New 99designs logo vs original logo: -10% trial conversion
Commentary byLiftArchiveOur analysis, not theirsReplacing the original Bingo Card Creator logo with an off-the-shelf logo from 99designs dropped sitewide conversion to free trial from 11.05% to 9.92%, about 10% worse.
Patrick McKenzie ran this sitewide test and published it on the Kalzumeus blog in 2010. About 8,500 people saw each logo over two weeks, and the original won at 99% confidence.
- Conversion to free trial was 11.05% with the original logo and 9.92% with the new one, about 10% lower in relative terms.
- Significant at 99% confidence, with about 8,500 people per version and bots filtered out.
- The test measured trial sign-ups, not purchases, so the effect on revenue is not known.
The change
Only the logo in the header changes, on every page.
Simplified mockup drawn by LiftArchive from the shared screenshots.
Control
The original, custom-made Bingo Card Creator logo in the site header.
Variant
A new logo bought from the 99designs store, shown sitewide. Each visitor was randomly assigned one logo and kept seeing it on later visits.
Why it likely lost
Our reading of the result, not the practitioner's claim.
- The original logo was less generic, which may have made the product look more distinct.
- A logo is one of the first things visitors see, so it can shape trust in the whole site.
- The new logo may have fit the product's audience or look less well than the one made for it.
Caveats: The conversion rate is sitewide and includes a lot of search traffic that could use free content without signing up. The test measured trials, not purchases, and the data is from 2010.
Who ran it
Patrick McKenzie ran this test and published the results (Kalzumeus (patio11) blog). The numbers are theirs; the summary and commentary on this page are ours.
Read the original case study on Patrick McKenzie →
How to run this test on your site
- Hypothesis
- If we replace our logo with a new design, then trial sign-ups will change because the logo shapes first impressions of the product.
- Primary metric
- Free trial sign-up rate, sitewide
- Guardrails
- Paid conversion from trial
- Bounce rate on landing pages
- Returning visitor behavior
- Who should test it
- Small, transaction-focused SaaS products where brand confusion during a short test is a small risk. Larger brands with strong recognition may decide the risk of showing two logos outweighs what they would learn.
Questions
Does a logo affect conversion rate?
It can. In this test, a new logo cut trial sign-ups by about 10% at 99% confidence.
The effect will differ by product and audience.
How do you A/B test a logo?
Randomly assign each visitor one logo and show it on every page and every visit.
Filter out bots and measure a conversion that happens often enough to reach significance, such as trial sign-ups.
Is a cheap stock logo good enough for a SaaS product?
Sometimes, but not always. Here, a custom logo beat a ready-made one, so it is worth testing before you switch.