TL;DR: Only buy this game if it is on sale and you absolutely love 4X games. If you want to spend a few dozen or hundred hours testing your patience, then this may be the game for you. Otherwise, perhaps gift it to someone you want to see spend hours on end beating their head against the wall.
This game is a great time sink, but there are lots of poorly executed concepts. Item durability, especially on arrows, is a joke. Your arrows can (and sometimes will) break before you run out of arrows in a single fight. The best magically enhanced arrows have less durability than the "heavy arrows" which themselves have to be repaired literally after every single serious fight or they will break during your next battle.
If item durability was the only flaw this game might still be a gem, but even concepts that are well thought out are not well developed. Exploration of owned territory is an interesting idea, but there is often no real reason to bother completely exploring areas. Your rewards for doing so are threefold. You receive a tiny immediate 2 gold increase in total income, you discover a large variety of new enemies to battle, and the settlements in that area can expand into the newly explored areas, eventually generating significantly increased income. This mechanic would function a lot better if there was a small amount of exploration that took place automatically in provinces with guards. The AI doesn't explore much manually, but it does have guards in every province so at the very least the AI would have some small amount of exploration done in their own territories. Instead when you take territory from an AI opponent every single province will be gripped under the stranglehold of "overpopulation". And that brings me to my second major frustration with the game.
The AI doesn't understand the mechanics, either in battle or on the world map. As an example of the flawed combat AI, ghosts can only be damaged by magic damage. However, if you try to autoresolve a battle with a ghost in it you will almost certainly be completely annihilated by the ghost because the AI will use all the physical damage it can against the ghost, while units capable of magic damage busy themselves with other tasks until your entire army is dead. This means that armies that are labelled as something you can fight without taking casualties are often impossible to defeat, where enemies that are labelled as "I will be trampled" can often be manually defeated without taking a single point of damage.
On the overland map the AI spams a granary in effectively every province. The granary itself generates no income, and takes up one of the limited building slots. The granary can be explanded to a stable which would give the AI some tiny additional income as well as allow it to move more quickly through its territory, but I have yet to see a province in which the AI had actually upgraded one. A storehouse on the other hand can add 10 gold per turn income to a province when it is fully upgraded, otherwise it will offer 5 at tier 2 or just a single gold per turn at tier one. Still, these are the most readily constructed source of income, while it takes awhile to be able to upgrade them to tier 2 you can quite readily increase your total income by a third or more by having one in every province. It's a crying shame the AI doesn't understand how to build a thriving economy, if it did perhaps there could be a provincial governer (or some such) that would automatically build the upgrades you wanted akin to automating workers in Civilization. Building a mall in every province you have is a chore when you have twenty provinces, but it's ludicrous to even consider bothering once you've gotten to the point where you have 100 or so provinces. If the AI could build wealth it might be able to field larger armies more often. I have thus far only rarely seen the AI taking advantage of units higher than tier 2, and there are 5 tiers of units that can be recruited.
As an example of something that is ridiculously easily fixed, and yet remains flawed for no good reason, just look to the autosave system. The autosave system autosaves at the end of every several turns, rather than the begining of every several turns. If they had taken the approach of labelling these saves Turn ___ or something along these lines this wouldn't really be an issue. However, they've gone the route of having 5 autosaves that roll over each other. Not necessarily a problem, but there isn't anything behind the scenes that tells the game that it has autosaved this turn in say the slot Autosave 4. So when you load Autosave 4, and end the turn, you will create Autosave 5. Does the war seem inevitable at that point? Load Autosave 3, but when you end your turn you will create Autosave 1. Get to the turn that originally was Autosave 5? Now it's created Autosave 2, and if Autosave 3 isn't early enough to head off your issue your older autosaves are gone. This is easily avoided using a rotation of manual save games, but it's just one tiny example of a problem that could likely be fixed by changing a few lines of code and remains in the game just to annoy players.
I wish I could introduce Snowbird to the concept of numbers that end in 0, it would be a lot less frustrating to see that I've done 2 damage or 5 damage to something with 50 health than that I did 0 damage four times in a row firing Ballistae at one skeleton (each with damage potential labelled as 0-5 on a 5 health skeleton). And that's just another in a litany of odd design decisions. If a flying creature flies onto a hill, it gains +1 range for its attacks. But it already flies, it's flitting about in the air, why on earth should the topography below it matter? If anything flying units should constantly receive a +1 range buff, perhaps losing it when they run out of stamina. There's a reason why it's a fairly common practice that ground based effects, positive and negative alike, are typically ignored by flying units in games. Siege engines are considered mechanical units for the purposes of healing spells, but they're not immune to being petrified by a medusa. There is literally only one unit, at least according to the wiki, who can gain the Repair ability so that they can repair siege engines, and it is the militiaman which is one of the weakest units in the game. Both the Ballista and (clearly described as self propelled, yet also capable of being petrified) Catapult are described as having been created initially by the Dwarves, but the Dwarf cannot gain repair. Interestingly the Dwarf can gain petrification immunity, and now I have nearly come to the end of this tirade. I am baffled anew by this game every time I play it, I understand many of its quirks, and yet I am consistantly scratching my head and asking "Why?". Why did they make every quiver of arrows in the game have the potential to break if you fight two back to back battles when you can very easily be required to fight at least three armies during a fortress siege (the guards, a defending hero, and the garrison. You will honestly likely fight five as two more armies rush home to commit suicide as an act of solidarity with their comrades). Early game a quiver of 10 arrows will break before you've even run out of arrows, apparently you break arrows like a drunk having a nicotine fit might break cigarettes.
This game has had a lot of thought put into it, but it seems to me that very little logic has gone into the development. I almost wish it had come out alongside Master of Magic and the first Civilization. At least then I could have more readily forgiven its faults and the games it inspired would be able to refine the good ideas and burn away the dross. As it stands at the end of the day there are better 4X games out there for your purchasing dollar.