Service field guide
Find friction after launch and decide which observations justify an experiment.
Use after observing actual customer journeys. On a new low-traffic site, begin with task testing and obvious defects; do not promise a conclusive experiment before checking sample requirements.
Funnel definitions, traffic sources, event quality, user observations and development capacity.
Ask how hypotheses are prioritized, which primary outcome is fixed before testing and how confounding changes are controlled.
Included because the cited agency sources describe relevant capabilities. This is not a service performance ranking; confirm the scope for your particular market.
Assess for a launch followed by work on acquisition and on-site conversion together.
Consider for an evidence-led market test where search and paid-media findings must meet.
Assess when paid acquisition and landing-page testing belong in the same workstream.
Assess when a launch needs a shared evidence base for search, media and site decisions.
Review when a team needs to distinguish AI-search experiments from established acquisition work.
Use these formats to request a comparable proposal. They are editorial brief structures, not quoted products or prices.
The number of markets and languages, existing asset quality, integrations, production volume, review capacity and handover depth. Request separate lines for agency work, media and external fees.
Field note
A multilingual campaign raises questions about mobile forms and measurement.
Field note
An airline launch connects strategy, acquisition and responsibility.
Use observed task failures and interviews to prioritize improvements, and label their evidence accordingly. Do not rename those observations a statistically proven uplift.
The review separates defects, observational findings and experiment conclusions; inconclusive results remain inconclusive.
Field note
Agree what decisions the first observations can support.