A guided EV charging decision tree — for stakeholders who need a real answer, not a black box.

Sizing EV charging infrastructure is the kind of question where the right answer depends on a dozen interacting variables. How many vehicles? What’s the duty cycle? Are users dwelling for eight hours or eight minutes? Is this an employer subsidy, a fleet operations decision, or a public-access build? ATI’s research team had already mapped the decision logic into a Lucidchart that ran the full length of the problem space — but a chart isn’t a tool a stakeholder can use on their own.

The brief: build that decision tree into an interactive web experience. Three verticals — Employee, Fleet, and Public. A transparent results page that shows the stakeholder why the recommendation came out the way it did, not just what the recommendation is. An admin surface where the ATI team can keep the data inputs (adoption rates, growth projections) and educational content current without filing a change request.

A research-center tool with a long, evolving decision tree has two failure modes: it gets hardcoded and rots the minute the team learns something new, or it gets over-engineered into a CMS the team can’t actually operate. The architecture below splits the difference.

Single deployment

Payload CMS 3 runs natively inside the Next.js app on Vercel. One application, one deploy, one URL. No separately-hosted admin panel for the ATI team to operate.

Config-driven engine

Each vertical’s decision tree — branches, inputs, formulas, constraints — is stored as structured JSON. Fleet and Public in Phase 2 are configured, not rebuilt. The same engine renders all three.

TypeScript formulas

Power, simultaneity, energy, efficiency, and charger eligibility live in a pure TypeScript utility layer — independently testable, easy to validate against the Lucidchart source of truth.

Stack

Next.js (App Router) · Payload CMS 3 · MongoDB Atlas · Vercel · Tailwind. The whole platform is in a technology the ATI team can hire for and ship against for the long term.

The Employee vertical ships first as a proof of concept — the full design system, the decision engine, and a complete working flow that ATI can put in front of stakeholders before we commit to the remaining two verticals. Four weeks from kick-off to staging.

Discovery

Decision-tree audit against the Lucidchart source. Every branch, input, and formula documented. Variables split into user-entered (number of employees, workday duration) and team-maintained (adoption rate, projected growth).

Design

High-fidelity Figma mockups for landing, the question flow, the calculation summary, the results page, and the educational content surface. Designed for transparency — every output traceable back to an input.

Decision flow

Full Employee-vertical interactive tree, with the branching paths the Lucidchart calls for (5-year vs. 10-year planning horizon, Level 1 / Level 2 / DC Fast Charging eligibility).

Admin portal

The ATI team edits adoption rates, growth projections, educational copy, and disclaimers through the Payload admin — quarterly or semi-annually, without a ticket to us.

Educational pages

Charger types, cost ranges, and post-estimate considerations as first-class content — not buried in a footer. The stakeholder leaves with both an estimate and the context to act on it.

QA & launch

Cross-browser and device testing. Formula validation against the Lucidchart source. Two rounds of revisions on ATI feedback before staging hands off to production.

With the engine and design system established in Phase 1, the remaining two verticals are configuration work, not a rebuild. Fleet brings its own inputs — average distance traveled, fleet size, dwell time — and its own formulas. Public brings a different set again, with the access and utilization patterns of an open-to-everyone charging deployment. Both slot into the unified tool experience so a single stakeholder visit can move between verticals without losing context.

The Phase 2 estimate is provisional by design. Once Phase 1 is in stakeholders’ hands, ATI will know more about where the friction is — and the Phase 2 build gets sequenced against what we’ve actually learned, not against what we assumed at the start.

A Lucidchart is a research artifact. A tool is a deliverable. The difference is that a stakeholder can show up alone, answer a series of clear questions, and walk away with a defensible initial estimate — and the ATI team didn’t have to be in the room for it.

Everything they need to keep that tool current sits inside the same web application: the data inputs, the educational content, the disclaimers, the about page. When the adoption-rate model shifts, the tool reflects it within a quarter. When the research expands to a new vertical, the engine accommodates it without a new project. That’s the long-term shape of the deliverable.