Frequently Asked Questions

Quick answers to the most common GeoLift questions.

Business Questions

Q: What problem does GeoLift solve?

A: GeoLift measures the true incremental impact of regional marketing campaigns. It answers the critical question: “How much additional revenue/conversions did my campaign actually generate?” by using advanced causal inference to separate campaign effects from natural market fluctuations.

Q: Why is this important for my business?

A:

  • Prove ROI: Obtain statistical confidence in your marketing returns

  • Optimise Budget: Identify which regional campaigns work best

  • Avoid Waste: Stop spending on campaigns that do not drive incremental results

  • Support Decisions: Provide rigorous evidence to leadership for budget allocation

  • Competitive Advantage: Make data‑driven decisions while competitors rely on guesswork

Q: Is this the only measurement tool I need?

A: No, GeoLift is one component of a comprehensive measurement strategy:

  • Best for: Regional campaigns, store rollouts, geo-targeted advertising

  • Complements: Media Mix Modeling (MMM), multi-touch attribution, A/B testing

  • Integrates with: Your existing analytics stack (Google Analytics, Adobe, etc.)

  • Use alongside: Brand studies, customer surveys, and other measurement approaches

Q: What’s the cost and expected ROI?

A:

  • Investment: Proprietary software license + 2-4 weeks implementation

  • Typical ROI: 5-10x return through improved campaign optimization

  • Payback Period: Usually within first campaign optimization cycle (3-6 months)

  • Long-term Value: Compound returns as you optimize multiple campaigns over time

  • Contact: Reach out for specific pricing based on your organization size

Q: How does the process work?

A: Simple 3-step workflow:

  1. Find Fair Comparison (1-2 days): Identify control markets that behave like your test markets

  2. Check Test Strength (1 day): Ensure your experiment can detect meaningful results

  3. Measure the Lift (1-2 days): Calculate actual incremental impact with statistical confidence

Total time: 1 week from data to actionable insights

Getting Started

Q1: What data do I need to run a GeoLift analysis?

A: You need:

  • Time series data: At least 12-24 weeks of pre-campaign data

  • Geographic units: Markets, DMAs, states, or regions

  • Outcome metric: Sales, conversions, or other KPIs

  • Treatment assignment: Which markets received the campaign

Data format: CSV with columns for date, geographic unit, outcome metric, and treatment indicator.

Q2: How long should I run my campaign to get reliable results?

A: Use the power analysis recipe to assess duration and detectable effects:

python recipes/power_calculator_sparsesc.py \
  --config data-config/power_analysis_config.yaml \
  --use-gpu --jobs -1

See outputs/multicell_power_analysis/power_analysis_results.csv for recommended durations and power by effect size.

Generally, 8-16 weeks provides good statistical power for most campaigns.

Q3: How many control markets do I need?

A: Typically 5–15 control markets work well. Use the donor evaluator recipe to select candidates:

python recipes/donor_evaluator.py --config data-config/donor_eval_config.yaml

Outputs are saved in outputs/multicell_donor_eval/ (CSV with scores and donor maps).

Analysis Issues

Q4: My pre-period fit looks poor. What should I do?

A: Try these solutions in order:

  1. Extend pre-period: Add more historical data

  2. Remove outliers: Exclude unusual periods (holidays, events)

  3. Check data quality: Ensure consistent measurement methodology

  4. Adjust donor selection: Use stricter correlation thresholds

# Stricter donor selection
evaluator.evaluate_donors(
    treatment_markets=[502, 503],
    min_correlation=0.8  # Increase from default 0.7
)

Q5: My results show no significant effect. What went wrong?

A: Common causes:

  • Low statistical power: Campaign too short or effect too small

  • Poor control selection: Controls don’t match treatment markets well

  • External factors: Market disruptions during campaign period

  • Data issues: Measurement problems or missing data

Check power analysis first:

if power_results.power < 0.8:
    print("Insufficient statistical power - extend campaign or add markets")

Q6: The effect size seems too large to be believable. Is this normal?

A: Large effects can be real, but verify:

  1. Check data definitions: Ensure consistent measurement

  2. Review external events: Major market changes during campaign

  3. Validate treatment assignment: Confirm which markets got treatment

  4. Run sensitivity analysis: Test with different time periods

# Sensitivity test with shorter post-period
results_sensitive = analyzer.run_analysis(
    treatment_start_date='2023-06-01',
    treatment_end_date='2023-07-31',  # Shorter period
    treatment_markets=[502, 503]
)

Interpretation

Q7: How do I interpret the confidence intervals?

A: Confidence intervals show the range of plausible effect sizes:

  • Narrow intervals: More precise estimates

  • Wide intervals: More uncertainty in the estimate

  • Intervals excluding zero: Statistically significant effects

Example interpretation:

Relative Lift: +15.2% (95% CI: +7.8% to +22.6%)

“We’re 95% confident the true lift is between 7.8% and 22.6%”

Q8: What’s the difference between absolute and relative lift?

A:

  • Absolute lift: Raw units of incremental impact (e.g., +1,000 sales)

  • Relative lift: Percentage increase over baseline (e.g., +15%)

Both are important:

  • Use absolute lift for revenue calculations

  • Use relative lift for comparing across campaigns

Q9: How do I calculate ROI from the results?

A: GeoLift reports statistical effect estimates (e.g. ATT, confidence intervals). ROI calculations depend on your business context (price, margin, cost). Combine effect estimates with your cost and unit economics to derive ROI externally.

Q10: When should I trust the results vs. be skeptical?

A: Trust results when:

  • Good pre-period fit between treatment and synthetic control

  • Statistically significant p-value (< 0.05)

  • Effect size is reasonable for your industry

  • Results are stable across sensitivity tests

Be skeptical when:

  • Poor pre-period fit or large residuals

  • Very large or very small effect sizes

  • High p-values (> 0.10)

  • Results change dramatically with small specification changes

Need More Help?