Field Notes

Test More Growth Ideas Every Week With AI Copilots

By Jason Kumpf, Strategy Advisor · October 28, 2026 · 5 min read

Test More Growth Ideas Every Week With AI Copilots

Generative AI lets lean marketing teams draft, launch and learn from far more ideas than a traditional calendar allows. Here is a practical loop, plus 12 experiments you can start this month.

The most valuable number in a growth program is rarely the conversion rate. It is the number of good ideas a team can test before the quarter ends. More tests mean more chances to find the message, offer or channel that breaks out.

Generative AI has changed that math. Tools such as Claude and Gemini now handle much of the drafting, variation and analysis that once filled a marketer's week. That frees people for the part software cannot do: choosing which bold idea deserves a shot.

Marketing is where AI is already working hardest

Marketers moved first. In McKinsey's global survey on the state of AI, fielded in July 2024 with 1,491 participants, marketing and sales was the business function most commonly using generative AI, at an average of 42 percent across industries. Technology companies led at 55 percent, followed by professional services at 49 percent.

The tools keep getting sharper. At Google Marketing Live on May 20, 2026, Google expanded Asset Studio, its Gemini-powered space for building and testing ad creative, with connections to design tools such as Canva and Adobe and the option to upload brand guidelines that steer every new asset. Google also introduced Ask Advisor, an agent built with Gemini that works across Google Ads, Google Analytics, Merchant Center and Google Marketing Platform.

The building blocks for rapid testing now sit inside the software marketers open every morning. The opportunity is to organize work around them.

Inside one AI-native growth team

Anthropic offers a useful case study from its own building. In a July 2025 write-up on how its teams use Claude Code, the company described a growth marketing workflow that reads a CSV file of several hundred ads, identifies the ones ready for a refresh and generates new variations that stay within strict character limits. Two specialized sub-agents handle the work.

The result: several hundred new ads in minutes rather than hours. The same team built a Figma plugin that produces up to 100 ad variations by swapping headlines and descriptions across frames. A copy-and-paste task that once took hours now takes about half a second.

The lesson goes beyond any single tool. A small team mapped the repetitive steps between an idea and a live test, then handed those steps to software. Their creative energy went back to strategy.

The fast-experiment loop, step by step

Teams that get the most from AI run a simple, repeatable loop. Each cycle should take days, not weeks.

  1. Frame the question. Write one sentence that names the audience, the behavior you want and the signal you will measure.
  2. Generate wide. Ask your AI assistant for 20 to 30 angles on the message, then have a person pick the five most distinct.
  3. Build fast. Use AI to draft the copy, image prompts, landing page sections and email versions for each angle.
  4. Launch small. Put each variant in front of a modest, comparable slice of your audience.
  5. Read and recombine. Have the model summarize results and customer comments, then carry the winning elements into the next round.

Human judgment anchors both ends of the loop: choosing the question and deciding what the results mean. AI compresses the middle, which is where most of the hours used to go.

12 experiments to run this month

Here is a starter list. Each one can be drafted in an afternoon with Claude, Gemini or a similar assistant.

  • Headline sprint. Write 25 headlines for your top landing page in five registers (curious, bold, practical, playful and aspirational) and test the top three.
  • Persona remix. Rewrite one core email for three buyer roles, such as a finance lead, an operations lead and a chief executive.
  • Question mining. Cluster the questions prospects ask most often in sales conversations, then turn each cluster into an ad or a FAQ block.
  • Offer framing. Present the same offer as a guide, a checklist and a short workshop invitation.
  • Subject line ladder. Generate 15 subject lines that move from literal to playful, then test the two ends of the ladder.
  • Visual directions. Use Gemini-powered creative tools to produce five distinct visual concepts for one campaign and let early engagement pick the winner.
  • Localization pilot. Adapt a high-performing page for a new region's idioms and examples, with a native speaker doing the final review.
  • Customer language. Summarize public reviews to find the exact words happy customers use, then put those words in your headlines.
  • Video hooks. Script ten different opening lines for a 30-second video and test only the first three seconds.
  • Leadership voice. Turn a recent internal talk or memo into five short social posts.
  • Interactive tool. Describe a simple calculator or quiz to an AI coding assistant and ship a first version on a test page.
  • Welcome-back note. Draft three tones of re-engagement email for quiet subscribers, from warm to witty.

Where to point AI first

Not every task deserves automation on day one. The best starting points share three traits: they repeat often, they follow a clear pattern and a person can review the output in seconds.

Ad copy variations, email subject lines, landing page sections and campaign summaries all fit that description. So does turning a long webinar transcript into a blog outline, a set of social posts and a short recap email. Each of these used to consume hours of skilled time that can now go to strategy and creative direction.

Save the hardest calls for people: positioning, the big creative idea and the decision about which results to trust. A useful rule of thumb is simple. Let AI produce the options, and let your team choose.

Habits that keep speed and quality together

Fast and polished go together when a few habits are in place. These keep the volume high and the brand consistent.

  • Give the model a short brand brief with your voice, preferred phrases and three examples of your best work.
  • Keep a human editor on every asset that reaches customers.
  • Check every factual claim against its original source before launch.
  • Log each test in a shared sheet so wins compound and the next sprint starts smarter.

A shared test log is the quiet hero of the system. After a few months it becomes a playbook of what your audience responds to, written in your own data.

Putting the Loop Into Practice

The opportunity is straightforward: a growth team can now test many more ideas in the same calendar without adding headcount. The advantage goes to teams that learn fastest, and AI makes learning faster for everyone willing to set up the loop.

Start with one channel and one sprint. Pick the page or email that matters most, run a two-week cycle with five AI-built variants and share what you learn with the wider team. Momentum, more than any single tool, is what turns a good team into a growth machine.

Sources

  1. McKinsey & Company, "Gen AI's broad reach," The Week in Charts, April 17, 2025 (drawn from "The state of AI: How organizations are rewiring to capture value," March 12, 2025).
  2. Google, "Google Marketing Live 2026" announcement collection, The Keyword, May 20, 2026.
  3. Anthropic, "How Anthropic teams use Claude Code," Claude blog, July 24, 2025.

More news from Growth Idea Group