These two case studies will look at CRM data cleaning, routing, and reporting at 6sense - and how we approached GDPR research and design for enterprise customers at Informatica.
A new AI-driven developer API and doc portal that provided curated APIs and content and reclaimed countless hours of lost productivity.
Canvas-based decision tree tool that allows Ops users the ability to keep their CRM records accurate through automated enrichment, de-duping, and more.
In 2019, companies scrambled to figure out how to stay in GDPR compliance. We helped several large enterprise customers adapt.
These examples from 6sense and Oracle show my approach to analytics and dashboards. Simplicity, relevance, and personalization are covered.
Interviews included 1-on-1s and system walk-throughs across many groups - to get a sense for horizontal themes, behaviors, and pain points.
I worked with the lead engineer on the team to establish basic expectations on process and time, and tradeoffs that may need to be made in the short-term to build momentum.
Primarily used as a solid starting point for conversation (with later validation from engineers), I leveraged Claude to help me map out common user flows to provide the basic context we would use to talk through how future APIs might be organized.
It did this by searching through and analyzing github repositories, internal docs, Google docs, and publicly-available Slack channels.
Groups agreed on proto file standards across groups, allowing a subset of attributes to be consistent horizontally. Claude agents would run daily and index the correct APIs based on creation and last updated dates, deprecation and version flag values, product category and subcategory, and description attributes.
For the portal, the determinism that would build trust with engineers sat underneath while AI assistants allowed conversational searches based on product context.
This process not only simplifies, automates, and improves productivity, the curation means the same API means the same thing in every group. While observability monitoring for drift is still necessary, proto files provide the deterministic grounding that make it much less likely.
While the initial AI Assistant primarily allows searches through indexed APIs and description fields, this would later include other document repositories after appropriate strategies for integration had been decided upon.
With most engineers relying on manual searches and lengthy Slack conversations on an average day, this was expected to reclaim massive amounts of unnecessary search and discovery time.
Even conservative estimates using average engineer salaries and average time wasted in searches, could reach hundreds of thousands of dollars saved and more customer needs addressed faster.
Through 1-on-1 meetings (virtual and our annual conference) and through CSM feedback, we learned from MarketingOps users that:
Aggregation of all feedback led to high priority themes we could target for the first release - to begin laying a foundation for future enhancements.
Side-by-side with our Audience Workflows product (omni-channel marketing automation) we explored different patterns that would provide a balance between simplicity, flexibility, and scale. The decision tree pattern was the only format flexible enough to adapt to any customer scenario imaginable, from continual branching logic to enterprise scale.
Like Audience Workflows, Data Workflows was offered free with any platform package and only billed users on specific actions taken (credits). Many of the automation features within Data Workflows stood to be a new revenue stream for the company.
Unlike Audience Workflows, which primarily focused its functionality around Marketing, Data Workflows was foundational to the entire platform and needed to work seamlessly with both Marketing and Sales Pillars.
Figma
After seeing the extent of features I would be designing, I organized Figma files based primarily on which logical groupings of features would grow over time (new nodes, integrations and templates).
Everything else was divided by logical grouping to help other designers navigate through the team folders.
Mockup Libraries for PMs and Engineering
With so many moving parts (80+ nodes, 30+ templates, etc.), I kept the team organized by creating Figma libraries, each with a Table of Contents that was organized with a hierarchy that matched the app layout.
Figma links became a defacto standard for each UI-related Jira ticket (so engineering would always have clear reference - and always from the latest version).
Through alpha, beta, and usability testing, we gained insights into how users thought about, and approached, workflow strategies - addressing pain points long before GA release.
We were also able to better understand which personas typically performed different tasks and where certain tasks became more of a collaborative team activity.
The following were common themes across verticals:
Users loved:
More attention was needed for:
While a strong functional foundation was developed for the GA release, the team already had post-GA designs that help users visualize and manipulate complex workflows, including node groups, expand/collapse for subflows, and editing the flow with an AI assistant.
With GDPR regulations being new to everyone, we needed three core pieces of information quickly:
By talking with companies onsite, at the RSA security conference, and at virtual Zoom follow-up calls, we learned that:
The biggest challenge to the team, resolved through collaboration and workshops, was mapping out cross-product flows across different data governance products. Compliance is rarely relegated to one individual - and is typically part of a larger workflow - with multiple touchpoints.
We also needed to consider how the functionality could be sold as a stand-alone solution, using ServiceNow templates and other third-party integrations.
The data privacy solutions at Informatica are now embedded within Data Catalog functionality and workflows.
While we wanted to offer a third-party option for customers (or at least simple methods of API integration with workflow tools like ServiceNow) most effort was placed on data privacy as a value ad to the cloud platform, the company’s best selling suite of products.
Because of the speed that large enterprises would need to adapt, selling Professional Services hours for initial onboarding was critical.
The success of our Customer Lifecycle Management app became a double-edge sword. That success had more leaders wanting to influence changes to the dashboard for their own department’s needs. With several Ops leaders adding to the mix, the end result became an experience that was visually overwhelming, inconsistent in format, and prevented at-a-glance insights.
I setup a short series of meetings with PMs and Ops leaders to discuss a different approach. It would require users to make tough decisions on what metrics are actually needed immediately after login and whether charts are actually critical for this type of dashboard.
Through those discussions, I was able to recommend groupings that better organized the main themes of the dashboard. I suggested a default filtering and sorting (hiding the advanced filters by default) and used a simple, consistent format for drilling to deeper details.
6sense Data Workflows dashboards needed to be able to accommodate two primary use cases. As a user, I can see what needs attention and
Being able to launch a workflow directly from the dashboard helps users complete the circle from creating a flow > to measuring success > to creating or fine-tuning a flow for even better success in the next published run.
Oracle CSMs monitored customer usage from two perspectives: