Enabling AI-powered sales outreach
Similarweb
SAM (Sales Ai Module) simplifies outreach by using AI to transform real-time traffic data into personalized, ready-to-send sales messages, eliminating manual research.
Timeline
March - December, 2024
responsibilites
Ideation, research, user flows, end-to-end design
TEam
PM, developers, data team, brand designer, marketing manager, general manager
The problem
The challenge
The SAM (Sales AI Module) output was initially text-heavy, making it hard for users to identify what mattered. My role was to distill the information into a clear, engaging interface that highlights what’s important—making it easy for users to adopt within their existing workflow.
🔍
Accelerate insight discovery
✉️
Streamline personalized messaging
📈
Boost reply rates
💰
Improve retention of paying accounts
🪑
Increase average seats per subscription
Product goal
research
Before I joined the project, my fellow product manager had already conducted interviews with SDRs and ran initial research on direct competitors.
Once I joined we conducted an also research on non direct competitors, and internal guerrilla research.
User research (Guerrilla)
We spoke with 2 internal users and 9 stakeholders, the CEO among them.
From what we've heard we understood that Sales reps were spending too much time finding relevant insights and combining them.
They needed better data, surfaced fast.
This shaped our direction to start with lean and focused "talk point" text, and to explore competitors through market research.
“
When searching to add insights to personalized emails it can take 5-10 minutes per email.
Senior Business Development Representative
“
For them [recipients] to bother reading cold email it has to be short and clear, the image helps
Team Manager, Sales Development
Key takeaways from user interviews
Competitor research
Beyond direct competitors like Apollo and Lavender, we also looked to leading AI products—like Grammarly.
From the direct competitors, we understood the industry standards and how can we differentiate our product.
From the non direct competitors it was important for us to see how the big players in the AI industry designed their AI products.
This research informed key decisions around clarity, scannability, and how "Talk points" should appear.
Direct vs. Non direct competitors
Analysis
Designing for how sales reps actually work
The insights were clear:
Sales reps didn’t need writing help — they needed relevance and ready-to-send copy
The design had to be structured, scannable, context-aware, and require minimal cognitive effort, based on how reps work.
After wireframing, we focused on several possible UX directions. We needed to test them and choose the one that delivered the most clarity and speed.
Internal user interviews
We explored several UX directions: a wizard, tabs by insight topic, and collapse/expand flows. In the first iteration we choose the collapse expand.
After internal conversations we developed the new concept. Showing by default 3 talk points, with a simple text about the insights it's based on.
This concept was the most optimal thanks to its clear UX, minimal CTAs and distractions, and the strong value it delivered to users.
Exploration of concepts
Solution
Launching a lean, insight-first MVP
Based on our research the first version of SAM (Sales AI Module) focused on delivering fast, usable insights — without overwhelming users.
After a successful beta launch with a slack channel for immediate feedback, We launched a minimal but powerful core: talk points, regeneration with a clean interface.
From the beginning, we saw users actively engaging with the design and using all available options. For example, the regenerate button accounted for 40% of all user clicks.
After the MVP we had several releases, improving and adding new capabilities step-by-step.
first release
Graphs for visual impact
We added compact graphs to each talk point so sales reps could quickly understand trends — all designed to fit a tight layout.
This release boosted usage by letting users copy the text and graph together, making it easy to paste directly into LinkedIn.
Second release
Preferences & competitor control
Users could now select competitors and fine-tune AI preferences. We designed the experience to feel simple and intuitive.
Preferences became the second most-used feature, while competitor selection ranked fourth in usage.
Third release
Context-aware onboarding
We rolled out onboarding: users approve their value prop (if signed with corporate email), metrics, and traffic location upfront. SAM now felt more personal right away.
This led to a 32% onboarding completion rate.
releases 4-6
Smarter, more customizable output
We added support for languages and message lengths, and expanded SAM (Sales AI Module) to all plans with clear package-based limits (B2C).
Outcome
User growth fueled by value
We launched the MVP in May 2024 (After beta release in April), and then six more releases followed. Each one made SAM (Sales Ai Module) more useful, more flexible — and more loved by sales reps.
By December, we saw an 853% increase in users.
Feedback like: “That’s actually amaaazing”, “This saves me 10 minutes per lead”
Users are viewing more pages per session than average, showing strong interest.
REFLECTION
Context as a design principle
This was one of the most rewarding and challenging projects I’ve worked on. I led the design from concept to execution — turning complex AI logic into something clear, scalable, and genuinely useful.
If I could do one thing differently, I’d take a more holistic view early on — thinking beyond the product to the user’s full workday: their tools, habits, and mindset. That context could have led to smarter entry points and smoother interactions.
