01
Start with a customer problem
Do not begin by collecting every possible data point. Begin with a specific problem such as helping a returning customer find a previous category or preserving a delivery preference.
Write down the benefit, the minimum data required and how success will be measured.
- Recently viewed products
- Saved cart with a clear expiry
- Relevant category recommendations
- Remembered non-sensitive preferences
02
Collect less data, more deliberately
South African businesses processing personal information need to consider POPIA and their responsibilities as a responsible party. Privacy notices should explain what is collected and why.
Separate service communications from direct marketing. Give people practical controls rather than relying on vague consent wording.
- Document the purpose
- Limit retention
- Protect access
- Provide an opt-out where required
03
Keep the default experience strong
A first-time or anonymous visitor must still be able to browse, understand delivery and returns, and complete checkout. Personalisation should enhance the journey rather than repair a weak base experience.
Avoid hiding price, stock, shipping or return information behind a profile.
- Make search and navigation useful
- Show total costs early
- Keep checkout focused
- Explain recommendation logic when helpful
04
Measure value and harm
Track whether a feature improves completion, repeat use or product discovery. Also monitor incorrect recommendations, support complaints and opt-outs.
Run a controlled test and stop a feature that adds friction or erodes trust, even if it increases one short-term metric.
- Define a primary metric
- Set guardrail metrics
- Review different device sizes
- Check accessibility and performance
Sources
Primary references
Technical and regulatory points were checked against the following primary or official references.
This guide was prepared with AI-assisted research and drafting, then checked against the cited sources and reviewed for unsupported claims before publication.