/* global window */
// Content object for the article renderer (blog/blog-article.jsx).
// Produced by the Writer→Editor→Humanize→Proofreader chain; draft in blog/pipeline/drafts/onboarding-clarity.md
window.__ARTICLE__ = {
  slug: "onboarding-clarity",
  laneLabel: "FIELD NOTES",
  kicker: "ONBOARDING",
  readMins: 8,
  dateLabel: "Apr 2026",
  title: "Onboarding clarity: where new users stall, and the gap between activation metrics and felt experience",
  deck: "Activation dashboards say users “completed onboarding.” The recordings say half of them never understood what they finished. A field note on the gap between the metric and the felt experience, and how to close it with evidence instead of funnel math.",
  tags: ["onboarding", "saas", "methodology"],
  toc: [
    { id: "contradiction", num: "01 · THE CONTRADICTION", title: "The chart is green. The user is gone." },
    { id: "number", num: "02 · THE BENCHMARKS", title: "The number that isn't the number" },
    { id: "blindspot", num: "03 · THE BLIND SPOT", title: "What the funnel can't see" },
    { id: "drift", num: "04 · WHY THEY DRIFT", title: "Why “completed” and “felt value” drift apart" },
    { id: "evidence", num: "05 · CLOSE THE GAP", title: "Closing the gap with evidence" },
  ],
  body: [
    { t: "h2", id: "contradiction", num: "01 · THE CONTRADICTION", text: "The activation chart is green. The user is gone." },
    { t: "p", html: `That contradiction sits at the center of most onboarding reviews I've watched. A product lead pulls up the dashboard, points at a healthy completion number, and asks why retention is still leaking. The dashboard says the cohort finished onboarding. The recordings say half of them never understood what they finished.` },
    { t: "p", html: `This is the gap between the metric and the felt experience. <em>"Completed onboarding"</em> is an event your instrumentation fired. It is not a state of mind. And the two diverge more often, and more quietly, than the dashboard lets on.` },
    { t: "pullquote", text: "“Completed onboarding” is an event your instrumentation fired. It is not a state of mind." },

    { t: "h2", id: "number", num: "02 · THE BENCHMARKS", text: "The number that isn't the number" },
    { t: "p", html: `Start with what the benchmarks actually say, because the headline figure flatters you.` },
    { t: "p", html: `Across SaaS, onboarding checklist completion runs low. A 2024 benchmark of 188 SaaS companies put average checklist completion at <strong>19.2%</strong> and the median at <strong>10.1%</strong> (Userpilot, 2024). So when a product team celebrates "most users complete onboarding," they usually mean most users <em>who completed it</em> completed it — survivorship dressed as success.` },
    { t: "p", html: `Activation tells a different, more honest story. A 2024 activation benchmark drawn from 62 B2B companies put the average activation rate at <strong>37.5%</strong> and the median at <strong>37%</strong> (Userpilot, 2024). Roughly four in ten signups actually reach the moment the product was supposed to deliver.` },
    { t: "p", html: `Here's the part that matters. Checklist completion and activation aren't the same measurement, and they don't track each other. A user can tick every box in the setup flow and never experience value. Another can skip the entire checklist, wander straight to the one feature they came for, and activate in ninety seconds. The funnel records the first as a win and the second as a drop. Both readings are wrong.` },

    { t: "h2", id: "blindspot", num: "03 · THE BLIND SPOT", text: "What the funnel can't see" },
    { t: "p", html: `A funnel is a counting machine. It records that step three happened after step two. It can't record <em>how</em> step three felt, or whether the user understood why they were doing it.` },
    { t: "p", html: `In recent B2B SaaS studies — AI-moderated interviews where we watch people meet a product for the first time — the stalls cluster in places the funnel reports as clean. A user reaches the "connect your data" step, the funnel logs the page view, and then nothing for forty seconds. On the recording that silence has a shape: the cursor circling, the quiet "wait, which one is mine," the tab-switch to go find a credential they didn't know they needed. The funnel sees a slightly slow step. The recording sees someone deciding whether this is worth it.` },
    { t: "figure",
      fig: { key: "transcript", props: { time: "00:41", speaker: "New user · data-connection step", children: `Wait — which one is mine? I don't think I have that. Hang on, let me go find it… do I even need this part?` } },
      ref: "FIG 01",
      caption: "A stall the funnel logs as a slightly slow step: the verbal hesitation pairs with a forty-second freeze and a tab-switch to hunt for a credential." },
    { t: "p", html: `The recovery cases are the ones dashboards miss entirely. Plenty of users stall, get briefly lost, then claw their way back to value on their own — not because onboarding helped, but in spite of it. The funnel can't tell a smooth path from a near-miss, because both end at the same logged event. You only learn which one happened by watching the middle.` },
    { t: "p", html: `Then there's the quiet disengagement: the user who completes the motions, says the polite thing, and has already decided not to come back. No rage-click, no error, no support ticket. Just a flat affect and a closed tab. That's the most expensive stall of all, and it's invisible to every metric you have.` },

    { t: "h2", id: "drift", num: "04 · WHY THEY DRIFT", text: "Why “completed” and “felt value” drift apart" },
    { t: "p", html: `There's a structural reason the divergence happens, and it isn't sloppy instrumentation.` },
    { t: "p", html: `A completion event is binary and cheap to fire. Value is graded and slow to arrive. You can finish a setup wizard in two minutes; you may not feel the product earned its place for two days. The research on time-to-value keeps landing on the same point: the faster a new user reaches a first genuine outcome, the more likely they stay, and most early churn happens in the first stretch after signup (Userpilot, 2024). The wizard ends at minute two. The decision to stay gets made later, somewhere the funnel stopped looking.` },
    { t: "p", html: `So a team optimizing for completion can post a rising number while felt value flatlines. They shorten the checklist, auto-skip the hard step, pre-fill the form. Completion climbs. Activation doesn't move, because none of it changed whether the user understood what they'd built or why it mattered. You can manufacture completion. You can't manufacture comprehension.` },
    { t: "pullquote", text: "You can manufacture completion. You can't manufacture comprehension." },

    { t: "h2", id: "evidence", num: "05 · CLOSE THE GAP", text: "Closing the gap with evidence" },
    { t: "p", html: `The fix isn't a better funnel. It's a different kind of evidence sitting next to the funnel.` },
    { t: "p", html: `When the dashboard flags a step where users drop, that's a coordinate, not a diagnosis. It tells you <em>where</em> to point a camera, not what you'll find there. The diagnosis lives in five layers the funnel doesn't carry: the source clip of the actual stall, the transcript moment where the user says what confused them, the behavioral signal of the forty-second freeze, the segment pattern that proves it's the structure and not one tired participant, and an honest confidence indicator about how sure you are.` },
    { t: "figure",
      fig: { key: "insight", props: { accent: true, title: `Name the credential the data-connection step actually wants`, clips: 11, participants: 9, segments: 2, confidence: 0.82, body: `Across new users, the data-connection step stalled because the page never said which credential it expected. The users who recovered did it by leaving to hunt for it; the rest went quiet. A label problem, repeating across segments — not a one-off.` } },
      ref: "EXHIBIT 01",
      caption: "The abstract drop-off as it reads in a NeroView report: clip count, sample, segment spread, and an honest confidence indicator." },
    { t: "p", html: `Run that on an onboarding flow and the abstract drop-off becomes a specific, fixable thing. Not "users stall at step three" but "eleven of the new users hit the data-connection step, couldn't tell which credential the product wanted, and the ones who recovered did it by leaving to hunt for it." One of those statements you can act on. The other is a shrug with a percentage attached.` },
    { t: "p", html: `This is the discipline worth keeping. Treat the activation dashboard as the smoke detector and the evidence trail as the walk-through that tells you what's actually burning. The metric tells you a room is hot. The clip shows you the door the user couldn't find. Fix the door, and the number moves on its own — but you fixed it because you watched it happen, not because you found a cheaper way to log the event.` },
    { t: "p", html: `The activation chart can stay green. Just make sure someone watched what it's counting.` },

    { t: "references", items: [
      { n: 1, html: `Userpilot (2024). "User Activation Rate Benchmark Report 2024" — average activation rate 37.5%, median 37%, across 62 B2B companies. <a href="https://userpilot.com/blog/user-activation-rate-benchmark-report-2024/" target="_blank" rel="noopener">userpilot.com</a>` },
      { n: 2, html: `Userpilot (2024). "Customer Onboarding Checklist Completion Rate: 2024 Benchmark Report" — average checklist completion 19.2%, median 10.1%, across 188 SaaS companies. <a href="https://userpilot.medium.com/customer-onboarding-checklist-completion-rate-2024-benchmark-report-8ebabebefb1f" target="_blank" rel="noopener">userpilot.medium.com</a>` },
    ] },
  ],
  related: [
    { href: "/blog/where-do-i-click.html", title: "“Where do I click?” Five ways onboarding flows lose people", meta: "Field · Onboarding" },
    { href: "/blog/what-counts-evidence.html", title: "What counts as evidence — and what's just a number", meta: "Index · Methodology" },
  ],
};