Service 05
Growth & Experimentation
Testing, analytics, and iteration once the thing is actually live.
In short
Sparza keeps working after launch, when the real data arrives. We instrument what matters, run tests that are actually powered enough to conclude something, and change the design based on what happened rather than what was argued in the review. Usually a retainer rather than a project.
You probably need this if
- You launched something good and have no idea which parts are working.
- Your analytics record everything and answer nothing.
- Tests get called after a week because someone wanted a result.
- The design has not changed since launch because nobody knows what to change.
What’s included
- Instrumentation. Measuring the handful of events that map to real outcomes, rather than everything the tool can capture.
- Analysis. A regular read of what is actually happening, written in plain language.
- Experiments. Hypotheses worth testing, sized so the result can be trusted, and called honestly.
- Design iteration. The changes themselves, designed and shipped, not just recommended.
- Reporting. What we changed, what happened, and what we are doing next.
How the work runs
Instrument
Audit the current measurement, fix what is misreporting, and define the metrics that matter.
Establish a baseline
Read behaviour honestly before changing anything, and build the experiment backlog.
Test and ship
A steady cadence of experiments and design changes, with results reported monthly.
What you get
- Measurement plan and instrumentation
- Prioritised experiment backlog
- Designed and shipped test variants
- Monthly written analysis
- Documented results, including the losses
- Recommendations for the next quarter
Related work
Common questions
Do we need a lot of traffic for this to work?
For classic A/B testing, yes — under-powered tests produce confident nonsense. Below that threshold we focus on qualitative research, funnel instrumentation, and changes justified by reasoning rather than significance.
Will you report a test that failed?
Yes, including the ones we argued for. A programme that only ever reports wins is not measuring anything.
Is this only available as a retainer?
Practically, yes. Iteration needs time to observe, change, and observe again. A three-month minimum is the shortest honest version.
Start here
Every engagement starts with a 45-minute call and a short written proposal covering scope, price, and dates. Tell us what you’re building, or compare plans first.