Campaign measurement Guide
Measure incremental impact when attribution is incomplete
Choose a creator campaign holdout, geographic test or observational comparison, with an experiment worksheet and a worked example of an inconclusive result.
Blog Tag
Research methods
Measuring without fooling yourself: incrementality when attribution is incomplete, confidence intervals on small campaigns, selection bias in repeat partnerships and declared survey methods.
10 articles
Start here
All articles
- Social listeningGuide Separate social listening from campaign monitoring Choose a mention alert or a research question, then connect both workflows for product complaints and campaign launches with a declared conversation sample.
- Social listeningGuide Turn social conversations into product questions Use public comments to frame product research questions. Code a hypothetical comment set, record sampling limits, and write a brief for the next product test.
- Creator discoveryReference Build a creator shortlist from customer interviews Turn consented customer interview notes into anonymized themes, search queries and creator selection rules, with a fictional theme-to-query worksheet.
- Creator discoveryReference Find micro-influencers beyond hashtag lists Find smaller creators through customer mentions, specialist communities, collaborations and referrals. Use a discovery log with duplicate checks and stop rules.
- Campaign measurementReference Measure brand awareness with a declared survey method Plan an influencer awareness survey with comparable samples, neutral questions and declared exposure. Interpret a hypothetical pre/post change with uncertainty.
- Campaign measurementReference Measure repeated creator partnerships without selection bias Compare returning creators with their own starting results, retain the full trial roster, and mark changes in timing, briefs, spend, and attribution.
- Creator outreachReference Personalize creator outreach from evidence you checked Turn a checked creator work sample into a relevant pitch opening. Use a research worksheet and six examples without invented praise or personal guesses.
- Social listeningReference Remove duplicate posts from a listening sample Separate duplicate rows, reposts, quoted commentary and repeat authors with a worked decision table that preserves useful discussion in a listening sample.
- Campaign measurementReference Use confidence intervals without pretending small campaigns are precise Work through a hypothetical 12-of-40 survey result, calculate its Wilson interval, and report sampling uncertainty separately from creator-audience selection bias.
