Start a Project

Case Studies

YESCARTA® (DTC)

Analytics, Heatmaps & UX Optimization

Summary (Problem → Solution → Outcome)

Problem
  • Paid (esp. mobile) traffic bounced; JS errors and rage-clicks disrupted journeys; missing conversions blocked optimization.
Meta:
  • Regions: US
  • Timeline: 12 months
  • Stack: GA4, GTM, Hotjar, Looker Studio (MVP)
Solution
  • A measurement-first reset: GA4 + GTM (consent) MVP with a governed event /parameter schema, plus Hotjar heatmaps /clickmaps/recordings
  • An error taxonomy, all surfaced in a Looker Studio MVP and a weekly Decision One-Pager
Outcome

Paid bounce

↓ 26pp

(from >80% to ~54%)

Mobile CTA reach

↑ 31pp

(after fold/CWV fixes)

Error clusters

↓ 43%

(JS + interaction traps)

Provider-finder starts

↑ 19%

(post ad→landing realign)

Snapshot

Audience

DTC (patients & caregivers).

Scale (12 mo)

~44.5k sessions; avg duration 6:37

Baseline friction

Overall bounce 64%; paid bounce >80%; mobile bounce ~73%; clusters of JS errors + rage-clicks on At a Glance and Clinical Results.

My role

Global governance & reporting lead: authored the KPI layer, rebuilt events/params with consent, stood up the MVP dashboard, and instituted the weekly Decision One-Pager.

What the heatmaps revealed

  • Below-the-fold CTAs on mobile starved key actions; attention pooled around cookie bar/header.
  • Clickmaps showed “Cookie settings” and global nav siphoning clicks on first view.
  • Scroll maps: <50% of mobile users reached critical evidence + conversation prompts.
  • Rage-clicks concentrated on data tabs/accordions (latency + interaction traps).
  • Behavioral split: direct visitors engaged; paid cohorts underperformed → ad→landing mismatch.

Problem → Solution → Outcome

  • Problem
  • Intent mismatch for paid visitors → fast bounces.
  • Fragile mobile templates (CWV + heavy scripts).
  • JS errors/rage-clicks derailed journeys.
  • Missing conversions in GA4 blocked optimization beyond CTR.
  • Solution (Data/BI first, then UX)
GA4 + GTM (consent) — MVP instrumentation
  • Metric dictionary + event schema: learn_view, provider_finder_start, pi_download, conversation_prompt_open, all with consent_state.
  • Standardized UTMs (campaign/source /medium /content / country/brand); stitched platforms→ BI with ±2% reconciliation target.
MVP dashboard (Looker Studio)
  • Portfolio → device → landing drilldowns; heatmap KPIs (scroll reach, click density) surfaced next to outcomes.
Hotjar integration (consent-aware)
  • Loaded only after Analytics consent; keystrokes off; inputs masked; controlled retention.
  • Dual tagging (Hotjar Events and dataLayer.push) kept GA4 and Hotjar in sync.
Paid→landing realignment
  • Mapped queries/creatives to evidence-backed claim blocks with fair balance; deep-linked ads to the section that answers intent first.
Error remediation + CWV
  • Error taxonomy + detectors; simplified header/consent layout; page-level LCP/INP/CLS targets.
Executive cadence
  • Weekly One-Pager (what changed → so what → now what) with scale/pause/reallocate rules; monthly allocation reviews.
  • Outcome (what changed)
  • Decision-ready reporting: single KPI layer; live MVP; weekly adoption by brand.
  • Optimization on qualified actions: conversions captured; paid cohorts tuned beyond CTR.
  • Behavioral lift: error clusters shrank; rage-click tiles cooled; mobile scroll reach to CTAs increased after CWV/layout fixes.
  • Trust in numbers: platform↔BI variance held near ±2%; actions & owners logged weekly.

What we built

  • Consent-aware event spine (step + action events tied to page modules).
  • Heatmap gallery (mobile vs desktop) to validate fold placement and content sequencing.
  • Error monitors with alerts and release notes.
  • One-Pager automation feeding highlights, risks, and next moves.

Methods

  • Cohort trees: device × channel × landing × error-state to separate content vs technical drop-offs.
  • Attention mapping: scroll-reach percentiles + click density → re-order content above the fold.
  • Attribution hygiene: governed UTMs; stable keys for MMM/MTA.
  • Decision logging: hypothesis → action → delta captured and rolled into playbooks.

KPI Layer

  • (excerpt)
  • Acquisition: non-brand share, paid intent-match score, landing readiness score.
  • Engagement: engaged sessions, clinical-content depth, PI view rate.
  • Action: provider-finder starts, PI/MG downloads, talk-to-doctor prompts.
  • Quality: JS error rate by template, rage-click rate, consent coverage, mobile CWV

Stack & Governance

  • Data/BI: GA4 + GTM (consent), Looker Studio (MVP), warehouse-ready exports (daily).
  • Behavior analytics: Hotjar heatmaps, clickmaps, scroll maps, recordings (consent-gated).
  • Ad platforms: Google Ads/YouTube (via CM360/SA360), Meta, LinkedIn.
  • Governance: KPI dictionary, GA4 schema, UTM standard, QA monitors, Decision One-Pager.

DTC Budget Impact & Allocation

  • — What the GA4/Hotjar + Looker data drove
  • Cut waste; fund intent. We throttled high-bounce paid placements and reallocated weekly (from Looker Studio real-time + GA4 step events) to deep-link ads into the best-performing modules—evidence sections, support resources, and provider-finder—optimizing to cost per meaningful action, not CTR.
  • Prioritize trusted health inventory. We shifted dollars toward evidence-led placements (e.g., Open Evidence content destinations) and endemic healthcare networks (e.g., Medscape/WebMD patient units) where intent and education depth were highest, with tight frequency caps and allowlisted contexts.
  • Fix mobile first, then scale. After resolving CWV/interaction issues and reducing the “cookie-bar tax,” we expanded YouTube/CTV and social—so multimillion-dollar reach landed on pages that actually convert.
  • Governed moves, every week. A standing Decision One-Pager set thresholds for scale / pause / reallocate, with ±2% platform↔BI reconciliation, ≤5% variance guardrails, and a 10–15% test tranche—letting leaders pivot big budgets with confidence.

Deliverables you can request

KPI Layer & Metric Dictionary

(Confluence + YAML/JSON)

GA4 Event/Parameter Spec

(with consent)

MVP Dashboard

(exec + ops)

Hotjar Pack

(heatmap gallery, rage-click clusters, tags, survey template)

Paid→Landing Map

(intent→section routing with fair balance)

Error Monitor Pack

(freshness, completeness, schema drift)

Decision One-Pager

templates (auto-filled)

Operating cadence

Weekly
One-Pager (15 min), under/over-delivery, reallocation calls.
Monthly

Allocation review (trends, experiments).

Quarterly

QBR (outcomes vs KPIs, roadmap & risks).

Compliance notice: Public figures/visuals are anonymized/redacted and shared for illustrative purposes in alignment with NDAs, MLR/OPDP/APLB, and HIPAA/GDPR.

Want this level of rigor on your DTC brand?

Book a 30-min Analytics Diagnostic—I’ll map your KPI layer, instrument conversions, deploy consent-aware Hotjar, and stand up the MVP dashboard + one-pager.

Book a fit call
Back to Case Index

BIKTARVY® — Global DTC • Analytics & Funnels

Calquence — HCP + DTC • Analytics & Orchestration

FAQs

Can we run branded on LinkedIn?

Possible, but we default to unbranded + route branded to HCP-auth portals. We follow platform policy and OPDP patterns.

18+ gating, no personal-attribute inferences, fair-balance/disclaimers, strict comment filters, PV/AE triage in ≤24 hours.

Qualified events, attention & completion, and outcome cohorts—rolled up in the Decision One-Pager; MMM/MTA exports provided.