Car Insurance Technology Not Joined Up

I'd love someone to explain why it seems impossible these days to get any degree of consistency from organisations, especially when they have so much computing power available to them.

I recently got my car renewal letter and, like many people do these days, I decided to do a market sweep.  Of course, this is easier today because of the innovation of the aggregators like Comparethemarket.com (others are available, of course).

Anyway, I got a list of quotes cheaper than the one I had received frommy existing insurer (which itself was commendably only £7 higher than the previous year).  Recalling an article I had read recently that suggested it could be even cheaper to go direct to the insurer's website, I then picked on the one I fancied (and at around £60 less than my current insurer, quite a good deal), and entered another quote request, instead of clicking through from the aggregator.  As far as I can tell, the main difference in the data request was the question of where the car is stored.  I entered 'in a garage', which is true and, I imagine, the option that is surely the lowest risk and should therefore keep the premium low.

Imagine my surprise when the premium returned was around £30 higher than the 'Comparethemarket' equivalent.  Just at that moment, a window popped up inviting me to chat.  I asked why the premium had calculated out differently from the aggregator site and, predictably, the lady couldn't help me, other than to say that it could have been the answer to the 'garage' question, which is preposterous, as I would have expected that to shave some cost off the premium, not add to it.

Anyway, apparently the best solution to the issue she could offer was simply to click through from the aggregator site, which of course I will did, and then saved the quote for now.  Of course, I'm also going to haggle with my existing insurer first, but the experience hardly fills me with confidence that these people are fully in control of their propositions, and the execution of those propositions, if there can be such a disparity in the quotes, using the same data.

Anyone else have the same experience?