Commission grounded conversion optimisation
The goal is to improve ordering, enquiries, bookings or registration. More clicks alone are insufficient. Specify actions, pages, assessment and implementation. A test may end without demonstrated improvement.
Define actions and quality
Describe the journey from entry to completion. Enquiries can be unsuitable and orders returned. Quality matters alongside counts.
Specify denominator and period: sessions, people or started forms yield different rates. Compare consistent data and limitations.
| Goal | Check quality too |
|---|---|
| Order | Payment, cancellation and returns |
| Enquiry | Validity and relevance |
| Booking | Confirmation and attendance |
| Registration | Valid account and intended use |
Check measurement
Check events, duplicates, redirects and devices. Changed recording may simulate improvement. Specify available data, missing data or problems.
Tools measure defined actions, not their business value automatically. Suitable access and bounded analysis suffice initially. No customer files, credentials or complete recordings in the enquiry.
Assess obstacles
Examine offers, expectations, navigation, mobile use, forms and errors. Where must people guess, repeat or correct without explanation? Check keyboard operation and accessible labels.
Qualitative assessment delivers problems and proposals, not a percentage uplift alone. Specify journeys, devices, languages and documented findings.
Derive specific changes
Hypotheses connect changes, expected benefits and measurable observations. Separate different interventions for interpretation. Colour or copy changes without identified problems are not a complete plan.
Specify priorities, effort and risks. More forms with less suitable leads may be undesirable. Errors, speed and returns can be useful guardrails.
Choose an appropriate method
Comparison tests need a question, variant allocation, reliable measurement and sufficient use. Have scale planned; no universal visitor minimum or duration applies.
With little or no use, assessment, user tests or justified improvements may come first. Unclear results should remain unclear. Account for concurrent campaigns and changes.
Separate assessment, delivery and testing
Compare pages, journeys, languages and tasks. Research, design, development, measurement and analysis may have different owners. Concepts do not automatically include implemented variants.
Separate initial work, care and tools. Agree checks and decisions on further tests. Extra collection or infrastructure needs a purpose and controlled costs.
| Item | Question |
|---|---|
| Assessment | Which journeys and findings? |
| Design | Which proposals and approval? |
| Development | Which changes and variants? |
| Measurement | Which events and corrections? |
| Testing / care | Which method, analysis and tools? |
Define work and acceptance
Describe goals, context, pages, languages and measurement. Actual approximate use helps testing scope; zero use for a new offer is possible.
Agree diagnosis, priorities, checks and acceptance. Assign approval, publication and issue response. Assess delivery and sound results without guaranteed uplift.
Frequently asked questions
What is a conversion?
The useful action defined for the work. Clicks, enquiries, valid orders and qualified contacts differ.
Is A/B testing always required?
No. Method and data must suit; assessment may come first.
Can tests be prepared without visitors?
Yes. Actual measurement needs suitable use and planning.
Is a higher rate always better?
Not without checking quality, denominator, period and adverse effects.
Can improvement be guaranteed?
No. Work and methods can be agreed, not a particular effect.
What should a quote include?
Assessment, journeys, design, implementation, measurement, tests, evaluation and care costs.
