Known-Strategy AI-ification Rule
Apply AI to a proven strategy before betting on an untested one
- Difficulty
- Easy
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
The rule starts with a marketing strategy that already has evidence behind it and uses AI to make that strategy faster, cheaper, more personalized, or more effective. Because the underlying activity is known to work, uncertainty is concentrated in the AI enhancement rather than in both the strategy and the technology. Teams define the existing performance baseline, build a lightweight AI-enabled version, release it quickly, and compare the result. Lower development and production costs make failure less damaging and iteration faster. A successful first project demonstrates tangible value to leadership and frontline teams, creating momentum for broader adoption. Once the organization has learned how to deploy and evaluate AI safely, it can take on more novel opportunities with less institutional resistance.
Origin
Kipp Bodnar and Kieran Flanagan developed the rule while discussing how marketers can reduce the outcome variance of their first AI initiatives. Extracted from Marketing Against The Grain.
Core principles
- 01Early wins create organizational confidence.
- 02Existing evidence narrows the range of possible outcomes.
- 03AI can reduce the cost and time required to reproduce proven work.
- 04Lower implementation costs allow faster feedback and iteration.
- 05Novelty should not be confused with business impact.
How to run it
- 1
Select proven work
Choose an existing strategy with historical evidence of business impact and a clear baseline.
Pro tip Look for repetitive work constrained by cost, speed, or personalization.
Watch out Do not choose a tactic merely because the team is familiar with it; it must actually work.
- 2
Define the AI advantage
Specify whether AI should improve speed, cost, scale, relevance, or quality.
Pro tip Choose one primary improvement so the experiment remains measurable.
Watch out A vague goal such as “use more AI” cannot establish success.
- 3
Build the minimum version
Use lightweight tools and limited resources to create the smallest useful AI-enabled implementation.
Pro tip Reuse existing workflows and distribution wherever possible.
Watch out Do not rebuild the entire system before testing the enhancement.
- 4
Ship and learn
Release the implementation quickly, collect feedback, and compare it with the established baseline.
Pro tip Track both business performance and the time or cost required to produce it.
Watch out Do not delay launch in pursuit of a perfect first version.
- 5
Expand from evidence
Scale the improvement if it works, or apply the lessons to another proven strategy if it does not.
Pro tip Publicize credible wins internally to increase adoption.
Watch out One failed enhancement does not establish that AI has no value.
In the wild
A company already knows that its first-conversion email sequence produces customers. Instead of inventing a new channel, it uses an LLM to adapt the proven sequence to each recipient’s company and behavior, tests it against generic copy, and ships the experiment without changing the broader funnel.
→ The team obtains a clean comparison and an early AI result without risking the entire acquisition strategy.
Common mistakes
Starting with the flashiest idea
A novel AI concept can fail because of the strategy, the implementation, or the market, making the result difficult to interpret.
Skipping the baseline
Without historical performance, the team cannot tell whether AI made the known strategy better.
Overbuilding before feedback
Heavy investment removes the cost and speed advantages that make the rule useful.
Is it for you?
Best for
It is best for teams seeking an early, measurable AI win that can build confidence and momentum.
Not ideal for
It is not ideal when established strategies are fundamentally broken or the opportunity depends on a genuinely new AI capability.
From the transcript
“you wanna take a known strategy and AI-ify it, because it's gonna give you a more predictable outcome.”
“It's taking something you already know that works and trying to do it better with AI, versus doing some net new thing with AI.”
“the thing that then matters is the time to getting it in the public and getting the feedback loop and iteration on it”
From the episode
Use This A.I. Marketing Strategy To Grow Your Business In 2023 (#114)