The Inside Story of How We Got 105K People Using Our Marketing Chatbot

In the crowded martech landscape, driving widespread adoption of a new tool is no easy feat. Marketers and salespeople are inundated with pitches promising to revolutionize their workflow. Separating the signal from the noise has never been harder.

So when we set out to build a chatbot specifically for marketing and sales professionals, we knew we had to take a radically different approach. Our north star from day one was creating a bot that would become a trusted, indispensable part of our users‘ daily work lives.

Today, I‘m excited to share that over 105,000 marketers and salespeople use our chatbot every month to surface insights, track key metrics, and streamline their workflows. Here‘s the inside story of how we got there.

Putting Users at the Center

Our single most important decision was defining a target audience for the chatbot from the outset. We knew that a generic, jack-of-all-trades assistant would be doomed to fail. To deliver meaningful value, we needed to focus relentlessly on a core set of users and use cases.

After extensive research and interviews, we honed in on marketers and salespeople as our primary persona. These teams are under constant pressure to drive results, but they often get bogged down wrangling data from dozens of siloed tools. The status quo of manual reporting and fragmented insights leaves a huge amount of efficiency on the table.

Our goal was to unify data from all the most important marketing and sales platforms – CRMs, marketing automation, web analytics, SEO tools, etc. – and make it accessible through a single intelligent interface. We envisioned a world where any marketing or sales question could be answered in seconds with a simple natural language query.

Some of the most common pain points we heard in our user research:

  • "I waste hours every week pulling data from different tools for my marketing reports"
  • "I get blindsided by pipeline issues because I don‘t have an easy way to proactively monitor key accounts"
  • "I miss important customer insights because I don‘t have time to dig through campaign data every day"
  • "I struggle to get buy-in for additional budget because I can‘t easily tie marketing activities to revenue"

These challenges resonated deeply with our team. We knew that if we could abstract away the tedious parts of data analysis and empower users to focus on high-impact work, we‘d be onto something big.

Building an Intelligent Marketing and Sales Assistant

Of course, building a best-in-class AI chatbot is easier said than done. To create a truly intelligent assistant, we needed to combine cutting-edge natural language processing with deep integrations into all the key sales and marketing platforms.

On the NLP side, we partnered with the team at OpenAI to adapt their state-of-the-art Transformers-based model for our domain. Through extensive pre-training on marketing and sales data, we were able to create a model that could understand and respond to a huge range of questions out of the box.

But NLP alone isn‘t enough to answer most real-world questions. To deliver on our vision, we needed to build a robust knowledge base that could translate natural language queries into API calls across dozens of different tools and data sources.

We started by integrating with the most popular CRMs and marketing automation platforms like Salesforce, HubSpot, and Marketo. Our chatbot can access full contact and company records, monitor pipeline changes, and track engagement across the buyer‘s journey.

Next, we tackled the world of web and product analytics. Integrations with tools like Google Analytics, Amplitude, and Mixpanel empower our users to easily access site metrics and investigate user behavior. They can ask questions like:

  • "How many trials did we start last week?"
  • "What‘s the average order value for customers acquired through paid search?"
  • "Which features are being used most heavily by our enterprise customers?"

For content marketers, we integrated with leading SEO platforms like Ahrefs and Moz to surface insights on keyword rankings, backlink growth, and competitor strategies. Our chatbot makes it easy to track content performance and identify optimization opportunities on the fly.

Ahrefs Integration Example

The end result is a chatbot that can answer an incredible breadth of questions by tapping into data across the entire marketing and sales stack. And because it‘s powered by cutting-edge AI, it‘s able to understand questions phrased in everyday language and provide thoughtful, nuanced responses.

Some examples of questions our chatbot can now easily handle:

  • "What were our top 3 lead sources last quarter?"
  • "Which types of content are driving the most organic traffic to our blog?"
  • "Alert me if any of our top 50 opportunities haven‘t had activity in the last 2 weeks"
  • "Which industries have the best win rates for deals over $50K?"
  • "Analyze the impact of our latest product launch on key activation metrics"

Empowering marketers and salespeople to get instant answers to their burning questions has been game-changing. Instead of spending hours pulling data and cobbling together reports, they can now have intelligent conversations with their data to quickly extract insights and take action.

Designing a Delightful UX

We knew that raw technical capabilities were just one part of the equation. To drive true adoption, we needed to craft an experience that users would genuinely enjoy.

A key part of that was investing heavily in the chatbot‘s personality and tone. We wanted every interaction to feel like talking to a trusted colleague – smart, helpful, and succinct. Some key principles we followed:

  • Err on the side of showing rather than telling. Wherever possible, the chatbot should respond with a chart, graph, or concrete data point instead of a wall of text.

  • Adapt the vocabulary and communication style to the individual user. The chatbot should pick up on terms like "MQL" vs "SQL" and mirror the user‘s preferred lingo.

  • Inject moments of warmth and humanity. Something as simple as "You‘re welcome! Let me know if you have any other questions." can make the experience feel much more personal.

  • Maintain context across conversations. If a user asks about "leads from Facebook last month", the chatbot should be able to handle a follow-up query like "How does that compare to Google?" without missing a beat.

Another critical aspect of the UX was setting clear expectations about what the chatbot can and can‘t do. During the onboarding flow, we explicitly call out the key skills and most common use cases. We also make it easy for users to access a cheat sheet of sample queries at any time.

Onboarding Example

By focusing the onboarding on a core set of high-value interactions, we‘re able to get users to that magical "aha" moment faster. Once they experience how much easier the chatbot makes their lives, they‘re much more likely to stick around and explore its full capabilities.

Growth Powered by Word of Mouth

With a product that solved a real pain point and an engaging, intuitive UX, we had a solid foundation for growth. But we also knew that driving adoption for a new type of productivity tool would be an uphill battle.

Our core insight was that the best way to convince people of the chatbot‘s value was to let them experience it for themselves. No marketing copy or product video could match the power of seeing the chatbot unearth a key insight in seconds.

So we leaned heavily into a product-led growth model, with a relentless focus on activation and engagement within the first few days after signup. Some of the key tactics in our playbook:

  • A guided onboarding flow that walked new users through 5 key interactions and encouraged them to connect their core marketing and sales tools. By reducing Time to Value, we were able to dramatically increase activation rates.

  • Targeted in-app messages that suggested high-value queries based on the user‘s role and connected tools. For example, a content marketer who connected Google Analytics might see a prompt to ask about their top traffic sources.

  • A robust library of templates and sample questions to help users discover new ways to leverage the chatbot. We made this library easily accessible within the bot UI and also promoted it heavily via email.

  • Shareable dashboards and reports that made it easy for users to socialise insights and collaborate with their teammates. We included prominent CTAs to invite colleagues and tracked the virality of different insights.

By getting users to that first moment of delight and helping them build a habit around the chatbot, we were able to drive strong retention and organic word of mouth. Over 60% of our signups now come from referrals and invites from existing users.

One of the most powerful drivers of word of mouth has been the "water cooler effect". When someone asks a question in Slack and gets an answer from the chatbot in seconds, it naturally piques the interest of other people in the channel. We‘ve seen entire teams adopt the chatbot bottom-up after one power user gave a quick demo to their colleagues.

As the flywheel started to spin, our growth accelerated rapidly. We went from 100 to 10,000 users in just 6 months. And in the two years since launch, we‘ve grown to more than 105,000 monthly active users across thousands of companies globally.

Growth Metrics

The Future of Work

Looking ahead, we truly believe that AI-powered assistants will fundamentally transform how we work. Today, knowledge workers spend over 20% of their time on repetitive data collection and analysis tasks. By abstracting that work away, we can unlock massive gains in both individual and organizational productivity.

But realizing this potential will require a deep understanding of the end user and a willingness to solve for their specific needs. Generalist chatbots and virtual assistants will only ever be a novelty. The real value is in building domain-specific AI that is uniquely tailored to the workflows and challenges of a particular role or function.

This is the approach we‘ve taken with our chatbot for marketers and salespeople, and we‘ve seen the incredible impact it can have. We‘re now applying this same playbook to build intelligent assistants for other functions like customer success, product, and more.

The rise of remote and distributed work has only accelerated the need for this kind of ambient, always-on support. When you can‘t tap your deskmate to get a quick question answered, having an intelligent chatbot in your corner makes all the difference.

At the same time, we‘re acutely aware of the importance of building trust and setting appropriate expectations. The goal of AI assistants should be to augment and empower human workers, not replace them. We see a future where knowledge workers can offload the tedious parts of their job to AI and focus on higher-order tasks that require creativity, empathy, and judgment.

As we look to the next stage of our growth, we‘re excited to keep pushing the boundaries of what‘s possible with conversational AI. With every new integration and algorithm update, our chatbot gets a little bit smarter and a little more valuable to the marketers and salespeople who rely on it every day.

But our north star remains the same as it was on day one: to build an indispensable assistant that makes our users‘ work lives simpler, more productive, and more fulfilling. If we can continue to deliver on that promise, I‘m confident the next 105,000 users won‘t be far behind.