#analytics
21 pieces filed under “analytics”.
First-party data without the surveillance
First-party data foundations for the cookieless era: consent-mode analytics, an honest take on server-side tagging, what to stop measuring, and labelling estimates.
Analytics for AI features: what to measure
Chat volume tells you nothing. The instrumentation stack we use for AI features: acceptance, correction signals, task completion, cost-quality overlays, weekly review.
Review your content like a portfolio
A quarterly content benchmarking ritual: sort every article into winners, flatliners and sleepers, then give each piece one of three decisions — prune, refresh or expand.
Share of search: a brand metric a CFO will accept
Brand health is usually measured with slow, soft surveys. Share of search turns branded query volume into a leading indicator finance teams actually trust.
Writing an analytics implementation plan before you tag anything
A measurement plan decides what to track; an implementation plan decides how. Naming, property hygiene, consent mode, staging QA and the 30-day calibration.
Migrating a site without burning its rankings
Domain changes, CMS switches and relaunches are risk decisions, not just checklists. How to forecast the dip, set expectations and know when to roll back.
Analytics pipelines you can actually trust
Dashboards nobody trusts get quietly ignored. Event schemas, validation at the edge, QA environments for data, and pruning metrics nobody has read in two quarters.
Write the measurement plan before you build
Analytics bolted on after the build answers yesterday's questions badly. The measurement plan: events, naming, consent, dashboards — before the first sprint.
Checkout friction: where conversions actually die and how to find out
Checkout abandonment is diagnosed with funnels, error logs and session data — not vibes. Where conversions actually die, how to measure each death, and what to fix first.
Funnel metrics: measure the movement, not the moment
Snapshots lie; cohorts and velocity tell the truth. How to define funnel stages across teams, pick the metrics that find the constraint, and run the weekly review.
Setting goals for a redesign that aren't vibes
How to turn 'make it feel modern' into measurable redesign goals: primary metrics, kill criteria, baseline capture, and reporting that survives launch fog.
Attribution is 90% noise: making decisions anyway
Last-click lies, multi-touch dreams, and what actually works for teams without a data science department: a decision framework for acting on noisy attribution.
Name your events before you buy the analytics tool
Event naming decides whether your data answers questions in year three. The taxonomy anatomy we design before any tool evaluation: verbs, properties, screens, migration.
Measuring brand without voodoo
Brand measurement for people allergic to brand science: aided recall, share of search, consistency audits and the quarterly one-pager leadership actually reads.
Your search box is a research department
Nobody lies to a search box. Mining internal site search — zero-results reviews, synonym dictionaries, query taxonomy — into a content roadmap that compounds.
Campaign naming conventions: boring is the feature
UTM and campaign names are infrastructure, not admin. A grammar that survives staff turnover, separators, tooling that enforces it, and cleaning old messes.
After launch: the first 90 days done properly
Launch day is the start, not the finish. How we run the first 90 days after a site ships: analytics baselines, bug etiquette, and the next engagement.
Analytics governance: tracking plans before tools
Analytics rots quietly until nobody trusts the dashboards. The fix is governance, not tooling: event naming grammar, ownership, QA rituals and the monthly data hour.
Activation metrics that mean something
Most activation metrics are flattering fictions. Find the behaviour that predicts retention, measure time-to-value honestly, defend it in the boardroom.
Attribution that admits what it doesn't know
Attribution is always lying a little. Model taxonomy, dark social acceptance, self-reported fields, geo holdouts and decision-making without false precision.
Dashboard design: answer first, chart second
Most dashboards are chart galleries that answer nothing. The question-first method: decision inventories, alert thresholds, and the five-element view that wins.