We've been thinking a lot about WTP lately and have been researching and developing better ways to estimate it.
First, realize that as realistic as we try to make our studies, in a survey, respondents don't spend real money, they spend survey money - you don't have to work as hard to earn survey money, so it's easier to spend, and this inflates WTP.
One thing folks have recommended in classes when we've taught simple methods to estimate WTP is that rather than comparing just two products (one with the feature and a higher price and one without the feature at a lower price) you might want to simulate a scenario that contains your test product (without the feature and at the lower price) along with the none alternative and several (4-8) competing products. Get the share of your test product in this scenario and write it down. Now add the feature to the test product and lower the price until it's the same as it was before when you wrote it down. The difference in price you get now may well be a lower, more credible, estimate of WTP.