Joshua Franklin Wiedeman

Onsite Personalization

Onsite personalization changes what a visitor sees based on who they are and what they have done. A first-time visitor sees a different homepage than a returning customer. A visitor from a Facebook ad sees different copy than one from search. Someone who browsed running shoes sees running shoes, not hiking boots.

How it works

Most sites show the same page to everyone, whatever they came for. Personalization uses signals you already have (traffic source, location, browsing history, purchase history) to show a more relevant version. Even simple rules, such as different headlines for new and returning visitors, can help. Each rule still needs to be tested.

How to do it

  1. List the signals you have: new or returning, traffic source (UTM), location, device, pages viewed, and purchase history.
  2. Pick a content change for each segment. For example:
    • New visitors: reviews and an explainer.
    • Returning non-buyers: recently viewed items.
    • Past customers: related products.
    • Local visitors: local shipping times.
  3. Use a personalization tool that swaps content blocks on existing pages (for example Dynamic Yield, Optimizely, VWO, Mutiny, Nosto).
  4. Start with simple rules: headline by traffic source, recently viewed items, local shipping info.
  5. Keep a control group for each rule. A/B test the personalized version against the default.

What to measure

  • Personalized vs. default conversion rate: Conversion rate per segment for the personalized version versus the control.
  • Revenue per visitor: Total revenue divided by visitors, split by test and control.
  • Segment coverage: Share of total traffic that receives a personalized experience.

Best practices

  • Start with your three or four largest segments.
  • Keep a strong default page for visitors who match no rule.
  • Add rules one at a time and confirm each one helps before adding the next.

Common pitfalls

  • Getting too personal. “We noticed you looked at anxiety medication last Thursday” feels like surveillance. Personalize on behavior patterns, not sensitive details.
  • Building rules on assumptions. Untested personalization can lower conversion.
  • Creating more segments than you can maintain. A few well-built segments beat many weak ones.