it depends on the type of project, on an ecommerce for example you can measure the UX improvements by measuring the conversion.

Other indicators:

• time spent on pages
• number of actions (comments, ratings etc)
• time for specific action
• bounce rate
• traffic

and so on

LATER EDIT: and than again, there's always the user's feedback. It's subjective but if you have enough you can draw some conclusions from it.

I think there's a simple acid test:

• Does the client come back for more work?

This applies to internal consultants just as much as to external consultants. Repeat business is the best measure of quality, because it's the ultimate indicator that you're adding value in the client's eyes.

Although measures like ROI are correct in theory, in practice they miss other important elements of the engagement: such as how you gel with the design team, how clear your recommendations are and how your proposed solutions fit with the the client's culture.

If you're willing to read a book, then you will find "Measuring The User Experience" very useful. I've read it and I highly recommend it.

Return on investment (ROI) is one way to measure quality, because the business drivers for people who pay for UX consultants or staff include financial ones.

UX often has a massive return on investment, but UX employees rarely publish ROI results, for a variety of reasons.

I recently heard about an ideal project that demonstrates the value of UX. Here are the generic project details:

• The users take calls, all day, from customers who want to purchase or modify a service from the company. The old software used to set up the service had lots of excise [à la Cooper-Reimann in About Face 3]. The service has many options, so the users may be faced with many choices in the user interface as they respond to each customer's requests.
• The project involved reskinning existing software written decades ago, not designing new software, and not adding features. This allowed ceteris parabis before-and-after time-on-task comparisons. (Really, you couldn't have planned a better experiment to measure the effect of UX changes.)
• The code between the GUI and the legacy software added some error trapping, so the new GUI guides users in making fewer errors. This allowed ceteris parabis before-and-after comparisons of error rates. (Again, you couldn't have planned a better experiment to measure the effect of UX and usability work.)
• The software users were all internal, which meant all data to calculate ROI was available, including deveopment cost, user-training cost, operating costs before and after, as well as time-on-task and error-rate data. The number of users (a large number) was also known.
• A company employee told me that time on task dropped from 20 minutes to 5 minutes and that error rates dropped by about two thirds.

Since the company knows what its users (staff) are paid and how many transactions are taking place, it can calculate the decrease in daily operating cost. The company can also add up how much the project cost to design and develop. From there it's simple math to the break-even point (counted in in days, not months, in this case). The rest of the savings is the ROI.

Don't you wish we knew the exact amounts involved?

-=- Jerome

Check this article, it's quite interesting: Impact of Hci activities

It really depends on your site and business model. For example, on a research platform time spent on pages should be high for some pages (reading articles) and low for other pages (search forms - effectiveness of the algorithm to help users find what they want quickly).

One way I've been experimenting with is using satisfaction or usability scales. For example, regularly administering the SUS (System Usability Scale) on the site to get a sense of a baseline score and then setting goals for development projects relating to that. For example: we can't release unless we achieve a SUS score of 85. Then we can report post-release scores to stakholders describing quantitatively how much the experience was improved upon from before release. This is kind of hacked together.

Ideally, you would want to tie KPIs to money: speak the language of the people who invented KPIs and communicate in business terms. Example: ROI of usability testing or conversion rates for an improved user flow in an ecommerce setting, etc. The problem is that it is really hard to tie some UX work directly into profit analysis. Cost/benefit ratios are difficult to assess because UX work is often not the only factor in the benefit part of that equation and true cost can be a pain to calculate.

So my comments might not be so helpful... But maybe this discussion of Livia Labate's take on KPI's is: http://www.userglue.com/blog/2010/02/07/interaction10-livia-labate-ceci-nest-pas-une-kpi/

For B2B applications, one measure I use is referrals and positive tweets/blog posts. In other words, I try to measure the outcome of great user experience, which is that the customer has to tell other people how great the application is.

For websites, Adrian's list is pretty good.