SaaS Project: Building ML-Powered Email Prospecting Platform for B2B
My Role
I was the product manager owning end-to-end ideation for Autobound, a machine learning powered email sales enablement platform for outbound sales reps.
The situation
Autobound was built on the premise that outbound sales reps can be more productive with selling vs. spending 36% of their weekly schedule prospecting.
My core responsibilities included conducting research on user personas, industry, and competitors and writing PRDs to prioritize the most compelling user needs into becoming product features for our MVP. I came up with 3 value propositions as focal points for building features that would get outbound sales reps to trust Autobound into creating the emails for them:
Help users continue to learn and make bigger impact
AI powered prospecting that accelerates high quality lead generation and conversions
Always learning platform that keeps getting better with every use
User Problems
There were 3 user problems I was solving to maximize the success of Autobound:
How to get users on-boarded with the Campaign Review Flow
Understanding progress and next steps from the Dashboard
How to make the platform more fun and approachable
Understanding the Campaign Review Flow
The main way that our users would interact with Autobound is through reviewing the email campaigns and sequences created by the platform.
Click on image to enlarge. Campaign review flow.
Each campaign comprises of a series of emails, an email sequence - scheduled time and order of emails sent to a contact based on the contact’s activity with each email (clicked linked, no response, etc.), and a highly segmented list of contacts.
Users were required to review each campaign and to approve, edit/approve, or reject each campaign.
Successful on-boarding of the campaign review flow required trust points to be surfaced in order to aid the user in understanding the reasoning and impact for each email to guide users to either edit, approve, or reject campaigns.
Click on image to enlarge. Campaign Review Flow
Trust text in bold located in the blue boxes give users context for each email to increase acceptance of Autobound’s emails.
Click on image to enlarge. Email editor
Users can provide any edits to the campaign. The user feedback generated would feed into training the machine learning algorithm.
Click on image to enlarge. Approval window
Users who approve the email with/without editing will need to click on the CTA in this pop up to confirm. Once the user clicks “Approve”, the campaign will be active.
Click on image to enlarge. Rejection window
Whenever user rejects a campaign, there will be an optional field with suggestions as pretext for users to provide feedback. The feedback would help train the ML algorithm to improve on future campaigns.
Providing Actionable Insights in the Dashboard
The dashboard needed to provide an at a glance view of the progress that Autobound email campaigns were making.
Prioritization was given to providing aggregate metrics on a quarterly basis to showcase overall impact of Autobound.
Click on image to enlarge. Default screen is the “Overview” tab in the Performance Dashboard section.
Users could then drill down to specific campaigns to see performance metrics. Opens, Clicks and Replies are the engagement metrics that showcase that Autobound is doing its due diligence to generate engagement. Replies and Leads Generated were the ultimate KPIs.
Click on image to enlarge. “Per Campaign” tab where users can select specific campaigns to view
Click on image to enlarge. Different views available to see within each campaign
Left: Showcasing email replies trend line.
Right: Showcasing lead funnel and the percentage of prospects moving towards the bottom of the funnel.
Users can also compare and contrast between 2 specific campaigns to see which campaign did better to aid them in campaign review decisions.
Click on image to enlarge. Compare feature in Performance Dashboard
Left: Comparing Campaign 2 and Campaign 3 email replies trend lines
Right: Comparing Campaign 2 and Campaign 3 lead funnel conversions
Once the machine learning algorithm becomes more sophisticated, future iterations of the dashboard would provide insights into specific titles that outbound sales reps should get contacts for and upload into the CRM to further improve conversion rates using Autobound.
Gamification as a core driver of success
One aspect missing from our competitors was gamification. With Autobound, this manifests in rewarding users with badges when completing specific actions that would improve retention and increase successful outcomes when using Autobound. The badges would serve as a means to give distinction to achieving certain goals such as meeting quota using Autobound, or reinforcing certain actions such as editing and approving a campaign that led to improved conversions. Users can also compare their performance by seeing where they place in the sales leaderboards.
Click on image to enlarge. “Badges” tab in the Performance section
Users can see the badges they have accumulated.
Click on image to enlarge. Modal View
Clicking on the badges will open up a modal view of the badge that gives a description of what the badge signifies.
Click on image to enlarge. Leaderboard
Launching & Results
We launched the MVP to a select group of outbound sales rep beta users to review and approve campaigns and track the performance of campaigns. It also allowed for admins to create campaigns manually and push them to users.
There were 3 active, paying enterprise clients participating in the closed beta
They signed to pay a $ amount for each successful lead
After running the MVP for 1.5 weeks we found that:
The most active user has 8 campaigns; least active user has 3 campaigns
15 total active campaigns created, sent out to ~1.5K contacts
average cumulative performance:
20% open rate
2% click rate
1% reply rate
campaign approval rate well north of 95%, mostly using quick approval (rather than step by step)
very little editing of campaigns; usually minor details related to links
From the beta testing with our MVP, we concluded that:
Yes, outbound sales reps will actually allow someone else to create their sales campaigns once they trust them
Yes, using sales best practices and playbooks rather than business specific knowledge can create equal to or better campaign performance
For next steps, I put together a comprehensive 12 month product strategy detailing how Autobound can create compelling features to entrench itself with outbound sales reps and then grow its market share through building features to appeal to other sales related personas. The ultimate goal for Autobound is to become sales.
You can access the interactive demo here.