marketing growtha b testingcroexperimentationfunnel analyticsvwo optimizely
Complete evaluation framework
What to assess and how to score it
Review the evidence signals before interviewing. Then use the anchored descriptions—not instinct alone—to choose the score that best matches each answer.
01
Evaluation factor
Campaigns that performed
30% weight
Check which winning tests they personally shipped: page, hypothesis, variant treatment, and lift in conversion rate, revenue per visitor, or AOV, plus how long the win held.
Evidence to listen for
Names campaigns or programmes they ran, with the metric they moved and the baseline it moved from
Distinguishes their own work from the team's or the agency's
Knows the audience and the offer, not just the channel
Can describe a campaign that failed and why
Five-point scoring guide
1
Poor
No numbers; describes activity rather than results.
2
Needs Improvement
Vanity metrics only; attribution to their own work is unclear.
3
Satisfactory
Real campaigns with some numbers; baselines often missing.
4
Very Good
Named campaigns with metric, baseline, and clear personal ownership.
5
Excellent
Names specific tests on checkout, PDP, or signup flows with baseline, lift, confidence level, and post-launch holdout validation.
02
Evaluation factor
Audience and segmentation
25% weight
Probe how they segment traffic before testing: new versus returning, device, paid landing versus organic, and whether they analysed results by segment rather than aggregate only.
Evidence to listen for
Describes real customer segments and what differs between them, not demographic guesses
Has built lifecycle stages, journeys, or nurture logic that reflect actual behaviour
Knows where the audience data comes from and its limits
Writes for the segment rather than for everyone
Five-point scoring guide
1
Poor
One message for everyone; no segmentation thinking.
2
Needs Improvement
Segments by demographics only; no behavioural insight.
3
Satisfactory
Workable segmentation; lifecycle logic is basic.
4
Very Good
Behaviour-driven segmentation and lifecycle design with evidence behind it.
5
Excellent
Describes segment level effects, for instance mobile lift masking desktop loss, and adjusts targeting or rollout accordingly.
03
Evaluation factor
Measurement and testing
30% weight
Test statistical literacy: sample size and MDE calculation, peeking, sequential testing, Bayesian versus frequentist tooling in VWO, Optimizely, or AB Tasty, and GA4 funnel instrumentation.
Evidence to listen for
Tests deliberately rather than changing everything at once
Understands attribution limits and does not overclaim
Knows what sample size and duration a test needed
Can name a test whose result contradicted what they expected
Five-point scoring guide
1
Poor
No testing; claims credit for correlation.
2
Needs Improvement
Runs tests but reads them badly; overclaims attribution.
3
Satisfactory
Tests sensibly; rigour drops under deadline pressure.
4
Very Good
Disciplined testing with honest attribution and a result that surprised them.
5
Excellent
Calculates required sample before launch, refuses to call underpowered tests, and explains false positive risk in plain numbers.
04
Evaluation factor
Working with the business
15% weight
Assess how they source hypotheses and get them built: session replay in Hotjar or FullStory, support tickets, user testing, plus negotiating dev and brand sign-off on variants.
Evidence to listen for
Works with sales, product, or operations rather than throwing campaigns over a wall
Briefs designers and writers well enough to get usable work
Manages agencies or freelancers to a standard
Reports results to leadership without spin
Five-point scoring guide
1
Poor
Works in isolation; reports only flattering numbers.
2
Needs Improvement
Limited coordination; briefs are thin.
3
Satisfactory
Coordinates adequately; reporting is honest if basic.
4
Very Good
Strong partner to sales and product; briefs and reports clearly.
5
Excellent
Runs a prioritised research-backed backlog (PIE or ICE) and shows how losing tests changed product or copy decisions.
Put this rubric to work
Score every candidate against the same standard
Add these weighted factors to Hirevire and let AI evaluate recorded answers against your rubric.
