Sparkcentral - Digital Customer Support SaaS Solution
Sparkcentral's mission is to help brands exceed the expectations of today’s mobile, hyper-connected and increasingly impatient customer. Sparkcentral is a Saas startup helping companies like Delta, Slack, Uber and Netflix offer exceptional customer support over digital channels.
I was responsible for defining the vision and leading the delivery of SC's Reporting product in addition to shipping agent workflow, search, data API's and integration improvements.
Customer service industry
Traditional call centers, a $20 billion dollar industry with a CAGR of 52%, are being digitally disrupted.
Over the decarde, social media, the Internet of Things (IoT) and mobility are all changing the dynamics of how businesses serve their customers.
Enterprise organizations are moving away from traditional on-prem applications to cloud apps, and forced into evolving operations to remain connected to the ever-growing population of millenial consumers across a myriad of social and messaging channels.
At the same time, there's a growing gap between customer mindset and reality. Customers expect to consume a product or service how they want to, when they want to, and for them to be understood in a way that is personal and relevant.
A customer is 4x more likely to be disloyal after a high effort service interaction.
What determines 'high effort' you ask?
- If asked to switch channels to get their issue resolved
- If asked to repeat information about an issue to multiple agents
- Having to contact multiple times to get a timely response
- Having to interact with an agent who cannot resolve their issue/inquiry
The benefits for orgs adopting digital channels as a customer-service channel and leveraging efficiencies of the cloud to meet market needs are vast:
Happier agents (lower attrition rate)
- Brands have seen higher satisfaction in job in agents focused on delivering support on digital channels compared to other service channels.
Increased operational efficiency (lower TCO - Total cost of ownership)
- Cost effective to service customers on social ($1 per interaction) compared to phone ($10/interaction) or live chat ($3/interaction).
- Agents can handle more volume on social (60 inquiries) compare to phone (8) and live chat (15) per handle due to the nature of communication medium.
- Number of agents required to handle incoming volumes is significantly less on social (3) compare to phone (25) and live chat (13).
Happier customers (high CSAT and loyalty)
- 82% of consumers reported sharing their positive experience with others after having their issue resolved on social.
- Corporate Executive Board found that customers will leave you for not meeting their expectations
- Low-effort customer service interaction will have a positive correlation to returned and increased spend.
Case study: Analytics
Analytics is a mission critical part of the Sparkcentral platform. For some of our largest clients, at any given almost 1000 agents would be actively logged into Sparkcentral responding to customer enquiries.
Analytics allows managers to measure and optimize the effectiveness of their digital customer care team and operations.
Data from these Reports are used to pay/hire/fire/coach agents, build a case to grow contact center teams, operations and prove ROI.
Discover & Define the Problem
When I joined, Analytics was the companies' primary focus. Our existing offering wasn't meeting customer needs and contributing to customer churn and impacting new sales opportunities as it was not stacking up against competition.
Simultaneously, the company was in midst of the following significant changes.
1) A market repositioning to focus on attaining enterprise customers, a shift from small to mid-sized companies.
2) As a result of the market maturing and needs expanding, our product strategy needed to evolve to support private 1:1 messaging (Whatsapp, Mobile & Web messaging, Messenger, Wechat) between brand and end customer, a shift from public social (Twitter, Facebook) interactions.
3) Business expansion into the European market & support for GDPR compliance.
4) A system replatform to support 1) 2) & 3)
Research - customers
I spent hours 30+ hours interviewing primary users of Analytics to better understand their role + responsibilities, in addition to the "Jobs" they've hired Sparkcentral to help address.
The two personas identified were, Contact Center Manager and Business Analyst.
Contact Center Manager
Operational status & performance (31%)
- Maximize the number of customers an agent can assist
- Ensure digital care volume can be addressed with available agents without jeopardizing service-level-agreements (SLA's)
- Minimize the effort for a manager to continually ensure team are 'working the queue’
- Ensure time-sensitive, or high priority conversations are shared with stakeholders to avoid PR disasters
- Ensure agents are online and working when they're scheduled to be
Reporting analysis & comms (28%)
- Minimize the time required to prepare reports on digital care operations and activity
- Minimize the effort required to disseminate reports on a recurring basis to business groups
- Minimize the effort required to analysis data and identify trends that drive business efficiencies
Team management & performance (30%)
- Minimize the effort required to identify top/low performing agents
- Ensure agent activity is tracked to ensure they're productive and working as expected
- Ensure agents are trained and equiped with neccessary resources to effectively resolve a customer's enquiry
Business forecasting & planning (11%)
- Reduce business costs (resources, equipment etc) due to a more optimized and smooth running contact center
- Grow digital care operations in the organization
- Ensure trends (seasonal vs spontaneous) are incorporated into budgets and annual planning
Operations optimization & business reporting (50%)
- Minimize the effort required to analysis and prepare real-time and historical reports
- Ensure all data is available 24/7, is reliable and accurate
- Minimize the time to build contact center dashboards to share knowledge, and support digital care teams
- Ensure internal groups are supported with data to make business decisions (Eg. WFM, HR, Management)
Business process standardization (50%)
- Minimize the effort required to extract raw data to systematize reporting across all business apps/tools in a similar way
- Minimize the time required to manupulate data for business analysis
Research - Users
Data revealed several pain points with existing offering.
Pains - CCM
- Time wasted on repeat tasks such as generating reports
- Process inefficiencies (ie. manual select of 100 agent names to export data)
- Lack of management understanding in the value of digital care
- Lack of data / insights to drive business decisions
- Unclear what action to take to drive positive change
Pains - Analyst
- Lack of trust in data quality, reliability
- Lack of data to measure, optimize key business metrics
- Hard to integrate reporting into existing business processes
Research - Internal stakeholders
The underlying event data is the critical component of a reporting solution. To draw an analogy, building a house on a fragile foundation will only guarantee it's demise.
To better understand the system, I spent hours with my lead Data Engineer to learn about current architecture and foundation, system limitiations and metric logic.
- Unscalable architecture that won't address enterprise business requirements and GDPR
- Significant technical debt (calculation logic inconsistencies, dirty code, difficult to iterate) resulting in partial data loss and duplication.
- Performance issues and unscalable frontend/UX.
Discussion with sales revealed gaps in our offering compared to competitors.
I conducted analysis in Salesforce (our internal CRM system) to help to quantify impact and discover lost reasons.
- Lack of configurability (dashboard/report builder), interaction (data compare, filters, drill-ins), usability (shareability, permissions) and lack of insights to help answer key questions.
- More than $2M in sales revenue at risk due to lack of Reporting. In addition to $1.5M ARR at risk of churn.
- Several customers with a combined $1M ARR had recently churned to due inconsistencies in data in Reporting.
Customer support interface with customers on a daily basis that offers a constant stream and source of customer feedback for PM's.
1/3 of all customer support enquiries received were Reporting related. Unfortunately, due to how the system was architected, it was difficult for Support to answer recurring questions such as:
Inability to support customers' inquiries led to customer distrust in the data, and reduced use.
In addition, I conducted analysis in Zendesk (internal ticketing system) and Jira (bugs, defect count) to quantify the impact to build a case to help executives understand the enormity of the problem.
- Challenging to resolve reporting related questions due to system limitations. If a customer inquired about the accuracy of a metric eg. "What is causing the Avg. Response time outlier?", "Why wasn't Handle time calculated for this conversation?" or "I see 3 contacts waiting in queue, but Dashboard says 10." - it was extremely difficult for CS to be confident in the data, and over time lose trust with customers.
Competitor analysis further supported feature / functionality gaps.
Crafting north star
A north star is forward looking, years out, and a statement that the team can get behind.
The north star for Analytics was to: Measure and optimize the effectiveness of social and messaging customer service.
Measure --> Surface relevant and reliable data highlighting what happened
Optimize --> Support decision making through facilitated action
Effectiveness --> Operational Metrics: Response Time, Handle Time etc. Financial Metrics: Cost saving, customer deflection, CSAT to churn / loyalty
A north star sets the long term path for a product. To address the first phase Measure, I led the following:
Design, Develop & Deliver the Solution
- Rebuild trust by delivering accurate reporting data
- Ensure we're defining and delivering the right metrics and KPI's to support operations
- Reduce customer churn due to reporting offering and improve customer satisfaction
I presented current state with findings from internal/external interviews to the executive team together with prioritized themes and a plan on how I would approach each phase.
Clarifying what isn't in scope is equally as important as what is included. I made sure the initial focus for phase Measure was existing customers, and as a result new business opportunities was deprioritized.
To avoid further customer churn, feature development for 2Q's would be focused on Phase 1: Measure, retaining customers by addressing the top pains:
1) Data inaccuracies and integrity
2) Access real-time queue stats to better organize teams & drive tactical decisions
2) Access volume and key performance stats to identify trends and guide strategic decisions
Data inaccuracies and integrity
- Unexplainable anomalies: When drilling into the data to understand the cause of a spike in metric eg. Avg. Response Time or Handle Time, a manager would not be able to identify the conversation that caused the anomalie. Over time, this led to distrust in data.
- Metric calculation logic inconsistencies: The formula to calculate a metric varied across digital channels, the logic was poorly documented and not maintained as new product functionality was introduced that would affect the workflow.
- Data loss & duplication: Engineering estimated ~15% of data ingested into was lost or duplicated, that had a concerning impact to data in reports.
- All critical metrics defined and built, with detailed specifications to support customer support in their customer comms
- Reporting infrastructure v2 built in a scalable way to support logic tweaks & new metrics down the road
All KPI's were benchmarked with customers and call center standards.
- Positive feedback from customers on metric logic calculations
- Point increase in CSAT
- Managers need in the moment visibility into queue stats and agent availability in order to ensure they have the people necessary to address incoming volumes across all digital channels
- A current view of queue status and key metrics, updating every one-minute interval
- Filterable by one, some or all digital channels
- Overview of channel activity, agent availability and conversations context (tag sentiment)
High fidelity mock
- 90% uptake within 2 days of launch. Managers kept the dashboard open all day on a 2nd screen to respond to influx of volume
- Customer feedback that they were able to address negative customer complaints in real-time and avoid a PR nightmare
Historical / trends reporting
- Managers need a historical view of key productivity metrics to identify trends in data to ensure customers' enquiries are reoslved in a timely manner, and within SLA eg. historical analysis could highlight volume spikes during weekend, and help to build a business case to support customers 24/7
- Managers need a way to manupilate the data to compile their daily/weekly/monthly reports
- Reports that highlighted top metrics including volume, avg. response time, avg. resolution time and team performance.
- Ability to easily export data to support drill-ins
- 80% users (Manager) DAU
- Highly manual to create internal reports using data from Sparkcentral
- High effort to analyze data across multiple digital channels
A simple interface to export all report data across one, some or all channels. This raw export gave managers and analysts the flexibility to organize data in preparation for meetings, or reports.
- Managers saved 10-15 hours a week due to process efficiency
- Happy customers! Positive CSAT