
You're receiving Revenue Activators because you subscribed.
No longer want these? Unsubscribe here.
THE ACTIVATION
Why your AI campaign cost is unknowable - and why that’s not the problem.
For as long as marketing budgets have existed, planning has been a discipline of bounded uncertainty. Paid media is a quoted CPM. Agency work is a quoted retainer. Content production is a quoted day rate. Software is a quoted seat fee. The variance lived on the return side - will the campaign perform? - not on the cost side. The cost half of the equation was always locked.
The agentic era broke that. The biggest variable in most AI-native marketing functions is no longer paid media or content production. It's the AI tokens burning underneath every campaign - and the per-campaign total is the product of three unknowns: how cleanly the AI drafts the work (per asset), which model tier the orchestration layer picks (per pass), and how much new ground the team will cover with the tools by the time the campaign actually runs (per quarter, always upward). Three unknowns multiplied together isn't a forecast. It's a wish.
Most CMOs are responding by scoping down. Smaller bets, smaller variance, smaller returns. Hedging on the input side because they can't model the cost side. That's the trap I want to name, because it's the wrong response - and it took me a while to see why.
“Productivity is a denominator metric. Capacity is a possibility metric. You cannot manage one with the discipline of the other.”
Here's what changed for me. The black-box variance is only a planning problem if you're still running a productivity function. Productivity is a denominator metric - outputs per dollar of input. The dollar sits at the bottom of the fraction. If you can't measure the dollar reliably, the metric breaks, and the planning breaks with it.
Capacity is a different metric. Capacity asks what becomes possible that wasn't possible before. The dollar stays in the equation; it just stops being the central question. The central question becomes: which capacity unlocks are worth the variance, and can I defend that choice clearly to the people who pay for it?
My CMO, Karthi Ratnam, just published the operator-chair version of this argument - honest piece, worth your time - and it surfaced something we've been working through internally for months. The marketing functions struggling most with agentic-era planning are the ones still applying productivity discipline to capacity tools. The functions thriving have made the worldview shift, often without naming it. Issue 2 is about naming it.
One Move for This Week: Sit down with your CFO this week. Don't bring them a revised budget. Bring them a revised framework. Tell them you want to move how you brief marketing spend from a productivity model (cost-per-output) to a capacity model (possibility-per-quarter). Ask them what evidence would let them accept the shift. That conversation, more than any forecast, is the planning unlock for the next 18 months.
THE PROMPT
Framework 1.2: The Persona-Specific Value Translator
From the Revenue Activation Playbook
What it does:
Takes the elevator pitch you built from Issue 1 and adapt it for each buyer persona in the deal. The AI analyses your persona cards (or descriptions), names what each buyer is measured on, and rewrites the core pitch in their language with proof points that fit their function. CFO, VP of Sales, end user - same product, three different conversations.
Why it matters:
A rep who delivers the same pitch to every buyer isn't ramped. They're reciting. The difference between Hayden (85% of quota) and Griffin (10%) in our diagnostic wasn't product knowledge - it was Hayden's ability to shift language, emphasis, and proof points depending on who she was talking to. Same training. Same product. Different translation muscle. That muscle is what Framework 1.2 builds.
Time required:
20 to 30 (depending on the number of personas)
What you'll need:
Your core elevator pitch from Framework 1.1 (Day 1). If you haven't completed Day 1 yet, do that first - you need a solid core pitch before you can translate it.
Persona cards or written descriptions of your key buyers. Ideal: formal persona cards from marketing. Acceptable: a written description of each buyer's role, priorities, and concerns. Minimum: the job titles you need to pitch to.
Claude, ChatGPT, Gemini, or CoPilot. Any of them work.
How to use it:
Start with your Day 1 pitch. Have it ready to paste.
Gather your persona materials - cards, descriptions, or at minimum the titles.
Open your AI tool. Either start a new conversation or continue from Day 1 if you want to maintain context
Upload persona cards if you have them, or be prepared to describe each persona.
Copy and paste the full prompt below. Include your core pitch in the Inputs section.
Review the persona summaries the AI returns. The AI summarises each persona before generating pitches. Correct any misunderstandings here.
Review each persona-specific variation. Does it sound like something that persona would respond to?
Note the trigger phrases. These are gold. Write them down and listen for them in real conversations.
Practise switching. Use voice mode to practise pivoting between personas. Have the AI play different buyers in sequence.
Copy-Paste this Prompt
You are a B2B sales messaging strategist who tailors value propositions to different buyer personas.
CONTEXT
One pitch doesn't fit all buyers. A CFO cares about cost savings and ROI. A VP of Sales cares about quota attainment and rep productivity. An end user cares about ease of use and daily frustrations solved.Same product. Same company. Three different conversations.
YOUR JOB
Translate a core value proposition into persona-specific language. Emphasise what THIS buyer cares about, using words THEY would use, connecting to outcomes THEY are measured on.
THE STRUCTURE
For each persona, create a pitch variation that:
Opens with their world - reference their role, their priorities, or a challenge specific to their position
Frames the problem in their terms - use language and metrics they care about
Connects your solution to their success metrics - how does this help THEM win?
Proof that fits - results or examples relevant to their function
EXAMPLE
Core Pitch (from Day 1):
“My company, [Insert Company Name], helps B2B sales teams hit quota faster by making training actually stick. Many companies try to accomplish this by running generic workshops and hoping reps remember what they learned, which often leads to wasted training budgets and reps who revert to old habits within weeks. We take a different approach, combining the analytical power of AI with supportive human coaching. The AI analyses actual sales calls to identify exactly where each rep needs to improve. The coaching ensures they actually change. Together, our clients have cut ramp time in half and improved win rates within 90 days.
CFO Version: “Sales training is one of the biggest line items that's hardest to tie to results. Most programmes show completion rates, not revenue impact. We take a different approach - using AI to measure exactly which skills are improving and how that correlates to closed deals. Our clients typically see a 3–5x return on their training investment within the first year, with clear attribution from skill development to quota attainment.”
VP of Sales Version: “Your reps are sitting through training, but are they actually getting better? We use AI to analyse real sales calls - not self-assessments - to pinpoint exactly where each rep is losing deals. Then we coach on those specific gaps. The result: ramp time cut in half and measurable improvement in win rates within 90 days.”
End User (Sales Rep) Version: “Most sales training feels like a waste of time - generic advice that doesn't apply to your actual deals. We're different. We analyse YOUR calls, find the specific moments where deals stall, and give you targeted practice on exactly what you need. No more sitting through training that doesn't apply to you.”
YOUR TASK
Based on the core pitch and persona information I provide, generate persona-specific pitch variations. For each persona, provide:
The adapted pitch (75–100 words)
Key messaging shifts - what changed from the core pitch and why
Trigger phrases — 2–3 phrases this persona is likely to respond to
STEP 1: ANALYSE MY MATERIALS
I may have attached persona cards, ICPs, or other documents describing my buyers. If so, analyse these first and extract: titles and roles in the buying process; what each persona is measured on (KPIs, success metrics); their likely concerns and objections; language patterns - how do they talk about this problem? Summarise what you learned about each persona in 2–3 sentences before proceeding.If I haven't provided materials, ask me to describe each persona before generating pitches.
STEP 2: REVIEW THE CORE PITCH
I've provided my core elevator pitch (from Day 1) below. Identify the core value proposition, the primary differentiator, and the proof points available.
STEP 3: TRANSLATE THE PITCH
For each persona, adapt the core pitch by shifting emphasis to their priorities, using their language and metrics, connecting to their definition of success, and selecting proof points most relevant to their role.
INPUTS:
Core Pitch: [Paste your Day 1 elevator pitch here]
Personas to target: [List the personas - e.g., CFO, VP of Sales, End User - or attach persona cards]
Additional context: [Any industry-specific language, known objections, or priorities for these buyers]
Why this prompt works
Most reps don't naturally make the persona shift. They deliver the same pitch to everyone and wonder why it lands with some buyers and not others. The four shifts the prompt enforces - opening, problem framing, solution connection, proof selection - are the moves a great rep makes instinctively. The prompt makes them explicit, so a newer rep can practise them with structure before they become reflex.
The other thing the prompt does well: it produces three variations, not one. Most reps won't use any version verbatim. They'll take the phrases that feel natural and combine them. That's the point. The variations are starting material; the translation muscle is what they build by working with them.
If you ship this with your team and it works, reply to this email. I'd love to see how you customised it.
You're receiving Revenue Activators because you subscribed.
No longer want these? Unsubscribe here.
THE ACTIVATOR
Timea Bara
VP, GTM Enablement at Everlaw | San Francisco
Second spotlight in the Wednesday Women × GTM Buddy Revenue Activators series
Timea Bara is one of the most-respected revenue enablement leaders in the Bay Area, and one of the people whose careers most precisely trace the arc this newsletter is about. Fourteen years in revenue, starting in sales, then to sales training at Meltwater, then seven years at LinkedIn building out the Marketing Solutions Sales Readiness team from its second hire, then Head of Global Revenue Enablement at Nextdoor, then Multiplier, and now VP of GTM Enablement at Everlaw.
Alongside the day job, she co-leads the San Francisco / Bay Area chapter of WiSE - Women in Sales Enablement, mentors at RE:WORK TRAINING, and serves as an investor at inVest Ventures. She is, by every measure, exactly the kind of executive whose work shows up in revenue outcomes and rarely in revenue narratives. That recognition problem - the architects of revenue infrastructure who get coded as “support” - is what the Revenue Activators series exists to fix.
Timea's POV on what activation actually requires:
Your network is your net worth, and relationships matter even more in the age of AI.
That line is hers, and it's the through-line of how she's built her career. Timea's view is that the next decade of enablement is going to be about combining AI horsepower with the human skills AI is making more valuable, not less - trust, judgment, the relationship work that closes deals when the analytical layer has done its job. She has been arguing this for two years before the consensus arrived at it.
If you want to see Timea articulate the arc from sales training to revenue activation in her own words, her 2025 talk at the Sales Enablement Collective Summit in San Francisco - and her earlier appearance on Gong's Reveal podcast on personal development and learning culture - are both worth your time.
Follow Timea
THE SIGNAL
Three external links worth your time this week
1. “The Black Box of Token Pricing Is Making Me a Worse CMO” by Karthi Ratnam
My CMO just published the operator-chair version of the argument I made in Honest Math. She names what the cost-unforecastability problem actually breaks at the CMO level - the smaller bets, the degraded CFO conversation, the slower decisions — and then makes the move that took me months to land internally: this is a productivity-framework problem on a capacity-framework function. If you only read one piece on agentic-era marketing planning this quarter, read this one. It's honest in a way most CMO writing isn't.
2. Bessemer State of the Cloud - Revenue Per Employee benchmarks
Look up the Revenue Per Employee benchmarks across public SaaS companies. The chart that matters: the spread between the top quartile and bottom quartile has widened materially since 2023. AI-native operators are pulling ahead, and the gap is showing up in efficiency ratios, not in growth rates. That's your competitive context heading into 2026 planning.
3. Anthropic on consumption pricing and prompt caching
If your team is going to operate intelligently inside the variance I described in The Activation, they need to actually understand the cost structure of the tools they're using. Anthropic's documentation on prompt caching and batch processing is the cleanest practitioner explanation of how to manage AI cost at the operational level. Not a marketing read - an engineering read that marketing and ops leaders should know.
Got something we should signal in this issue? Reply with it. We read everything.
THE SKILL
Token economics, explained for marketing and ops leaders
You're going to spend the next 18 months managing AI spend more carefully than you've managed any cost line in marketing's history. Here's the minimum vocabulary you need to do it without being misled by a vendor's pricing slide.
How AI cost actually works
Every interaction with a large language model has two cost components: input tokens (what you send the model) and output tokens (what the model sends back). Output tokens cost roughly 3–5x what input tokens cost. That asymmetry matters because most cost-control conversations focus on “how many words am I sending?” when the real cost lever is “how much am I asking the model to write?”
Model tiers compound the math. A frontier model can cost 30 - 60x what a smaller model costs per token. Most production agentic workflows mix tiers - cheap models for retrieval and routing, expensive models for reasoning and drafting - and the per-task cost is determined by the mix, not by any single model's pricing sheet. When a vendor quotes you “AI included,” ask which tier.
Then there's consumption growth, which is the slow-burn version of the problem. Per-token costs have come down meaningfully year-over-year. Total spend has gone up anyway, because every team that gets comfortable with the tools asks the tools to do more. The same campaign that used the visual-generation agent for hero images last quarter is using it for every supporting graphic, every social variant, every reel storyboard this quarter. Capability didn't get more expensive. The use case expanded.
Why this matters for you
If you're paying a vendor for “AI features,” you're often paying for the cheapest tier of model running on retrieval tasks. The genuinely expensive reasoning work - the work that actually moves the needle - typically requires either a higher tier or an entirely different platform. Ask which.
Cost forecasting tools that quote you a per-call number are honest about observability and misleading about prediction. Per-call cost is real; total spend variance is structural. The variance lives in usage growth, not in unit cost.
The CFO question worth being able to answer: “What's our cost per generated artefact, end to end, including all agent passes?” If you can't answer it, you're observability-only - and the worldview shift in The Activation is the move you need to make this quarter.
The question to ask your vendors:
“What's your cost per generated artifact, end to end, including all agent passes - and how does it change across the model tiers you use?” If they can't answer it, they're observability-only. If they can answer it but the number is suspicious, they're using a cheap tier for the work you'd want a heavier model on.
That's issue two.
Reply and tell me what landed and what missed. I read every response. The early issues get sharper because of what readers send back.
Three things you can do right now:
Forward this email to your CFO. The capacity-vs-productivity conversation is the highest-leverage one you can have with them this quarter, and this issue is built to start it.
Take the Revenue Activation Assessment. It diagnoses where your team's capacity is most constrained across the Five Levers — revenueactivator.ai/assessment
Follow Timea Bara on LinkedIn. She's the second Revenue Activator in the series and her POV on AI + human skills is worth your feed.
See you in two weeks.
Sreedhar Peddineni
CEO and Co-Founder, GTM Buddy
LinkedIn