Enterprise Account Strategy With AI: The Good, Bad and Ugly
Key Points
- Both of my savings levers were rounding errors to the CEO.
- The Frame Trap: AI polishes whatever frame you hand it.
- A friend's reframe turned it into a margin plus growth plan.
Key Points
- Both of my savings levers were rounding errors to the CEO.
- The Frame Trap: AI polishes whatever frame you hand it.
- A friend's reframe turned it into a margin plus growth plan.
I was proud of my account plan for a large public company, polished from v3.2 to v3.28 with AI agents checking every number. It had two levers. The first handed 20 percent of their time back to 10 R&D people at $250K each, which I estimated was worth about $500K. The second moved part of their AI workload to cheaper models, which I estimated would cut COGS by roughly $22M a year.
Unfortunately, both were rounding errors at their scale, and only loosely tied to the company's current strategy and business issues.
AI agents gave me my deepest account research yet
I ran three Grok Bot agents. One researched the account, one kept a living account app current, and one played their stakeholders so I could rehearse objections. They chewed through the 10-K, earnings calls, analyst day transcripts and the CEO's LinkedIn posts, and every claim got a label: Real, Illustrative, Hypothesis or Projected, so none of my estimates could pass for a sourced fact. (Also read: The Proof Premium)
I sized the first plan to my own product
I started from what the AI platform I was positioning did well, a classic salesperson move I'd mock in anyone else. People I trust took it apart quickly, without softening a word.
| What I did | What I realized |
|---|---|
| Sized the COGS lever at roughly $22M a year | Still a rounding error, and not aligned to a CEO agenda built on growth |
| Built a $500K R&D time card: 10 people, $250K each, 20% of their time | Even smaller thank the COGS lever at that company's scale |
| Started from my product | Their 2026 and 2027 initiatives were the right start |
| Packed in 27 dense slides | The slides were crowding out the conversation |
| Wrote in an engineering build champion | I had not earned that person yet |
| Thin discovery, no C-level access | I was pitching before I had listened |
One of the people I trust, told me to start from the company's own 2026 and 2027 initiatives and work backward to revenue gained, revenue protected and cost saved. I was sure I already had. I'd started there, wandered into the cost-saved corner with my ~$22M COGS lever and parked the whole plan in it, until my friend dragged me back out to see the bigger picture.
AI polished the wrong plan twenty-six times
The Frame Trap is the pattern in which AI keeps improving work inside a frame nobody has questioned, so every version looks more convincing while the frame stays wrong.
Picture a hiker in fog who only ever steps uphill. They will reliably reach a summit, but only of the hill they started on, and in fog a foothill could feel like Everest. Every version was another confident step up my foothill, with me planting flags.
The agents caught stacked savings the math didn't support, an overstated case study, a wrong executive title and an account app that disagreed with the deck. Any one of those could have torched every number on the page, yet no agent asked whether the page was about the right problem, because I never gave one that job.
Another AI pass costs almost nothing, while reframing costs a meeting and someone senior saying the problem is wrong.
The fix turned a margin pitch into a C-level growth plan
The rework turned my product pitch into a strategic, C-level plan that opens with one line: "Save ~$22M (margin). Make $300 million (growth)." My reworked pitch answers my failure table one row at a time.
| Before (v3.28) | After (CEO version) |
|---|---|
| Started from my product | Starts from their 2026 and 2027 initiatives |
| Two rounding-error levers | Save $22M COGS Make $300M Revenue |
| 27 dense slides | 9 sparse slides |
| Pitching before listening | A first slide titled "What we heard: your agenda" |
| An assumed engineering build champion | A champion found and earned before any bakeoff |
That listening slide is built on the CEO's own priorities, including their public view that the task should decide the model, this shows I am listening. An org map shows AI agents expanding every team's capacity instead of cutting headcount, because leaders have to sell the plan to their own people. (Also read: Enterprise AI Is an Organizational Design Problem in Disguise) Before-and-after workflows move Customer Success from firefighting to protecting and growing revenue, measured in dollars kept and expanded, and a loop turns customer feedback into shipped features faster.
The 27-slide version is still in a folder, a polished summit of the wrong hill. I'm grateful to the people I trust for showing me the taller ridge. I can't wait for the next person to tell me I'm on a foothill.
Key Takeaways
- Size an account plan to the executive's agenda, because even a correct $22M COGS saving can be a rounding error to the people who sign.
- AI agents make outstanding researchers and lousy judges of framing, so break the Frame Trap with feedback from people who haven't seen the last ten versions.
- A C-level plan opens by playing back the customer's agenda and pairs margin with growth, and it earns a champion before asking for a bakeoff.
- Label every claim Real, Illustrative, Hypothesis or Projected, because C-level readers judge the whole page by its weakest number.
FAQ
What is the Frame Trap?
The Frame Trap is the pattern in which AI keeps improving work inside a frame nobody has questioned, so every version looks more convincing while the frame stays wrong. The way out is feedback from someone outside the work.
How should you use AI to build an enterprise account strategy?
Use agents for research, a living account record and stakeholder role-play, and label every claim. Start from the customer's own initiatives, then let fresh eyes tear the plan apart before the next round of polish.
Why label claims as Real, Illustrative, Hypothesis or Projected?
Executives judge a plan by its weakest number, so labels keep your estimates from posing as sourced facts.
About the Author
Martin Goetzinger is an Enterprise Account Executive who has spent his career in enterprise software sales, helping large organizations such as Apple, Microsoft, and Verizon connect data, insight, and action. His work focuses on transforming how businesses measure success and create customer value through technology.
Outside the enterprise world, he writes about the five forces he believes are reshaping everything: AI, blockchain, energy, personalized health, and robotics. Not from a purely technical lens, but from a human one as to how these technologies will redefine work, wealth, and well-being.
He is based in the U.S. and publishes at www.MartinGoetzinger.com.
Disclaimer
The views expressed in this article are the personal opinions of the author and are provided for informational and educational purposes only. Nothing in this article constitutes investment advice, financial advice, legal advice, or any other form of professional advice. Do not make investment or financial decisions based on the content of this article. Always consult a qualified professional before making decisions that affect your finances, business, or livelihood.
