Why this matters
Data visuals are useful only when they lead to action. As a Data Visualization Engineer, you present insights to product managers, marketers, and operations leaders. Clear, prioritized recommendations with concrete next steps turn charts into decisions, pilots, and measurable results.
- Turn a dashboard finding into an experiment plan
- Prioritize competing ideas with a simple impact/effort score
- Communicate trade-offs and what to do next, by whom, and by when
Concept explained simply
A recommendation is your best, practical answer to: “Given this data, what should we do next?” Next steps are the small, concrete actions that move the recommendation forward.
Mental model: Signal → Option → Impact → Feasibility → Decision → Next steps
- Signal: The insight (what changed, where is the gap)
- Option: 2–3 actionable ideas that could address it
- Impact: Expected metric movement and who benefits
- Feasibility: Effort, cost, risks, and dependencies
- Decision: Pick the best option with a clear rationale
- Next steps: Owners, timeline, and how success will be measured
A simple framework: the Actionable Recommendation Canvas
- Problem: One sentence describing the gap or opportunity
- Recommendation: The action you propose (use strong verbs)
- Why it works: 2–3 data-backed bullets
- Expected impact: Metric, magnitude, and direction
- Confidence: High/Medium/Low and why
- Risks & assumptions: What could derail this
- Owner: Role or team
- Timeline: Now / Next / Later or a specific date window
- Next steps: 3–5 concrete actions
- Metrics to monitor: Leading and lagging
- Decision needed: From whom and by when
Prioritization: pick what to do first
When you have multiple options, score quickly and transparently.
RICE (Reach, Impact, Confidence, Effort)
RICE score = (Reach Ă— Impact Ă— Confidence) / Effort.
- Reach: number of users/sessions affected in the period
- Impact: relative size of change per user (e.g., 0.25=small, 1=big)
- Confidence: 0–1 based on data quality and precedent
- Effort: person-weeks or story points
ICE (Impact, Confidence, Ease)
ICE is faster: (Impact Ă— Confidence Ă— Ease). Use when rough ordering is enough.
Now / Next / Later board
Classify by urgency and effort. If it takes <1 week and unblocks others, it’s a Now.
Crafting clear next steps
Use unambiguous, time-bound actions.
- Examples of strong verbs: Align, Decide, Instrument, Ship, Launch, Pilot, Rollback, Re-run, Document
- SMART check: Specific, Measurable, Achievable, Relevant, Time-bound
One-slide template (fill in)
- Recommendation: [action + scope]
- Why: [top 2 data points] → [target metric + expected lift]
- Plan: [Now/Next/Later] with owners
- Risks/Assumptions: [top 2]
- Decision: [ask] by [date]
Worked examples
1) Product funnel drop-off
Signal: 38% drop at payment step after adding 2FA; error rate up 3.2%.
Recommendation: Ship a two-click 2FA fallback (SMS) for failed OTP retries.
- Why it works: 42% of failures are timeouts; SMS completion 2Ă— higher in similar flows
- Expected impact: +3–5% conversion to paid; Confidence: Medium
- Owner: Checkout team; Timeline: Now (1 sprint)
- Next steps: Instrument retry reasons; A/B test fallback; Monitor error rate and approval rate
2) Marketing CPA creeping up
Signal: CPA +22% month-over-month; search bids auto-increased; view-throughs flat.
Recommendation: Cap bids on 5 expensive keywords and shift 20% budget to high-ROAS creatives.
- Why it works: Top 5 keywords account for 63% of spend, 28% of conversions
- Expected impact: CPA -10–15%; Confidence: Medium
- Owner: Paid acquisition; Timeline: Now (this week)
- Next steps: Pause two lowest-ROAS keywords; Launch two new creatives; Review after 7 days
3) Support backlog
Signal: Median first response time increased from 3h to 9h; 26% tickets are password resets.
Recommendation: Add self-serve reset flow and auto-close resolved reset tickets after 48h.
- Why it works: Removes 26% load; benchmark shows 60–80% self-serve success
- Expected impact: Median FRT back to <4h; Confidence: High
- Owner: CX tools; Timeline: Next (2 sprints)
- Next steps: Spec reset flow; Instrument success events; Pilot with 20% traffic
Communicating up and across
- Lead with the ask: Put the recommendation and expected impact first.
- Use numbers sparingly but precisely: one metric to win the room.
- Offer an option B to show you considered trade-offs.
Email/brief template
- Subject: Recommendation to [verb + outcome] — needs decision by [date]
- Top line: We recommend [action]. Expected [metric] change: [value].
- Evidence: [2 bullets]
- Plan: [Now/Next/Later] with owners
- Risk & mitigation: [1–2 bullets]
- Decision: Request approval / budget / alignment
Handling uncertainty
- State confidence and why: data quality, sample size, precedent
- Use pilots: start small, set guardrails (e.g., max CPA, error budget)
- Define stop conditions: what would make you pivot or rollback
Common mistakes and how to self-check
- Vague verbs: Replace “optimize” with “reduce time-to-first-byte to <200ms.”
- No owner: Assign a role or team, not “someone.”
- No metric: Name one leading and one lagging metric.
- Over-precision: Ranges are fine; avoid false certainty.
- Ignoring effort: Use RICE/ICE to compare ideas fairly.
Self-check checklist
- One-sentence problem and one specific recommendation
- Expected impact with a metric and direction
- Confidence level with a reason
- Owner and timeline
- 3–5 next steps, each starting with a strong verb
Exercises
Exercise 1 — Prioritize and recommend: checkout drop-offs
Scenario: 100k weekly sessions reach cart; 60k reach payment; 37k complete. Option A (simplify address form) Reach=30k, Impact=0.2, Confidence=0.7, Effort=2. Option B (2FA fallback) Reach=60k, Impact=0.25, Confidence=0.6, Effort=3. Option C (free shipping banner) Reach=100k, Impact=0.1, Confidence=0.8, Effort=1.
Tasks:
- Compute RICE for A, B, C
- Pick one recommendation
- Write 3 concrete next steps, 1 owner, 1 success metric
Hints
- RICE = (Reach Ă— Impact Ă— Confidence) / Effort
- Pick the highest RICE, but check risks/assumptions
- Use strong verbs: Instrument, Launch, Decide
Exercise 2 — One-slide executive summary
Scenario: Email unsubscribe rate rose from 0.2% to 0.5% after a template redesign. CTR unchanged. Draft a 6-line slide using the template.
Hints
- Lead with the recommendation
- Call out expected impact range, not a single exact number
- Include a small pilot or A/B as the first step
Exercise checklist (use for both)
- Problem and recommendation are one line each
- Impact range and confidence are stated
- Owner and timeline are named
- 3–5 next steps are action verbs
Practical projects
- Dashboard to action: Pick one dashboard, choose 3 insights, and ship a one-page brief with a recommendation, RICE table, and Now/Next/Later plan.
- Pilot playbook: Create a template for pilots (scope, guardrails, success metric, stop conditions). Use it on a small experiment.
- Quarterly prioritization: Facilitate a 60-minute session to rank 5–8 initiatives using RICE and produce a single-page roadmap.
Learning path
- Before: Interpreting charts, identifying signals, basic statistics
- Now: Turning insights into decisions with clear next steps
- Next: Experiment design, causal inference basics, stakeholder communication
Who this is for
- Data Visualization Engineers who present insights and need stakeholders to act
- BI/Analytics professionals translating dashboards into roadmaps
- Anyone asked “So what do we do next?” after a chart
Prerequisites
- Comfort reading core product/marketing metrics
- Basic estimation (ranges, not exact forecasts)
- Familiarity with A/B tests or pilots
Next steps
- Apply the canvas to one current insight and share with the team
- Run the quick test below to check your readiness
- Note: The quick test is available to everyone; only logged-in users will have their progress saved
Mini challenge
Pick one live metric that moved last week. In 10 minutes, draft a one-sentence recommendation, a 3-bullet Next steps list, and a single success metric. Share it for feedback.