Creation of a turnkey digital telehealth platform

When we set out to build a telehealth platform we wanted to make it intuitive, easy to use, metrics-based, and expandable. It had to have most of the features of common Electronic Health Records (EHR) applications because it would be sharing information with them. So we included Patient Health History, Family Medical History, Medications, and Appointments integrated with Zoom. As Remote patient Monitoring was bursting on the scene we took a deep dive on graphical representation of vital signs trends with alerts triggered by boundary settings.

The next big part of it was to add a CRM capability since we expected to receive leads that would need to be sold on the idea of daily monitoring. Drawing from our experience at HotChalk with student recruitment, we built a stage-status workflow and linked it to an email service provider that sent emails and SMS messages. Matching patients to providers weas done by creating care groups within the multi-tenant setup.

The UI/UX design was farmed out to an agency with experience in healthcare and we were very pleased with the process and the result. In record time we had a world class digital telehealth platform with remote patient monitoring that is flexible enough to handle Direct-To-Consumer, Clinics, Hospitals, Home Health and Residential Care.

Student recruitment and enrollment platform (SRM)

The problem presented to me at HotChalk mostly revolved around time. A university partner had been contracted with and they expected us to have Student Relationship Management capability on day 1 and we were behind. When I arrived we had spent a few months modifying an open source CRM system for Online Program Management (OPM). But it was rife with problems, poor reliability and uptime, difficult setup, and a call center integration that was slow and inadequate.

We switched offshore teams from one in China to one in Vietnam with a Silicon Valley presence. Then we hired a DevOps contractor who would rebuild the stack and do all the releases so the development team wouldn’t be interrupted and also so they wouldn’t have access to personally identifiable information (PII).

Within a few months the uptime improved to 99.99%, a new call center integration was created that met the sub-minute callback speed needed to enroll patients before they left the HotChalk lead generation websites. Enrollment yields and enrollment counselor efficiency improved dramatically and degrees began flowing to thousands of online students, mostly teachers who were required to have Master of Education Degrees.

K-12 Learning Management System (LMS)

In 2004 we set out to build one of the first Learning Management Systems (LMS) for K-12 classrooms. We were so early that that name hadn’t been given to this sector yet. We started small with tools for Lesson Plans and Assignments, and later add Quizzes, Gradebook, Alerts. The online environment would connect Parents, Teachers and Students and have views for each. A problem with being so early was that many schools didn’t have Internet access yet. And most administrators and teachers were unfamiliar with technology. So we had to provide infrastructure and education.

By going SaaS we avoided the lengthy and political school budgeting process but we had to make money. So we partnered with NBC Universal and McGraw-Hill to sell premium content, program clips, news videos and continuing education courseware.

Many thousands of teachers signed up but it never became a commercial success until our lead generation websites like LessonPlansPage.com came online. The we pivoted to university partnerships to enroll out database of teachers and our website visitors in online degree programs.

Agile Transformation

I was hired at a privately owned financial services software company to improve the PMO process and Quality of execution. And that was about all the direction I got from the CEO/Owner I reported to. He showed me a list of what he called ‘Initiatives’ that were actually hundreds of product features that had been bottlenecked in development. There were 35 Agile Scrum Teams all working on the same platform, divided into 6 departments, each led by an Engineering and a Product Director. My approach was two-fold:

First I worked to improve the Agile Development Process. I replaced the Agile Coach with someone who was willing to try some non-standard approaches. Since there was no way to get to all 35 teams each two week Sprint period, we decided to use an online survey for the teams to do self-assessments 2-3 times a year. And as the Coach visited the teams he would update the in-person assessments, focusing on the teams that needed the most help such as re-training. The scorecards we created are shown below.

Next I took the list of Initiatives and facilitated a ranking process with the engineering and Product executives. We ranked them based on effort needed to complete, strength of customer need, and fit with other aspects in the release. We rimmed the list to a manageable number and broke it into releases. We needed a visual to measure progress. The home-grown ticketing system has no UI for this. The company had a rotation of co-op students and one was assigned to me and I asked him if he could build a dashboard on top of the data ll the teams were entering in the system. Working with and learning from some of the senior engineers he built a prototype. Then three more co-ops followed him and added refinements and expanded the dashboards. The result staked up well to any commercial product out there and was fit to our needs.

New product began to flow again for the first time in years, customers were happy, new ones were added and the company was acquired by a large financial institution.

Customer Preference Analytics

The idea for Venalytica came from observing my wife try to find the right coffee maker. When most brands were advertising slick control panels, timers and IoT, all she wanted to was pout water in and get pot of coffee in seconds. She had a different set of weighted preferences than other consumers.

Likewise, I was repeatedly frustrated when buying electronics off the web. The filters that were set up to narrow or widen your choices seemed to me to be unsatisfactory. Either I’d end up with too many equally weighted choices or none. And like my wife, I placed a different amount of importance on some features that others. I wanted a weighted result set like you used to get from a search engine.

Having used Simple Multi-Attribute Rating Technique and Analytical Hierarchy Process for Portfolio Management, it seemed to me that some technique could be blended with weighted choices to provide a ranked result set. At the same time the accumulation of data from these preferences could be used in merchandizing and product design. I connected a POC model for laptops to the BestBuy catalog and tried it out. The picture below shows one version of the concept.

Program Execution

In 2000 most enterprise software platforms still required the installation of local thick client for all users. This created an IT nightmare for the various platforms that the product had to be compatible with and a tremendous expense. I was brought in as employee #1, second to the founder, to manage an offshore team building a browser-based product portfolio management (PPM) suite. We would initially focus on a program management module but would also design a portfolio management piece, a resource management add-on, and a connector collection.

The program management part would be based on the waterfall or Phase-Gate model in use by most product companies at the time. A scheduling engine would enable a hierarchical representation of Phases, Tasks, Milestones and Summaries. To the standard Microsoft Project Gantt Chart we added an icon for deliverables, documents needed for completion of a development task and branded our approach as Deliverables-based Program Management. The idea was to complement the many products in use for document workflows.

To test our assumptions we acquired three pilot customers, one literally was an aircraft maker, one was a hydrogen fuel cell manufacturer, and the third one was an automotive supplier. This was enough to attract interest from a Bill-of-Materials software provider that also had taken a web-only posture and who felt our product would complement theirs. After a short co-development contract our product began selling through their customer base and they acquired us.

Industrial Experimentation

In an earlier life I developed expertise in Design of Experiments (DoE) where I used statistical analysis, Taguchi Methods and signal-to-noise, to identify the variables that had significance to the key output of systems. I put this to test in printing waste and make-ready, in glass and dinnerware production, and at the manufacturer of steel shelving for home goods megastores.

In one experiment, the cost the of powder coat process was killing the company due to the labor needed for touch-up as well as the wasted expensive powder that never adhered to the shelving parts. A cross-functional team was assembled to brainstorm the variables, set up and conduct the experiment, and evaluate the results. A welder on the team suggested that the parts needed to be fixtured inside the spray booth in a way where they more perfectly centered and align to the spray guns. Belt speed, pressure, temperature and other systems variables were also assessed and optimized. The result was far less less powder wasted, fewer people needed and a far more uniform high quality finish resulting in millions saved per year.

A second experiment was conducted on pickled and oiled cold rolled steel. This product cost more per foot than cold rolled steel that is not pickled and oiled. But the reason to use the more expensive product was to reduce dust and tool wear. The experiment was a side-by-side comparison of tool wear on 6 punches. While the difference in tool wear was significant, it did not prove economically beneficial. And millions in savings per year would result in switching to cheaper steel, performing daily cleanup, and buying tools more frequently.