Improving User Experience

with AI

Product

A sales engagement platform that helps sales teams with cold outreach.

Client

Mailshake

Type

Product Design

Year

2023

Project Overview

The problem

Numerous unresolved user pain points can now be easily addressed. Solving these will position our product more competitively in the market.

The goal

Leverage new AI capabilities to solve user pain points at a low cost, thereby improving the end-to-end user experience.

Responsibilities

I initiated and developed this project independently in my free time, without direction from leadership.

Section 2

Understanding the User

2.1 Interviews

Foundational interviews

Qualitative interviews with sales professionals from mid-sized sales teams to understand their needs, pain points, processes, and experiences with sales engagement software.

Interviewed 24 participants in roles ranging from managers to individual contributors. Sales team size ranged from 3 to 100 sales roles, primarily from SaaS companies.

Continuous product discovery

Our product team interviewed a new user each week to gather ongoing customer feedback and insights.

2.1 Personas

2.3 User Journey Maps

Section 3

Concepts

3.1 Sequence Follow Up Suggestions

Why?

Improve Retention

Improve retention by adding morevalue in the product.

Improved Reply Rates

A Yesware study showed that a series of follow-upemails can significantly increase reply rates. Itshowed that sending one follow-up email couldimprove reply rates from 22% to 33%.

Close More Deals

Velocify's research showed that potential customerswho were pursued with a high degree of persistence— over a longer period of time were more likely toclose.

What we've heard

“More guidance on how to strategically use these tools”

- Sr Manager

“Managing nurture campaigns”

- SDR

“So many sequences, it’s hard to keep up”

- Sales Engineer

Priming

SDR persona

You are a Sales Development Representative with over 20 years of experience with an expertise in cold multi-channel outreach, lead generation, lead nurturing, outbound sales, copywriting, and email deliverability best practices. You are proficient in prospecting through email, LinkedIn, and phone.

With email signature

I’m using software that automatically inputs my email signature (including my name). Therefore when I ask you to write or rewrite an email, do not generate a space for my name at the end of the email.

Text replacements & fallbacks

I’m using software that leverages text replacements. Here’s the format used for text replacements: Begins with “{{“, has the “keyword” in the middle, and ends with “}}”. Here’s an {{example}} text replacement.

This text replacement software also has fallback text. Meaning if the keyword isn’t found in the datasource, the software will use the fallback text instead of the text replacement. Fallback text is placed within the same text replacement format and indicated with “:” after the keyword. Here’s another {{example:fallback}} of a text replacement with fallback text.

From now on, feel free to reuse any text replacements/fallbacks that I mention when I ask you to rewrite content, but do not create any new ones. Do you understand?

Prompting

Follow up suggestion

This is the first email in the sequence: Subject: "Inquiry" Body: “hey {{first}}, looks like you have a compact sales team. Are you considering automating your outbound process at this stage?”. Please suggest a follow up message using the same voice and keeping it roughly the same character length.

Output

Chat reply

Subject:

Body:

“Streamlining Sales?”

"Hi {{first}}, just circling back on my previous message. Wondering if you've thought about the potential benefits of automation for your dedicated sales team?"

Sketching it

3.2 Email Personalization from LinkedIn Profile

Why?

Increase Revenue

Potential to increase revenue either as a new product or added to high revenue plans.

Existing Competition

Competitors include Humantic, Crystal, and The Predictive Index.

Higher Open Rates

Research has consistently shown that personalized emails have higher open rates than generic ones. For example, a study by Experian found that personalized emails can deliver six times higher transaction rates.

Improved Click-Through Rates

According to a study by the Aberdeen Group, personalized emails improve click-through rates by an average of 14%, and conversions by 10%...additionally, personalized emails can increase sales leads by 20% and drive a 18 times higher revenue than non-personalized emails.

What we've heard

"“I do a lot of personalization”

- BD & AE

"I do a lot of personalization research”

- Sr SDR

“How do I personalize at scale?”

- Sales Engineer

“Personalization is a pain point”

- SDR

Limitations

When providing only the LinkedIn profile url, the personalization results were limited to the about profile information.In order to get the personalization to reference recent posts made by the account, I needed to provide the posts manually.

Priming

SDR persona

You are a Sales Development Representative with over 20 years of experience with an expertise in cold multi-channel outreach, lead generation, lead nurturing, outbound sales, copywriting, and email deliverability best practices. You are proficient in prospecting through email, LinkedIn, and phone.

With email signature

I’m using software that automatically inputs my email signature (including my name). Therefore when I ask you to write or rewrite an email, do not generate a space for my name at the end of the email.

Text replacements & fallbacks

I’m using software that leverages text replacements. Here’s the format used for text replacements: Begins with “{{“, has the “keyword” in the middle, and ends with “}}”. Here’s an {{example}} text replacement.

This text replacement software also has fallback text. Meaning if the keyword isn’t found in the datasource, the software will use the fallback text instead of the text replacement. Fallback text is placed within the same text replacement format and indicated with “:” after the keyword. Here’s another {{example:fallback}} of a text replacement with fallback text.

From now on, feel free to reuse any text replacements/fallbacks that I mention when I ask you to rewrite content, but do not create any new ones. Do you understand?

Prompting

Personalization

Please rewrite the following email, adding personalization from their LinkedIn page. For personalization, look for interests in the recent LinkedIn posts made by the account.  When rewriting, retain the voice and a similar character count length. Don't add text replacements that I haven't provided you.

Email to rewrite: “hey {{first}}, looks like you have a compact sales team. Are you considering automating your outbound process at this stage?”  

Recent LinkedIn posts: “Post1” “Post2” “Post3” “Post4” “Post 5”

Last 5 posts provided

Output

Chat reply

Subject:

Body:

“Embracing Automation in Outbound Sales, {{first}}”

"Hi {{first}},

I couldn't help but notice the vibrant energy and commitment to excellence your team displayed during your recent events in Denver and Tenerife, not to mention your continued pursuit of AI integration within your CX. It's clear that your team isn't just compact, but impactful.

Given your forward-thinking approach to AI and customer experiences, have you thought about the role automation could play in your outbound process? I believe it could complement your team's innovative spirit and ambitions quite well.

Looking forward to hearing your thoughts.

Best regards,"

Sketching it

3.3 Reporting Synthesis for Sales Coaching

Why?

Increase Revenue

Potential to increase revenue eitheras a new product or added to high revenue plans.

Existing Competition

Competitors Gong, andSecondNature.AI, and sort of Avoma.

What we've heard

“Amount of activity for a booked meeting”

- Sr Manager

“Babysitting SDRs who aren't working”

- SDR Manager

“Knowing how many leads a SDR needs to be sent in order to reach their goal for the year”

- SDR Manager

“AE rejecting SDR leads? Or SDR sending trash leads?”

- SDR Manager

“Amount of meetings for a qualified lead”

- SDR Manager

“Why was a sale lost?”

- SDR Manager

Testing it

Limitations

- Only used the last 14 days with previous period.
  - Individual Coaching (Won leads, New leads, Lost leads, Contacted prospects).
  - Team Coaching (Won leads, New leads, Lost leads, Contacted prospects).
  - Team Coaching (Sent, Opens, Clicks, Replies, Bounces, Unsubscribes).
- Unknown usefulness of the feedback it provides.
- There’s still more to explore!

Privacy

- Looked at the SME’s outbound sales team for data.
- Removed the names of the sales people in order to protect PII.

3.3.1 (Part 1) Reporting Synthesis - Individual Coaching

Priming

Manager persona

You manage a team of sales development representatives. You want to help coach your team as a whole to improve productivity (as well as individual team members). Please consider the following data and provide coaching suggestions for the team as a whole as well as individual team members:

Prompting

Data provided: Won Leads, New Leads, Lost Leads, Contacted Prospects

Output

Chat reply

Salesperson 1

This salesperson seems to have the highest number of leads lost. To improve, they might need to work on their follow-up and closing skills. Additionally, this person seems to have a high number of contacted prospects but relatively lower conversion, implying a need for more effective engagement strategies.

Salesperson 2

They have a high number of contacted prospects and generate a good number of leads, but their won leads are relatively low. This suggests they might be struggling to close deals and would benefit from training in closing techniques and negotiation.

Salesperson 5

They show a balanced performance with respect to won leads and lost leads. However, they could improve in terms of generating new leads. Additional training in prospecting and lead generation strategies could be beneficial.

Salesperson 8

Despite having relatively fewer contacted prospects and generated leads than some others, they have a good number of won leads, suggesting they are effectively converting their leads. They could share their best practices with the team.

Salesperson 9

This salesperson has lower numbers in terms of contacted prospects and generated leads, but they have a relatively high number of won leads, which is a positive sign. Coaching could focus on increasing the volume of their outreach and prospecting efforts.

Salesperson 3, 4, 6, 7, 10, 11, 12, 13

These salespeople seem to have no activity in the period. If they are new to the team, they should start with basic training in prospecting and sales skills. If they are not new, it is crucial to understand.

3.3.2 (Part 2) Reporting Synthesis - Team Coaching

Output

Chat reply

Lead Conversion Rate

It seems like the total number of new leads generated is quite high compared to the number of leads that are won. Therefore, the team as a whole can work on improving their conversion rates. Coaching can focus on improving the team's sales skills such as effective communication, rapport building, and negotiation techniques.

Consistency in Prospecting

It seems like there are several days with no new leads generated (especially on July 1 and 2). This could indicate that prospecting efforts are inconsistent. Regular, consistent prospecting is crucial to maintain a healthy pipeline.

Handling of Lost Leads

The team also has a significant number of lost leads. This can be an area of improvement. The team could benefit from a training session to handle objections, follow-up strategies, and how to re-engage lost leads.

Prompting

Additional data (1 of 2)

What can insights can you tell me about my teams performance based on theses engagement numbers?

Data Provided: Sent, Opens, Clicks, Replies, Bounces, Unsubscribes (2 of 2)

Output

Chat reply

High Volume of Emails Sent

Your team sends a large volume of emails, particularly on June 23, 26, 27,28, 30, and July 3, 5, and 6. This indicates a high level of outbound activity, which is a positive sign for lead generation and engagement efforts.

Email Open Rate

The open rate seems reasonably good, especially given the high volume of emails sent. However, the unique open count drops significantly on weekends(July 1 and 2) and seems to be lower relative to the emails sent on June 24,suggesting potential areas to optimize timing or content for better engagement.

Click-Through Rate

The unique click count is relatively low, especially compared to the number of unique opens. This suggests that while people are opening your emails, they're not necessarily engaging with the content or following through on calls to action. Consider testing different types of content, messages, or CTAs to improve engagement.

Reply Rate

The unique reply count is also relatively low compared to the number of unique opens. Your team might need to make your emails more personalized or engaging to motivate recipients to respond.

Bounce Rate

The number of bounces is a bit high on some days (June 23, 26, 27, 28, and July 4, 5, and 6). This could indicate that some emails are being sent to invalid addresses. It's important to clean the email list regularly to keep bounce rates low and improve deliverability.

Unsubscribes

The number of unsubscribes is relatively low, which is a positive sign. However, July 5 had the highest number of unsubscribes, which might indicate the content or frequency of emails on that day was not well-received.

Recommendations

- Improve the content of your emails to encourage more recipients to click through and reply. Personalize the emails as much as possible and use compelling CTAs.
- Test different send times to find when your recipients are most likely to open and engage with your emails.

...

- Regularly clean your email list to reduce bounces and improve deliverability.
- Monitor unsubscribe rates closely and adjust your content or frequency accordingly to maintain your subscriber base.

...

- Consider segmenting your email list based on behavior or other factors to improve engagement.
- Use A/B testing to identify the most effective strategies for improving email engagement.

3.4 Email Sentiment

Why?

Table Stakes

Table stakes for a lot of our competitors.

Existing Competition

Competitors include Salesloft, Outreach, Instantly, ActiveCampaign and many more.

Time Management

Prioritize time better by focusing on positive first.

Manage Expectations

An indicator to manage SDR expectations.

What we've heard

“Need to see what's going on my my prospects each day”

- Sales Lead

“Rearranging priorities”

- Sr SDR

“Managing to-do list and calendar”

- Sr SDR

Limitations

We only provided the most recent prospect email reply (email body) for sentiment analysis, not the entire conversation.

Priming

SDR persona

You are a Sales Development Representative with over 20years of experience with an expertise in cold multi-channel outreach, lead generation, lead nurturing, outbound sales, copywriting, and email deliverability best practices. You are proficient in prospecting through email, LinkedIn, and phone.

Prompting

Determine sentiment

You classify sentiment in the following buckets: "Strongly negative, Negative, Neutral, Positive, and Strongly Positive".

Please determine the prospect’s sentiment based on their following responses: “Response”

Output

Chat reply summary

Negative Examples

- “Stop”
- “please stop”
- "Zero interest”
- “I am not interested.”

Neutral Examples

- “OPT OUT”
- “Hey! I'm already working with Salesperson on this. Thankyou!”
- "I'm not interested at this time.”
- "I am not available today. I would prefer after July 6th.Thanks,"

Positive Examples

- "No worries. I do it all the time. LOL"
- "Share away brother”
- "We're happy Outreach users for now -- no rip and replace anytime soon. Best of luck out there. -Prospect"
- “Yes please do”

Section 4

Going Forward

4.1 Takeaways

Impact

The concepts I explored were met with enthusiasm from leadership and were added to the product roadmap. As a result of this unsolicited work, I received a raise, which underscored the value and impact of my contributions.

What I learned

Revisiting the user journey with new technology capabilities can add significant value at a low cost.

4.2 Next Steps

Create Designs for Each Concept

Design how each concept will look within the system.

Build Prototypes

Collaborate with the development team to build prototypes for testing.

Test for Desirability

Test the desirability of features with our sales team and a select group of customers, ensuring they understand it is a rough prototype.

Implement Concepts

Implement the concepts that pass desirability testing.