When was the last time you came across a new Artificial Intelligence tool that promises to take your marketing to the next level? Most likely, it was this morning when you were scrolling through your LinkedIn feed.
If you feel overwhelmed by all the AI content, you’re not alone. Most of these “AI must-do” lists sound unrealistic and confusing. Growth marketers are bombarded with “What’s the new hot AI tool?” and there is less content about what tools to pick and how to implement them.
In our first two articles, we defined growth marketing and showcased how to think like a growth marketer. And guess what? These fundamentals don’t change with artificial intelligence.
In this article, you’ll learn the 2 Artificial Intelligence applications that matter most for growth marketers. No overwhelming lists. No tool comparisons. Just a clear starting point that you can implement today.
Table of Contents
The Two Artificial Intelligence Applications That Matter Most
Artificial Intelligence. tools can do countless things in marketing, from content creation to ICP mapping. But if you’re starting out, most of these tools can be distractive. Based on what we learned from thinking like a growth marketer, we will be focusing on two AI applications to create the most leverage and amplify growth marketing mental models.
Artificial Intelligence for Experiment Velocity
Learning is a crucial aspect of growth marketing. The faster you can test hypotheses and validate learning, the faster you can compound knowledge. And this is when AI comes to amplify testing velocity in three ways:
- Variant generation: Generate 10 email subject line variations in 2 minutes instead of 30;
- Audience segmentation: Identify micro-segments for targeted initiatives;
- Results analysis: Spot patterns across experiments faster.
The result: You can run 3-4 experiments in the time it used to take to run one. Meaning 3-4x more validated learnings in the same period.
Why does it matter for your funnel?
Remember the AAARRR funnel analysis from the previous article? Let’s say you identified your biggest conversion drop-off. You reported a 30% churn after the first paid subscription, meaning a drop-off from activation to retention.
AI-powered experiment velocity means you can test solutions to that bottleneck faster:
- Instead of one follow-up email test per week, you’ll test three different options;
- Instead of waiting 6 weeks for statistical significance, you’ll run parallel tests with smaller segments;
- Instead of guessing what works, you’ll access patterns faster to see what works.
Faster experiments lead to faster improvements, resulting in faster compounding growth.
How to implement Artificial Intelligence for experiment velocity this week?
Here comes our easy-to-implement 4-step framework that you can start today.
Step 1: Pick your testing scope.
Based on the previous article, you might come across your drop-off point. Either way, you need to pick a test focus and start from there. Preferably pick one variable at a time to avoid overwhelm.
Step 2: Set up your Artificila Intelligence testing workflow.
Start by picking your Large Language Model (LLM) of choice, for instance, the freemium version of Claude. Then go through prompts that reflect what you are looking for. An easy way is to reverse engineer the prompts by asking the LLM what prompt is the best to use for a specific output.
For example, you can ask Claude, “What’s the best prompt to use to navigate bottlenecks in my funnel, where I reported a 30% churn rate from activation to retention?”
This way you will be sure to ask the LLM the right prompts for better output.
Step 3: Run your first Artificial Intelligence workflow.
Here comes testing time. Pick the bottleneck you have already designed as a test focus, then go to your LLM of choice (Claude, for instance) and test with prompts until you find the right one.
Once you find the right prompt, ask the LLM to generate variations for your next step (e.g., different headlines, CTAs, or tweaks in the email body). Set these variations up, and start running the tests.
Step 4: Document the learning
As we know, learning is a crucial component for growth marketing. And in this step, we focus on what we are learning from running these AI workflows. Each experiment will give you insights on what to focus on and what to ignore. Allowing your team to accumulate learning.
Step 5: How to measure success
Here comes the finale! Learning while running the show is crucial, yet defining metrics to measure success will enable you to benefit from your learnings.
As for this part, velocity metrics will help define success. For instance, assessing the number of validated experiments per month before AI implementation and after running the AI workflow can inform you and your stakeholders about the amplifying role of artificial intelligence..
Impact metrics can also be a good translation of what artificial intelligence. helped boost. For instance, after running emails with different copies, your activation rate goes from 30% to 35% over 8 weeks.
Artificial Intelligence for Personalization at Scale
As we learned in the previous article, retention-first economics really matters. Meaning keeping customers is more valuable than acquiring new ones. But to keep your current customers, personalization must be a priority to meet each customer where they are and address them with the right messaging at the right time.
But as we all know, personalization, especially at a large scale, is time-intensive. Imagine how exhausting it is delivering customized onboarding, tailored emails, and adapted content to each customer, let alone the risk of human error.
AI makes personalization more accessible by focusing on:
- Dynamic content: Adjusting messaging automatically, based on user behavior, industry, role…
- Behavioral triggers: Personalizing touchpoints based on where users prefer to get in touch with the company
- Segment-specific journeys: Create different activation paths instead, based on the patterns of existing customers.
The result: Each customer gets a relevant experience without you manually adjusting each step and messaging outlet.
Why does it matter for your funnel?
In growth marketing all funnel stages are equally important. Yet, personalization is quite critical for the following two stages:
- Activation: Nowadays, users are more aware of generic messaging and can cut through the phony and salesy communications more than ever. Users are looking for value and empathy. The sooner they can come across the value and personalized communication that address their concerns, the higher chances you have to convert them into paying customers.
- Retention: According to McKinsey, 71% of consumers expect personalization, especially post-pandemic, where everyone started using technology more than anytime. Customer expectations are not the only factor to adopt personalization in this stage, but according to another McKinsey article, companies that personalize CX drive 40% more growth than companies that go for generic messaging with current customers.
How to implement Artificial Intelligence for personalization this week?
Step 1: Identify your personalization priority.
Detect a stage or two at most that creates the most drop-offs, and start the personalization efforts from there. It’s not one size fits all; it might be both activation and retention, as it might be acquisition. Most importantly, avoid personalizing the whole funnel at once.
Step 2: Personalize high-impact touchpoints
Personalizing all assets and campaigns can be a hassle. That’s why you should define the leaking touchpoints in each stage. For instance, if activation rates are not moving, try personalizing onboarding emails with great copy. And if retention is weak, try personalizing nurturing emails with industry-focused content.
Step 3: Set up your Artificial Intelligence personalization workflow.
For this workflow, we need a second app to join the stack. In our B2B scope, emails are more popular; hence, the need for an email platform. So, in addition to using Claude to generate prompts and content ideas, we will be using HubSpot as an email platform. As it integrates email personalization features and allows running A/B tests.
The workflow can take multiple formats, but we suggest the following one:
- Create 3-5 segments (by industry, company size, region, …)
- Use both Claude & HubSpot to generate personalized versions for each segment.
- Generate multiple versions of the personalized emails to run them in A/B tests.
Step 4: Run and measure
It’s showtime! Now that everything is set up, you can run the workflow in your email platform, using both its embedded personalization features and what you generated through your LLM of choice.
In this step, it’s also important to define success metrics to assess if your efforts are paying off. For instance, define the percentage of new users who complete activation milestones and assess how personalization enabled an increase in this metric.
Conclusion
Growth marketing in the Artificial Intelligence. era isn’t about using every single tool out there, but it’s more about using AI strategically to amplify the fundamentals of the discipline. Starting with velocity was a warm-up to test AI capabilities with no to little budget and prior knowledge. Then we tackled personalization, which is a crucial lever of growth, and integrating LLMs in your email marketing platform can be a recipe for success and optimization.
Before you close this article, we would like to let you know that all you need is 2 hours of implementation time for one and 30 minutes weekly to tweak your workflows. If you can commit 30 minutes a week for the next 30 days, you’ll start your first step in this journey. And if not, please bookmark this article for future use.
In our next article, we will cover how to use artificial intelligence. to deliver personalization at scale, optimizing the customer experience with the same workload and headcount. Diving deeper into dynamic content, behavioral triggers, and segment-specific journeys that improve activation and retention.
