Modeling redesign / usability conversion rates

what real world figures have you had?

[webdevsf]

I'm in the process of trying to justify some redesign changes to my website. In order to do this, I have to make some projections on how much conversion rate will be increased by doing a redesign.

IMHO, there are several areas that could be modeled.

1. Increasing product selection
2. Usability enhancements (improving page flows, UI etc)
3. Graphic redesign enhancements

Right now, I'm focusing on #3. I think #2 is the easiest to measure (so not bothering in this message), and #1 is something that I am not focusing on because of the nature of my business.

So how to model graphic redesign conversion rates? I am not trying to get an exact number. I know "it varies a lot" - so if that's your comment than you can save it. :)

But those of you who are graphic designers should have these figures, in order to justify your services to new clients. I am looking for range here, based upon what kind of redesign.

A) Major site overhaul - new colors, logos, fonts, layout, messaging, from "home grown" web layout to very professional.

B) Minor site changes - updated colors, logo, minor layout changes.

My guess is that A) should increase conversion rate anywhere from 50%-200%, whereas B) would increase it from 10% to 50%. If not, it would be hard to justify the costs associated with doing a redesign.

Anyone (especially web designers and product managers) have any real world experience here on sites that they've worked on and how the redesign affected conversion rate?

Much thanks,
WDSF

[TallTroll]

I know you don't want to hear this, but... "it varies a lot". For some sites, the look and feel is critical to shepherding the visitor through to the "success" (sale/subscription/whatever) page. For others, it is irrelevant.

Given that fact, it becomes impossible to hang a general number on any given change. It depends far too much on the demographics of the sites visitors.

For example, imagine that a car manufacturer is introducing a new model. If they have the savvy, there will be 2 sites for promotion, one to promote to men, and one to women. The one for men will have a Flash front end, be loaded with pictures of the car driving through rubbish, suspicious piles of cardboard boxes etc, and have strong, blocky colours (men look for edges, and geometric shapes), and have technical specs everywhere

The one for women will be pastelly, rounded and will emphasise the benefits of ownership.

In each case the design is very different, and targets a different group. Targetting for say, kids of primary shcool age requires different design values again.

The University of Wichita [psychology.wichita.edu] publishes extensive papers on usability in software etc, and would prove an excellent research resource for the specifics of your situation