Revenue Analytics Dashboard for Verve Studio

A multi-dimensional revenue analytics dashboard for Verve Studio, a subscription creative-software company, combining current performance monitoring, historical trend analysis, and seasonality-adjusted forecasting.

Designed and built an end-to-end revenue intelligence dashboard for Verve Studio's e-commerce and growth team. The project started from a wireframe I sketched out defining exactly how the dashboard needed to look and behave — which filters mattered, what "top-level" performance meant, and which funnel and segment views the business needed — and I carried that spec through to a working interactive prototype.

The dashboard unifies three views that are usually scattered across separate reports — real-time KPI monitoring, historical trend analysis, and forward-looking forecasts — into a single, filterable interface, with drill-down across 9 business dimensions (region, market, login/entitlement status, product, segment, purchase type, plan type, surface, device, and marketing channel) and dedicated explorer views for funnels, segments, and geography.

Role: Data Analyst / Dashboard Designer (solo) Tools: Requirements gathering & wireframing (Excel), React, Recharts, statistical forecasting (trend decomposition + seasonal indexing) Deliverables: Interactive dashboard prototype, filter/data framework specification, forecasting methodology

Verve Studio sells a suite of creative applications (Verve Photo, Verve Design, Verve Motion, Verve Docs, and the Verve All-Access bundle) across consumer, education, team, and Docs-only segments. The growth team needed a single view of revenue health that answered three questions at once: "How are we doing right now?", "How did we get here?", and "Where are we headed?" — without stitching together separate weekly reports, historical exports, and finance forecasts by hand. Funnel performance was especially hard to see clearly, because Verve runs four structurally different purchase funnels (a guided-selection path, a direct-recommendation path, a plan-commitment path, and a trial-to-pay path), each converting differently by segment.

I owned the dashboard end-to-end: turn my own wireframe into a concrete data and filter taxonomy, decide what "current," "past," and "forecast" should each mean operationally, choose a forecasting method appropriate for a seasonal, growing revenue base, and design/build the interface itself — including dedicated explorer views for the four funnel types and four customer segments.

1. Framework design, from wireframe to spec. Before touching any code, I laid out the full requirements in a wireframe: which filters the dashboard needed on top (region, market, login/entitlement status, product, segment, purchase type, plan type, surface, sub-surface, device, marketing channel), what the top-level KPI row should show (Visits, Web Orders, CVR%, Net Revenue with a QTD/Weekly toggle), and which dedicated views were needed beneath it — funnel views per purchase path, segment-level funnel and conversion trend views, a product/segment share table, and a geographic view. That wireframe became the direct spec for the build.

2. One global filter bar, not per-widget filters. I kept the filter bar global across every chart on the page, rather than local to each widget, so a single selection (e.g., "EMEA, Education, Mobile") re-contextualizes the entire dashboard — KPIs, trend chart, funnels, and segment mix all update together instead of requiring the user to re-filter each panel separately.

3. Defining "current," "past," and "forecast" as one continuous system. Rather than treating these as three separate reports, I designed them as one narrative on a single timeline:

Current trends — KPI cards (Visits, Orders, CVR%, Net Revenue) with a Weekly/QTD toggle, each showing period-over-period delta and a 16-point sparkline for immediate directional context.

Past trends — a 104-week (2-year) actuals series, giving enough history to distinguish real seasonal patterns from noise.

Forecast — a 12-week projection that visually continues directly off the actuals line, so the viewer reads it as "what happens next," not a disconnected chart.

4. Forecasting methodology. I used a seasonality-adjusted trend model rather than a naive moving average, since Verve's revenue has a clear annual cycle (holiday-quarter lift, January dip):

Fit a linear trend across the historical series via least-squares regression.

Deseasonalized the series (actual ÷ trend) and averaged by calendar month to derive 12 monthly seasonal indices.

Projected forward by re-applying trend(t) × seasonal_index(month(t)), with a confidence band that widens with forecast horizon.

Surfaced the seasonal indices themselves as a small reference strip beneath the chart, so stakeholders can see why the forecast bends the way it does, not just the resulting line.

5. Funnel Explorer — four purchase paths, one comparable view. Verve's checkout experience isn't one funnel; it's four, depending on how a visitor is guided to a plan. I built a tabbed Funnel Explorer covering the Guided Selection funnel, Direct Recommendation funnel, Plan Commitment funnel, and Trial-to-Pay funnel — each with its own stage sequence, drop-off rates, and a weekly conversion-rate trend, so the team can compare "is Trial-to-Pay converting better than last month" against "is it converting better than Guided Selection" in the same interface.

6. Segment Explorer — per-segment funnel and conversion trend. Beyond the top-line revenue chart, I added a segment-level explorer for Consumer, Education, Team, and Docs, each with its own funnel and its own weekly CVR trend, plus a Logged In / Logged Out toggle — since login state changes conversion behavior meaningfully across segments (a logged-in Team buyer converts very differently from an anonymous Consumer visitor). This mirrors the "which segment is punching below its weight" question stakeholders actually ask in review meetings.

7. Geo view and segment mix. I added a geography view comparing visit share against revenue share by region, flagging regions that over- or under-convert relative to their traffic — and a segment mix table comparing entry share vs. order share across Verve's product line, surfacing products that draw disproportionate traffic but under-convert, or vice versa.

8. Interface decisions. I used a restrained, data-forward visual language (navy/teal/amber palette, monospace numerals for scannability) so the design supports fast reading of dense metrics rather than competing with them, and built the KPI-to-detail hierarchy so a viewer can get the headline number in under 3 seconds and the "why" in the next 10.

Turned a written wireframe spec into a fully interactive prototype covering every view the requirements called for: top-level KPIs, historical trend, forecast, four funnel types, four segment breakdowns, geo view, and product/segment mix.

Replaced what would otherwise be three-plus fragmented spreadsheet reports with a single filterable source of truth.

Reduced the analytical "why did revenue move" question to a single dashboard view, by pairing the trend line with funnel, segment, and geo context in the same screen.

Forecast transparency (visible seasonal index) gave stakeholders a way to sanity-check projections against known business cycles, rather than treating the forecast as a black box.

The hardest part of this project wasn't the charting — it was resisting the urge to build a separate dashboard for every stakeholder question. Designing one filter framework and one explorer pattern that could answer questions across four funnel types and four segments, with current/past/forecast as one continuous read rather than three disconnected reports, was the actual analytical work; the visualization layer came after that decision was made.

Note: Verve Studio and all product names in this project (Verve Photo, Verve Design, Verve Motion, Verve Docs, Verve All-Access) are fictional, created for this portfolio piece. The underlying filter taxonomy, funnel structure, and forecasting methodology reflect real dashboard design work and are illustrated here with mock data.