Visual Entertainment and Technologies Forum
Purchasing AI - Printable Version

+- Visual Entertainment and Technologies Forum (http://www.ventdev.com/forums)
+-- Forum: GearCity (http://www.ventdev.com/forums/forumdisplay.php?fid=1)
+--- Forum: General Discussion (http://www.ventdev.com/forums/forumdisplay.php?fid=2)
+--- Thread: Purchasing AI (/showthread.php?tid=3732)



Purchasing AI - AT - 03-11-2018

My question here is the somewhat broad "what determines how many vehicles are sold", particularly in cities where there is a very large amount of competition. I just bought the game a few days ago (running Test Branch 1.23.1 new main map), and my expectation would be that the main drivers of sales (1900 start) would be price and overall rating, specifically the overall rating that fits the vehicle type. But I consistently notice that cars with higher ratings and lower prices are not outselling their rivals, and particularly even wild outliers in price can often get really decent sales. Going into a sandbox game, I developed cars which cost me over 2k and sold them for the cheapest in the big market. Somehow sales were nominal. I proceed to max out my branch spending and try again. Sales increase but not by too much. I figure the key is marketing, dump max into 3 categories of marketing... sales double and I do become the market leader but not by a terribly big margin. I figure the problem is therefore my brand image, so I create a new sandbox game and start with higher (though accidentally not higher than the AI just close 40 vs. 50) in luxury, and higher into the regions of image that the AI seems to have none of as well. I quickly develop a reasonably balanced car, sell it for approximately the same as the AI and sales are again pretty mediocre. Now I wonder if there are thresholds where if, say, my Cargo is not at least X certain customers just disregard my cars completely? I'm not entirely sure. And as a final test I "observed" an AI game and let it run about 7 years from 1900. The best seller (~100) is a 1903 car with a super low HP engine and pretty mediocre everything. A new 1907 model priced the same with higher values in almost every category sells about 20. Perhaps it's unfair to compare these since I don't know what the AI was really doing, but it seems to me that customers are picking cars off of some weird factors that are not obvious to me (and overall rating alone seems to have no bearing on anything).

So I wonder if I'm missing something fundamental about how the game works and why these supposedly mediocre cars are selling so well, or if the AI is actually being too generous in its purchasing decisions. I tried adjusting for many possible "X Factors" and found none. If my car (or another's car) is vastly superior to the field, why does it have such low market share? Is it actually a distribution problem related to dealerships, or what else could be happening here? I'd like to play with 300 AI because it makes the game feel more alive, but I can't actually figure out how to get the lion's share of a city's profits when a city has any competition. Any help on the topic would be appreciated.


RE: Purchasing AI - Eric.B - 03-12-2018

There are well over a hundred variables that are taken into effect when vehicles are sold. So I can not break them all down for you in one post. You'll have to wait for the "In-Depth Manual" for a break down of the formula.

But I will do my best to point you in the correct direction.

Quote:and my expectation would be that the main drivers of sales (1900 start) would be price
Vehicle price is important in getting the maximum amount of people who can afford it interested in purchasing the vehicle. For example, using modern money values, if you were making $20,000 a year, you would not even bother shopping for a new Ferrari or Rolls Royce. Instead you would look at new cars between $15,000 and $25,000. There are a lot more people who can afford Honda Civics than Ferraris.

So lowering the price increases the pool of potential customers. You can actually see this pool by clicking the "Line Graph" button where you enter prices for your branches.

Pools overlap somewhat, but not always. Again using modern money, while someone who makes $150,000 a year can afford a $20,000 Honda Civic, they are not interested in purchasing it. Why? Because their economic demographic is too high. So depending on how much price distance between your vehicle and the competition is, you may not even be competing for the same customers.

Secondly, the lower the price of your vehicle is, the more competition it has in the used market. The $20,000 new Honda Civic Sedan has to compete with the $20,000 4 year old low mileage Jaguar that cost $85,000 new.

Quote: and overall rating
The overall rating does not factor that much in vehicle sales. Using the same cars above, a Honda Civic would have a higher overall rating than a Ferrari. It's a much more useable vehicle. But if you were in the market for a performance car, and had unlimited money, which would you buy? As such the overall rating is not that important. This is why we show importance stars when using the advanced designer. The ratings effect on consumer AI is weighted for the type of vehicle they are. A very rudimentary example of this rating is the Specific Vehicle Type Rating. This is what you would use to compare the quality of your vehicle to other vehicles of the same type.

Quote:. But I consistently notice that cars with higher ratings and lower prices are not outselling their rivals, and particularly even wild outliers in price can often get really decent sales. Going into a sandbox game, I developed cars which cost me over 2k and sold them for the cheapest in the big market. Somehow sales were nominal.
The main issue here is you're selling cars for 5x (or more) per capita in 1900. This is equivalent to $250,000 in today's money. There is very little difference in the amount of people who make $250,000 per year and people who make $1,000,000 a year. Specially in early game years. Instead, try doing a vehicle that costs $500. And see how well it does against competition selling for $2000. You won't be competing in the same economics group, the masses will be able to afford it, and you'll get more sales.

Quote: I proceed to max out my branch spending and try again. Sales increase but not by too much.
Branches sell vehicles to dealerships, dealerships are third party companies that sell vehicles to customers. Increasing the sliders of the branch will take years to build up and increase the amount of dealerships your company supplies. The more dealerships you have, the more sales you'll make.

Quote: I figure the key is marketing, dump max into 3 categories of marketing... sales double and I do become the market leader but not by a terribly big margin.
Marketing also takes several years to build up to max value. It just doesn't happen over night. It is also only effective against a lot of competition. So it sounds like whatever market you are in is flooded.

Quote: I quickly develop a reasonably balanced car, sell it for approximately the same as the AI and sales are again pretty mediocre.
Depending on the vehicle type, balanced may not be a good idea. Phaetons don't need luxury. Microcars don't need performance. Supercars don't need cargo space. Etc.

Quote: Now I wonder if there are thresholds where if, say, my Cargo is not at least X certain customers just disregard my cars completely?
Cargo space matters for some vehicle types. Such as pickup trucks, vans, luxury cars. Just as smoothness of the engine matters for luxury types vehicles. I suggest trying the designer in advance mode and reading the help buttons for more details.

As for minimum requirements, the only minimum requirements are minimum top speed. Which you will get a warning for in advance designer. If you're building $2000 unit priced cars, you're not hitting this limit. The other minimum is for fuel type popularities. You can view these in Reports -> Reports 1 -> Body and Fuel Type Demand. If you're not using Gasoline, you're limiting your sales.

Quote:And as a final test I "observed" an AI game and let it run about 7 years from 1900. The best seller (~100) is a 1903 car with a super low HP engine and pretty mediocre everything. A new 1907 model priced the same with higher values in almost every category sells about 20.
By 1907, there is more competition in the game and an mini economic depression. There is no wonder that the sales fell.

Quote:but it seems to me that customers are picking cars off of some weird factors that are not
In short, the customers all follow the same formula.
-Consumers are pooled by vehicle type, demographics (wants), and economic demographics.
-Each car of a type, in each city, is given a buyers rating for each pool. If you're selling sports cars, then performance and power are more important than luxury or safety. The overall rating weights luxury and safety the same as power and performance. Consumers do not.
-Additional information effects your buyers rating, such as the number dearships you have, marketing, image ratings, what type of fuel you use, if your car is lower than the minimum speed, the price, etc.
-Used vehicles are given a buyers rating
-All buyers ratings is summed together for the entire type for that pool.
-Your buyer's rating is divided by the sum, that is the max percentage of sales of all customers for that pool you can get.
-The price of your vehicle is then compared to the economic demographics you're in. The number of sales are extrapolated from that.
-Consumers that don't purchase a vehicle are recycled to the start and it's repeated again.

So no black magic funkery going on. Everyone is playing by the same rules.


Quote:So I wonder if I'm missing something fundamental about how the game works and why these supposedly mediocre cars are selling so well, or if the AI is actually being too generous in its purchasing decisions. I tried adjusting for many possible "X Factors" and found none. If my car (or another's car) is vastly superior to the field, why does it have such low market share?

Without actual details from your save game, I can't really tell you what is going on. Some obersvations:
1) $2000 unit priced car is not cheap. There aren't going to be many customers.
2) It sounds like there is a lot of competition in the city. This effects everyone equally. New cars in the market does not increase the number of customers. If you sell 50 cars, and I sell 50 cars. And a third company jumps in. All of a sudden you're selling 33 cars, i'm selling 33 cars, and they're selling 34 cars.
3) Age of the design matters, if your designer is older than 5 years, its time to build a new one.
4) Fuel Type is very important.
5) Every vehicle type has important ratings. Maximize those ratings.

If you do a search on the forums, specially the steam forums, it'll have several post of me answering similar questions.

if you want to upload the save game on here or email it to me, I'll be more than happy to point out what you're doing wrong http://wiki.gearcity.info/doku.php?id=troubleshooting:steam_installfolder#sending_a_save_game . But most likely it's high priced vehicles + lots of competition. Try for going for lower priced vehicles if you want to top the sales charts. Also overall rating does not matter much, focus on specific vehicle type rating.


RE: Purchasing AI - AT - 03-12-2018

Let me first thank you for writing up such a detailed reply. I hope some of the particulars of my question were relatively unique, and were not something that a precursory search could have easily answered for me, because I did do some research on Steam.

Let me clarify one or two small things that, judging from my reply, I did not explain very well. Mainly, was that I understand overall rating is not [very] important—my questions were more so directed at using the Vehicle Type Specific Overall Rating (exact term is probably not right) as a general indicator and whether or not there were some thresholds on specific values that had to be met. You did answer this for top speed, I believe you mean the red text warning in the advanced vehicle creator. Is it true that the other red text warnings work in similar ways, effecting a decrease in overall buyer rating in a way that is not necessarily reflected in Vehicle Type Specific Rating? If this is true then perhaps my low (relative) sales were caused by this—I assumed that Vehicle Type Specific Rating was by far the most important factor for vehicle desirability and that if that rating was still good then the customers were content with my "red text" flaws. But if that's not true (and it seems to not be) then that would explain a lot.

Also I should clarify very briefly that all my comments regarding "low sales" are relative to the total of cars being sold: I understand that the low GDP of the early years makes total sales low.

Your explanation of the Customer purchasing formula was very helpful (and largely what I was looking to hear). It seems that with these extra factors it should be possible to eke out good sales even facing a lot of competition, but I just wasn't establishing myself as superior as well as I thought I was. I hope to be able to read the "In-Depth Manual" someday soon, because I can tell a lot of effort and research went into determining your formulae.


RE: Purchasing AI - Eric.B - 03-13-2018

(03-12-2018, 05:21 PM)AT Wrote: Let me first thank you for writing up such a detailed reply. I hope some of the particulars of my question were relatively unique, and were not something that a precursory search could have easily answered for me, because I did do some research on Steam.
I've explained most of it before. But it would be buried in bits and pieces in 8 years worth of posts. So it's fine. I do not mind typing this stuff out. I write quickly, although a bit sloppy at times. Forgive me if I get a bit terse when I write long replies, specially since it's 3am Smile .

Quote:Mainly, was that I understand overall rating is not [very] important—my questions were more so directed at using the Vehicle Type Specific Overall Rating (exact term is probably not right) as a general indicator and whether or not there were some thresholds on specific values that had to be met.
Yes, the type rating is a general indication of how well the generic consumers for that vehicle type sees the vehicle. However it is not 100% accurate. As mentioned before, some customers may want a sporty minivan, some might want a luxurious one, etc. Also the value is not recalculated with new values at as states degrade. So the values displayed may be a little lower than they actually are for older vehicles. While some stats fall at a constant fixed rate, such as safety, other stats like the fuel rating are tied to specific specs. Therefore they do not decline. Sadly there is not enough processing power in the game to recalculate all the type values anytime ratings decline, thus the type rating falls at a standard rate. If you're working on more spec oriented cars (power, fuel, cargo), this might be more an issue.    


Quote:You did answer this for top speed, I believe you mean the red text warning in the advanced vehicle creator. Is it true that the other red text warnings work in similar ways, effecting a decrease in overall buyer rating in a way that is not necessarily reflected in Vehicle Type Specific Rating?
The Type Rating is only for the individual ratings in the game that have the stars, IE they're abstracts (or used to be before changes to the game were made). Off the top of my head, Safety, Luxury, Dependability, Performance, Power, Quality, Cargo, and Fuel Economy.

So no, the minimum top speed requirement, nor anything else beyond those ratings are in the Type rating.  

As for how the minimum top speed system works. (your speed / minimum speed) * number of customers that would have bought your car if you hit the minimum speed.  This is a penalty added to the game to prevent people from making 1hp cars and selling them for $100 in 2020....


Quote:If this is true then perhaps my low (relative) sales were caused by this—I assumed that Vehicle Type Specific Rating was by far the most important factor for vehicle desirability and that if that rating was still good then the customers were content with my "red text" flaws. But if that's not true (and it seems to not be) then that would explain a lot.
The type rating only factors in the "star" ratings in the designer. This I believe is explained in the help buttons. It does not include everything else, not limited to:
Branch Selling Efficiency
Number of Dealerships
Marketing
Price
Fuel Type Popularity
Engine Smoothness
Gearbox Smoothness
Minimum Top Speed
Vehicle Image Rating
Company Image Ratings
City's infrastructure ratings (effects various things with chassis and vehicle dependability)
City's fuel prices
Global Fuel Prices
Global interest rates
Used vehicle effect
Sub-demo desires
How well you designed for your selected demo focus
Amount of competition directly effecting you because they target the same demographics (where other vehicle on the market may not)
Little bit of randomness.
Etc....


Quote:Also I should clarify very briefly that all my comments regarding "low sales" are relative to the total of cars being sold: I understand that the low GDP of the early years makes total sales low.
Sure, but without numbers I can't really tell how low you're talking about.
Are you selling 40 and they're selling 50? Yet your car is rated 35 and theirs is 28? What about all the other companies? How many overlap you. Are you out selling everyone within 10-20% of your price range? What about these lower rated vehicles out selling you, how big are their dealer networks? Etc.

Quote:Your explanation of the Customer purchasing formula was very helpful (and largely what I was looking to hear). It seems that with these extra factors it should be possible to eke out good sales even facing a lot of competition, but I just wasn't establishing myself as superior as well as I thought I was. I hope to be able to read the "In-Depth Manual" someday soon, because I can tell a lot of effort and research went into determining your formulae.
No worries, and if you have any more questions feel free to ask.