A reach-out tool for the onboarding ops team. Surfaces who's stuck, when, and what they told us — so calls are quick, informed, and well-timed.
Currently tracking the v8 onboarding flow (view live).
What's on the page
- KPI strip.
- Four cards across the top — signups today, bank-connects today, 7-day completion rate, live count. Each one carries an industry benchmark line so you always know if the number is good or bad.
- Queue view (default).
- Sortable, filterable table of signed-up users who haven't activated. Sorted by priority (Ltd + post-aha first), then by most-recently-stuck. Click any row to drill in.
- Kanban view (toggle above the board).
- Same dataset, spatial layout. Seven lanes following the v8 four-JTBD spine — Anonymous · discovery → Anonymous · profile + tax aha → Anonymous · meet the team → Just signed up → Sending €1 invoice → At bank connection → Activated. Useful for spotting clusters ("five Ltd users sitting on bank-connection right now").
- Events view (toggle above the board).
- The same Kanban board, but each lane header also lists the GA4 events that fire at that milestone (specified, not yet live). Kept as a separate view so the plain Kanban stays uncluttered for daily ops. Full list on the Event spec page.
- Funnel chart.
- All 22 onboarding steps as a vertical bar chart, biggest drops highlighted red. Toggle between "all sessions" and "signed up only". Click any step to filter the queue to people stuck there.
- Segment breakdown.
- Signup → activation rate sliced by Q1 situation, top worry, sector, and biz type. Surfaces which buckets convert and which leak.
- Benchmarks panel.
- Each headline metric next to its published industry range with a verdict chip (above / at / below), so you can tell good from bad at a glance. Our values are from the mock data; the ranges & sources (Mixpanel, Userpilot, HBR, Heap, OpenView) are real.
- Data scan.
- Ranks funnel gaps by the real drop computed live from the session data (biggest leak first), benchmarks each against a cited industry range, and names the v8 screens in the gap. The "why" is a clearly-tagged hypothesis to verify, never stated as fact — and it re-ranks automatically when real GA4 data replaces the mock.
- Paywall funnel.
- The post-onboarding trial→paid journey (day-4 nudge → day-7 → subscribed). Card-up-front trial, so "subscribed" = stayed and the gap is churn. Reach %s are illustrative — the paywall is built under /paywall/ but events aren't firing yet.
- Session drawer.
- Slide-in panel opened from any queue row or Kanban card. Shows the user's full questionnaire answers, screen timeline with per-screen dwell, and a notes log so the team can record outreach attempts.
Design decisions
- Built for the ops team, not product or sales.
- The page is optimised for individual user drill-down + a daily call list. Product gets the funnel + segment views; sales gets a "qualified-and-stuck" filter. But the primary user is whoever's calling people today.
- The win is "bank connected", not signup.
- Signup is easy to hit but doesn't mean anything — lots of signed-up users never come back. Bank-connect is the first moment Bullet delivers real value (live transactions, tax categorisation). That's our activation event.
- Identity flips on signup (step 15).
- In v8, create-account sits late — step 15 of 22, after the whole meet-the-team trust block. Sessions before that are anonymous: they show up in the funnel so we see top-of-funnel drop-off, but they can't appear in the reach-out queue because we have no way to contact them. The flip side is the queue only ever covers steps 16–21 (send-invoice through compliance-calendar).
- Stuck threshold: 1 hour idle.
- Industry research is clear — response rates to outbound outreach decay sharply after the first hour. Someone who signed up at 9am and didn't connect by 10am still has Bullet on their mind. Wait 24 hours and you've lost most of them.
- Manual outreach, not automation.
- v1 surfaces the list, the team calls. Automated re-engagement emails come in v2 once we know the manual playbook works.
Industry benchmarks used in the metrics
- SaaS activation rate (signup → first value): B2B median 25–30%, best-in-class 60%+. Mixpanel Product Benchmarks 2024.
- Drop-off at signup step: 40–60% of visitors who start onboarding don't complete signup. Userpilot SaaS Onboarding Report.
- Outbound response rate by speed-to-call: <5 min ~ 30–50%; 1 hour ~ 10–15%; 24 hours ~ 3–5%. Harvard Business Review lead-response research.
- Time-to-activation, B2B SaaS: simple products 5–15 min; complex products 1–3 days.
- 7-day retention from signup: ~30% B2B SaaS median. Heap onboarding cohort study.
Not in v1
- Automated re-engagement emails — v2 once manual playbook is proven.
- Slack alerts when high-value users drop — v2.
- A/B test comparison views — v7a vs v8.
- Cohort retention — v8.
- Anonymous-user device fingerprinting — privacy debt, deferred.
Changelog
- 10 Jun. New Setup review lane after Activated — the post-onboarding phase. Mariana verifies each activated customer (bank feeding, Bullet account configured, formed correctly) and answers their questions before accounting access goes live. Cards show verification progress + open questions; the session drawer gets the checklist, the question thread, and a “grant accounting access” action. Mock data, same as the rest of the board.
- 7 Jun. Remapped tracking to v8 (now the source-of-truth onboarding). Topbar pill, live + Overview links, About copy, the Data scan's screen labels, the Event spec and Settings notes now all point at v8. The 22-step flow itself was already v8-shaped (v8 forked from v7a with the same spine), so the funnel/lanes/events are unchanged — this is a label + link sync. Dated entries below are kept as history.
- 1 Jun. Added a Data scan — ranks funnel leaks by real drop (computed live), benchmarks each, names the v7a screens, and pairs each with a labelled hypothesis + how to confirm it. Distinct from the design-review Live scan.
- 1 Jun. Built the paywall flow (/paywall/) and added a Paywall funnel here (trial → paid). Topbar now links the paywall too. Illustrative numbers — events specified, not firing.
- 31 May. Added a Live scan — a plain-English, screen-by-screen review of the v7a flow (where it leaks + what to do, by priority). Added a Benchmarks panel — each headline metric beside its industry range with an above/at/below verdict chip. Also made the whole dashboard mobile-friendly.
- 30 May. Added a Settings page (top bar) to add/edit/reorder the GA4 events with no code. Edits apply live across the dashboard, Events view, and spec page (saved in your browser); Export hands a developer a drop-in
events-data.js to commit.
- 29 May. Remapped the dashboard internals to v7a: the funnel, lanes, and mock data now follow the real 22-step v7a flow (was still v3-shaped). Each Kanban lane now lists the GA4 events that fire at that milestone, and the session drawer shows a per-user event timeline. Added an Event spec link in the top bar. Events are specified, not yet firing — presentation mock.
- 28 May. Switched the tracked onboarding flow from v3 to v7a (the locked, current onboarding). Topbar pill, live link, and roadmap now point at v7a.
- 25 May. Kanban is now the default view. The Queue tab is still one click away in the top bar — flip whenever you want the table format.
- 22 May. Plugged into v3. Topbar shows the tracked version + a direct link to the live flow. Underlying FLOW screen paths now resolve to
/v3/screens/….
- 22 May. Removed the in-dashboard Recommendations module. Recommendations belong in the conversation alongside real funnel data — surfacing them inside a mock with synthetic numbers undercut their authority.
- 22 May. Added Kanban view (toggle in top bar). Same dataset as the queue, grouped into 7 journey-milestone lanes.
- 22 May. v1 shipped — KPI strip, reach-out queue, funnel chart, segment cards, session drawer.