How to Be a CX Ally Series: Introduction with Salesforce Examples

This series is dedicated to my colleague, friend, coach, and collaborator...the late Parthy “PJ” Walker.?

Are you an active ally for diverse prospects and customers in your target market??

Most marketing execs and teams are active allies in their companies’ DE&I programs for employees. But for marketers, our scope and stake in diversity and inclusion is much broader. Marketers own diversity and inclusion of the market and customer as well, so we must become inclusive Customer Experience Allies.

What does being a CX Ally mean? A CX Ally has these abilities:

• Identifies the points of bias risk in customer engagement processes, CRM systems, data, scoring, qualification, analytics, AI/ML, and reporting. May extend to products, offers, pricing, service, and support.
• Calculates the revenue impact—pipeline, profits, NLV—of resolving that embedded bias.?
• Masters methods to reduce systemic bias in people, processes, data, and software.

Those are the topics I will explore with you in this series on “How to Be a CX Ally.”

Common Areas of Systemic, Unconscious Bias in CX and CRM

In this first post, I’m going to explore several points of bias risk in customer experience…areas vulnerable to exclusionary practices that can erode company success in a marketplace. For the last few years, I have been analyzing the cost of bias in CX and building solutions on the Salesforce Platform to overcome unconscious, systemic bias in systems and software…so that companies can reach their full marketplace potential, generate more revenue, and retain more customers. I have discovered that Go-to-Market functions are hampered by the same decades of embedded of bias in people, processes, data, systems and software that our Human Resources colleagues struggle against.?

A simplified list of bias risk areas in go-to-market/customer journeys and ways CX Allies can track and resolve--we'll expand on each in future posts:

Awareness—KPI Brand Awareness. Most companies can’t reach nor are fully recognized by the complete, diverse audience they identify as their target market. What can a CX Ally do?

• Assess the database against marketplace diversity. Salesforce is adding diversity fields to make this easier, but you may need to create secondary objects or tables based on your market representation. For example: “Hispanic” is insufficient data to market successfully to the varied Hispanic cultures of Miami and the region.?
• Assess both traditional and social advertising reach by diverse subsegment. This isn’t merely the viewership, nor just the cast. CX Allies will look to content, creators, production teams, media outlet ownership, etc.

Marketing—KPI: Response. If a company is able to reach their full market, the next area to assess is response. Do diverse subsegments respond to your marketing efforts equitably?

• Develop data-driven profile for diverse subsegments.
• Ensure campaigns address the full, diverse marketplace. In Marketing Cloud, it’s a very doable step (small Java script) to create an automatic check on Data Extensions that compares the contact list against the diversity profile of the target audience.
• Adapt to the needs and preferences of the diverse segment/profiles. This may not just be language, format, content, channel, device, etc. In Healthcare, a 35-year-old Caucasian male and a 57-year-old east Asian female may both have Type 2 Diabetes, but a successful patient path for each will be very different.??Salesforce Customer Data Platform is absolutely essential to successfully engaging and adapting to diverse segments at scale and across the ensure customer journey.

Prospecting and Sales—KPI: Qualification. There are several KPIs to watch here but the highest risk area in my experience is fallout during qualification. If the full market is brand aware, responding to your offers…but don’t even enter the pipeline, that’s the red flag. AI is notoriously suspect here, but even the traditional lead scoring models have issues.?

• Monitor lead and opportunity scoring by diverse segment. Scoring models, algorithms and AI/ML are often built on historically biased data. The first fix is to ensure a cleansed data set is used to train tools such as Salesforce Einstein.?
• Post-qualification, you may have to train against unconscious bias in telemarketing and sales staff.?

Support and Service—KPI: Satisfaction. I am assuming marketing has the purview over the complete customer experience, which includes at least monitoring this stage of CX. This is why it’s so important for Marketing to define and manage the market/prospect/customer diversity information in a platform like Salesforce CDP and Customer 360. It allows a single source of data to measure DE&I along the entire customer journey in Salesforce Clouds as well as other platforms.?

• Track first call resolution and other metrics by diverse segment in Salesforce Service Cloud.
• Monitor metrics such as service call cancellations and work order add-ons by diverse segment in Salesforce Field Service Lightning.
• You may have to train against unconscious bias in support and field service staff.?

Advocacy and Loyalty—KPI: Retention and Recommendation. Granted much of customer satisfaction is built through the experience and journey…and managed by other functions…but Marketing usually has the monitoring and reporting role. Once you have the data organized by diverse segment, you should be able to provide this reporting:

• NLV by diverse segment.?
• NPS scores by diverse segment.

What Would a CX Ally Do?

Let’s explore marketing and sales risk areas in an industry example, Telecommunications:?

Internet Service Providers are given a near monopoly for providing data services to a geography. But they are regulated and must demonstrate good-faith coverage for all residents of that community. This is a good illustrative example because we have somewhat better data on diverse segments—age, race, ethnicity, culture, gender, different abilities—for geographic areas. Marginalized and underserved groups are very costly and many claim unprofitable for ISVs to attract and retain.?

Let’s start with the data. ISVs seldom have a database or reach for full market coverage into underserved segments. Try buying a current, cleansed list for ethnic groups, recent immigrants, non-English speaking, disabled, unbanked, lower wage earners. List brokers and their lists are built on decades of biased data collection favoring certain segments. So, the ISVs have little data on these more diverse segments (other than street addresses) in their database, don’t have customer insight, and it’s going to be very costly to get that info using traditional list-building activities. The only options are costly direct mail or general advertising.?

Now let’s look at the process. ISVs normally use credit scoring to qualify for an account. What do you think the credit score would be for unbanked, historically marginalized ethnic groups, or recent immigrants? Credit scoring bias in well documented (one of many studies…?https://hai.stanford.edu/news/how-flawed-data-aggravates-inequality-credit) and does not provide reliable guidance on whether someone is a good payment risk or not. So even if marketers can reach these underserved segments, current qualifying process and criteria will knock them out of the pipeline after expending prospecting and sales resources.?

The CX Ally at Work in Salesforce?

Market reach and the qualification process are reasons why the cost of acquisition for these customers is so much higher. Costs quoted are often >$50 per resident/prospect. How would a CX Ally solve for this? A CX Ally won’t be limited by List Broker limitations. A CX Ally conducts actual market analysis and will quickly discover free (yes, free) government data sources with open APIs are a terrific source of data for these underserved groups. The cost is merely the API connection…and, of course, ongoing cleansing and maintenance of data.?

And the qualification process? A CX Ally is focused on real outcomes across a diverse pipeline. Once we have the reach and response from the market, our analysis would show a drop in diverse leads and prospects passing through qualification. Our first job is to evaluate the scoring model, algorithm or AI against the diverse segments.?

• Which segments are falling out of the scoring model??We may find that differently abled people fall out. A common flaw in lead scoring models is higher scores for content or actions that rely on sight and hearing (videos, webinars, podcasts, demos, etc.). Which means a contact with vision or hearing disabilities would never score high.?
• Does the credit score actually impact NLV??Finding that the credit score doesn’t predict payment reliability, we’d define other predictive indicators. NLP has shown to be very valuable in predicting repayment in small loans and can be an easy set-up in Salesforce Einstein. More and more companies are using alternative credit predictors but be conscious and wary that you don’t violate regulations or privacy (https://www.gao.gov/blog/credit-scoring-alternatives-those-without-credit).
• Is the AI model based on biased data??It’s no secret that AI can amplify bias, but a CX Ally needs to know how to assess if bias is at play and how to fix it. We are all anxious to put Salesforce Einstein AI to work, but we need to be savvy enough about AI and ML to understand that Einstein, for instance, needs ~1000 data inputs to successfully train. What training data was used, was it biased, and does it reflect our actual market potential vs historicals? One fix is to review and scrub the training data. An advanced fix will be developing Aspirational Data Training Sets, which I will describe in a future post.?

While these seem like small changes and are actually low cost, an ISV did make similar changes and reduced the cost of acquisition for marginalized segments by 58%, substantially increasing margins. The great joy of being a CX Ally is that by ensuring we reach, engage and serve the full and diverse market means we also improve our company’s commercials.?

Topics Coming Up:?

• Building a Revenue Calculator for Customer Inclusion Investment
• Creating Customer Journeys for Diverse Subsegments
• Training Bias out of AI Using Aspirational Datasets