Skip to content
home

Conversion optimisation · 2 min read

Optimise the site or buy more traffic? Deciding with evidence

A framework for comparing the two, and the measurement each option needs.

By WebIQ Analytics · 12 February 2026

When growth stalls, the choice often comes down to two options: spend more on media to bring in more visitors, or improve the site so existing visitors convert better. Both can work. Comparing them properly needs the same thing for each: an estimate of the incremental effect, with a range.

The case for more traffic

Media scales quickly and predictably, up to a point. Every channel saturates: the next pound brings back less than the last. A mix model's response curves show how much headroom each channel has at today's spend.

The case for conversion

Improvements to the site apply to every future visitor, paid or not, and keep paying back without ongoing spend. But they take time to find, and many tests fail. The realistic gain is the size of improvements you can find, times how often you find them.

Illustrative: comparing two plans by their likely outcome and range.Compare scenariosScenarioSpendExpected revenueDifference from baseChance it beats baseReturn per £1Current planBase£2.4m£23.3mlikely £21.8m–£24.7m––£2.31Shift to search£2.4m£24.1mlikely £22.6m–£25.5m3.4% (1.1–5.9%)96%£2.66Cut TV 30%£2.1m£22.4mlikely £20.9m–£23.8m-3.9% (-6.8–-1.2%)2%£2.27Christmas push£2.9m£25.2mlikely £23.4m–£26.9m8.2% (4.0–12.6%)99%£2.18
Comparing two plans by their likely outcome and range.

A simple comparison

  1. Media option: from the response curves, the expected extra revenue for an extra £X, with its range.
  2. Conversion option: for the same £X spent on research, design, build and testing, the expected improvement, based on your own test history and the traffic reaching the pages you'd change.
  3. Compare like with like: over the same period, accounting for how long each takes to deliver.

What usually decides it

  • Channels near saturation favour conversion work.
  • Low traffic to key pages limits what testing can find, favouring media or bigger redesigns.
  • A poor-performing step with lots of traffic (a checkout, a lead form) is often the best opportunity for either.

Don't treat them as exclusive

The best plans usually do both: hold media at efficient levels and run a steady testing programme on high-traffic steps. Revisit the split each quarter as the response curves and test results update.

Experiments that tell you the truth about your site

We run research-led testing programmes on your site and funnels, designed to detect real effects and read out without wishful thinking.