Analytics Dashboard

Problem

Attentive is an SMS marketing platform for nearly 4,000 brands, driving ~20% of their online revenue and delivering billions of messages a week to 200M+ subscribers. But marketers could not understand their program performance, lost trust in the numbers, and left for competitors.

1. ~$1M a year in ops costs, from Jira tickets routed to solutions engineering and technical account managers for custom data.
2. Each CSM spent 1 to 4 hours every week fulfilling custom report requests for their clients.
3. The legacy Reports tab was painful and clunky to use, confirmed by a UX audit of annotated screenshots.
4. Marketers had a long tail of performance questions, and no self-serve way to answer even the most basic ones.

Solution

Rebuild Reports as an Analytics tab for deep dives, and reframe Dashboard as Home, a personalized command center. Accurate data and relevant insights first.

1. Research: interviewed customers on why they needed performance data and what they would do with it, audited the legacy Reports tab, and ran a competitor gap analysis against Klaviyo and Shopify.
2. One-day design sprint in July with the PM, Design Systems Lead, front-end eng, and 3 designers: defined the future of data viz at Attentive, and four principles, easy to understand, invites discovery, answers a question, inspires action.
3. Two reference designs, stacked bars vs side-by-side lines. Concept testing with CS picked lines: clearer at a glance, easier to handle responsively, and no accessible color palette for data viz existed yet.
4. Narrowed the metrics to four categories, in order: Subs, Revenue, Message Send, Billable Spend. Subs first, because revenue means nothing without subscriber context. Send and spend measure productivity per dollar. Revenue vs spend gives ROI for Campaigns vs Journeys.
5. Built on Nivo.rocks, the first chart library implementation at Attentive. Worked out the x-axes display rules with eng based on data points and time grouping, plus side nav updates and mobile responsiveness.

Outcomes

1. Beta launched Thursday, September 16, 2021 to 15 clients, measured with a 3-question Sprig microsurvey and Heap tracking of scrolls, clicks, and hovers.
2. 160+ clients rated Analytics 8.37 out of 10 in the October 1 CEO feedback survey, a sign of extreme interest.
3. The four metric categories answered the most common performance questions first. The rest of the long tail stayed open, a known next step, not a surprise.
4. Established the data viz principles and chart patterns the platform reused afterward.

Reflection

1. Trust is the product: accurate data and relevant insights had to come before anything else.
2. Simple first: as the first chart library implementation, the design had to be easy to grasp yet comprehensive enough to cover the top questions.
3. Data governance is the hidden work: finding the eng owner and the source of truth for metric calculations, like active subscribers, took real time.