Leveraging AI to build websites for small businesses within minutes
Generative AI · product design · Research
INTRODUCTION
In this rapidly evolving digital landscape, businesses need to create and maintain an online presence swiftly and efficiently. Traditional website development can be time-consuming and expensive, posing a significant challenge for small to medium-sized businesses with limited resources.
How can HubSpot help small and medium sized businesses create an online presence faster? This was the key problem we set out to solve.
ROLE
Lead Product designer
WHAT I DID
Complete product design, user research, user-testing and engineering handoff
TEAM
1 PM, 6 Engineers and 1 Designer (me)
DISCOVERY
The initial problem was discovered when noticing low user activation rates with free and Starter tier users of HubSpot’s Content Hub (previously CMS Hub). These users typically represented small and medium sized businesses. Users were creating websites but not publishing them and upon researching with these users, we learned that they:
KEY INSIGHTs
During discovery research we learned that the reason users were creating websites but not publishing them and the reason was because they:
Typically had zero to low experience with building websites
Had limited time and resources to build websites
Were in early stages of building their business and needed an online presence fast
design sprint
We kicked things by running a virtual design sprint with a mix of cross-functional stakeholders from various product groups. The group ideated multiple different ways to solve for this problem and came up with an initial concept.
concept validation
We ran a series of quick concept validation user interviews with participants representing our target persona - early stage small business owners with none/limited knowledge of building websites intending to sell products or services, collect leads or share content through blogs.
learnings
From concept validation we learned that:
There was huge excitement towards using AI to build websites!
Users were amazed by the amount of value being provided for free
AI being a new technology, users needed more confidence on how their information was being used to tailor the generated website.
On the heels of this wildly positive reaction to the initial prototype, I set out to iterate for a more concrete first iteration focusing on the following goals.
Design goals
Increasing value
Confidence using AI
Problem 1
Increasing value
With the first iteration for MVP, I set out to identify the core aspects of the experience that users found strongly valuable and contribute directly towards the success metric - user activation.
process
Ranking desirability and feasibility
Reviewing user feedback of the previous iteration, I prioritized features by feasibility and how desirable they were to users.
challenges
In trying to condense the experience, the challenge was understanding where to make tradeoffs especially while balancing user expectations technical feasibility and value to the customer in context of the core metric of activation and the costs per AI generation call.
Showing previews of the generated website
Previews of the generated website was a feature that users liked a lot but owing to heavy technical effort, this was scoped out for MVP.
Regenerating new versions
AI regeneration was quickly becoming a standard feature with all AI tools, however due to the cost of regeneration per site, this was scoped out of the free version.
Generating business names with AI
Although a high interest feature, this was scoped out since most users had some idea of what name they wanted to use for their business.
Problem 2
Providing confidence with using AI
A key concern that came up during user research was that users needed more confidence with using AI. Being such a new technology at the time, having clarity into how user inputs were directly contributing to the generated output was critical to user adoption.
solution
Moving fast for the MVP, I found that even simple solutions like adding context on the impact of a certain input had strong positive impacts to this challenge. In addition, I also worked with marketing to ensure that the marketing landing page provided enough clarity on what to expect from this free experience.
Final experience
The final MVP experience focused on value and helping users feel more confident providing inputs
The AI Website Generator was launched live in front of tens of thousands of HubSpot customers at the INBOUND conference!
Post launch we saw a huge rise in user activation rates, almost 2X times and it was exciting to see real users creating websites with AI Website generator for their business that focused on:
Addressing climate concerns with cutting edge technology
Using data analytics to drive innovation
And offering consulting services for product development and tech governance
