Preference Share + Regression

I have conducted a conjoint survey evaluating several combinations of user and devices, with the end goal of calculating a preference share of each possible device per user profile.

My initial plan was to calculate preference share using conjoint utilities, but there are too many interaction effects present between user and device attributes. I am trying to figure out a strategy (perhaps a regression analysis?) as an alternate way to calculate preference share.

Another complication – we would like to consider preference share per user profile for varying subsets of device profiles. For example, Devices A, B, C… Z may exist, but we would like to identify the preference share when only Devices A, C, and F are available on the market, and then re-evaluate as the market expands to include Devices A, C, F, and G.

Any suggestions would be welcome!
No problem. We've got 10 cell phone users profiles (65+, male, retired; 20-25 year old, female, student; etc) and 20 phone profiles - some phones are currently available, and some are under development. We would like to understand what the share of preference per user group is for various combinations of phones.

For example, for the 65+ population, what percent would select each phone? How would this change for the second scenario?
Scenario one: iPhone 5, iPhone 6, Android
Scenario two: iPhone 2018 iteration, iPhone 5, iPhone 6, Android


Omega Contributor
So you have current information about who has which phone and you want to create a model to be able to say males have 2 times greater odds of selecting phone A, persons over age of 65 are 3 have 3 times greater odds of selecting phone B.

Something like this? How many phones are there and how many people do you have data for?
I'm not sure if we are on the same page. Per user group (ie. 65+ males), X% would use an iPhone 5, Y% would use an iPhone 6, and Z% would use an Android. We then want to know how X/Y/Z% among this user group would change when a new phone is introduced to the market.

We are considering about 200 unique phone profiles, but would only determine % market share for about 8 phones at a time. (We are trying to understand how phone preferences would change depending on the characteristics of the phone released to the market). There are about 400 unique user groups.