Category: game design

Not marketing the game…but marketing *in the game*. As You may be aware, my car factory sim Production Line got its first taste of marketing as a feature in the last patch (1.26). The current setup is very simple, you can research marketing, then place down marketing facilities that generate ideas, but nothing else (yet). In the next patch, the games marketing system will be fleshed out more, introducing marketing campaigns which you can use to generate brand awareness. here is a screenshot of the resulting new dialog box:

This is all in, coded, tested, and works. So hurrah. SHIP IT.

Maybe not yet, as it requires balancing, and when considering how best to do this, I made the following list of potential balancing factors I should consider to get things right first:

1. The research cost of marketing as a base research item
2. The purchase cost of a marketing facility
3. The running cost of a marketing facility
4. The idea requirements of each campaign
5. The idea generation rate
6. The cost requirements of each campaign
7. The effects of each campaign and the variability of those effects.

Obviously this is effectively a system of counterweights, where increasing or decreasing one will impact all the others. There are reasons for/against changes to any one of these. Ideally marketing campaigns last long enough that they are not too ‘needy’ in terms of interrupting the player. The effects should be noticeable and strategic, but not unbalancing or sure-wins. Marketing will work great as a mid-to-late game feature, so it shouldn’t be attainable too early. It should be something that requires an investment, but not a crippling one, and one that generates a positive ROI in general.

Thinking about the real world, a marketing campaign is essentially a fixed cost, one that you hope makes a positive ROI when spread over a sufficient number of sales. I would like to engineer a system that means it makes sense to run a $500k print campaign if you are selling 10X cars, but maybe not worth it if you are selling just 1X. To put it ion car terms, if you have a single showroom in one city, spending $500k to increase footfall in that showroom by 50% wont be worth it, but if you have 10 showrooms and 10x the potential customers, it will be.

Right now we have a system where there are 3 car body types and 4 price points, so 12 customer categories. Those categories have no fuzziness, so if I want a mid-range compact car, no amount of bargain pricing or marketing will change my mind on that basic fact (I may introduce some fuzzy logic to this later). What this means is, if I have a single car model, then 11/12 of my customers are wasted anyway…

With a general ‘brand awareness’ campaign (the only type I’m working on this month), it will not make any sense (hopefully) to do this for only 1/12th of your customers. In effect, I’d be generating even more footfall from people who come into the showroom and say ‘I keep hearing about positech autos, but they don’t make SUVs in my price range’. A waste of money… If I have a fully kitted out range of cars, that calculation changes.

Lets examine my test numbers:

A print campaign lasts 24 hours and costs $500,000 to run. It boosts brand awareness (and thus footfall) by (on average) 10%. Current footfall is 5 customers per ‘sales interval’ (which is 4 seconds, or 4/120 of an hour) That means we get (on average) 150 customers per hour, and this campaign will mean a bonus 15 customers per hour. 24 hours means a bonus 360 customers. This means its cost us $1,388 per customer through the door.

Assuming that we offer cars to 50% of the possible market (note not all of the 12 categories have the same market share), then we have to double that to $2,777 per customer. Assuming an average car sale of $20,000, we are selling cars with a 13% marketing spend. That actually sounds pretty reasonable.

Of course, the real strategy for the player comes in when they have cars piling up in stock, or not. When you have excess stock building up, you can ALWAYS shift them by dramatically lowering the price. The strategy will be in knowing whether or not it makes more sense to lock in $500k to bring in *more* customers to view those cars, or to lower the price.

I like the way this is balancing out, because without explicitly saying “this is a late game feature”, it becomes evidently true because its not worth advertising to the mass market without having a mass-market range. The real problem is how I convey that to the player. Player-feedback is something that production line is currently bad at, and something I need to work on. I could maybe have a pop-up from your head of marketing that runs through those numbers… a dialog saying “our campaign has resulted in X new customers, so $X per customer, but only Y% of them found a model in our range they would consider. Maybe we need to expand the range?”

As usual; in complex sim games, it becomes a case of 25% code the feature, 25% test the feature, 25% balance the feature, 25% explain the feature. I guess I’m half way there for brand-awareness marketing…

Although in general I’m pretty happy with the design of Production Line, its clear that there are some issues with the way research is implemented in the game. Something I really like is the idea of a BIG tech tree, and players making decision as where to concentrate their resources in terms of competitive advantage. For example, you may focus on production efficiency, or maybe on product range, or maybe on costs, or on high-technology to making high-end cutting edge cars. Different strategies should work.

The trouble is, despite the big tech tree and different approaches, the research-choice decision (which should always be *interesting* as every decision in a decent strategy game needs to be), tends towards more of a nuisance in the late game than a joy. The reason is that research seems to get quicker and quicker, and the costs of doing it become trivial.

At the start of the game, I really need the player to be able to buy some research facilities, otherwise they cannot progress. The problem is, as we go from a tiny little factory to a big one, the cost of ‘one more facility’ becomes relatively trivial. Of course, I can make later techs more expensive in points to research, but you can’t go ‘too far’ in that direction without it automatically herding the player towards doing ‘lvl 1′ techs first and removing that initial flexibility. The flip-side is that if you *not* scale the tech costs, then later in the game, someone who has ignored a specific branch of the tree can normally just go click click click’ and research it in seconds. again: unsatisfying.

FWIW I note factorio has this problem to some extent too, although generally it works better overall because the mechanics of research there are way more involved, rendering the ‘cost’ of the facilities fairly minor compared to the manufacturing of tech ingredients.

Anyway, how do I achieve the following in my design?

1. Make tech tree choice interesting from the start, with multiple paths accessible.
2. Allow tech to continue to be researchable at a reasonable, but not annoying rate.
3. Prevent tech-spamming where the research cost becomes moot.

I’ve mulled over a lot of possibilities. here are some solutions that I have considered, either together or separately:

1. Have hard caps on the number of research facilities that can be built before some other (expensive) admin tech unlocks ‘advanced’ research, thus putting the brakes on research in the mid-game
2. Have variable costs to run or buy research (probably let scientist wages rise as more are hired), making research spamming non-viable in the late-game.
3. Allow research queueing, so that the player can ignore research for longer periods. (Not ideal, as each research SHOULD be a catalyst for production line and car model re-evaluation).
4. Have different ‘types’ of research require purpose built facilities. Maybe design-related research requires dedicated design studios? Maybe super-high tech research requires very expensive dedicated and large labs?
5. Reduce or prevent the immediate construction of any research facility, but require a construction or hiring people for staff. Maybe placing a research facility means it takes 3-4 hours before the staff can be located to fill the facility and start work?
6. Maybe introduce licenses, or patents that act as gatekeepers for research. To research reversing cameras perhaps you need to license a patent for it ($400,000) AND then research it once you have put that money down.

The problem with 1) is it seems very ‘gamey’ and arbitrary, and not intuitive for the player to understand. 2) sounds like it would actually make sense, although I need a decent way of letting the player know about the changes. 3) seems a quality of life improvement in general, but its also treating the symptom (research-popups are annoying) rather than the true cause (research happens too quickly in the late game, and does not seem to have enough of an impact to demand attention.

4) is interesting, and certainly one I’m attracted to. I like the idea of having to place down a design-studio facility, and to effectively research car designs entirely separately. I like the idea of a purely ‘design’ based arm of research for stuff like interior styling changes, new paint colors and types, and so-on. It also seems unlikely that someone who helps design the tire-making press is also working on voice-recognition software.

5) Sounds like its acceptable because build times for facilities are quite common in games, but then how do I justify the fact that everything else in the game is placed down instantly?

6) Could be interesting, and probably plays into a wider revamp of the research system where I need research-pre-requisites not to be limited just to other research items.

I need to get this right, so I don’t want to rush into a solution. I also feel this post is way too designy, I haven’t even tried to pimp the game. QUICK! Add a steam widget!

The hardest part of game design for Production Line has been the sales model. Not hard as in technically hard (frankly after 36 years of coding, not much is *technically* hard), but hard as in balancing the various needs.

The current system for selling cars in production Line works like this:

Each feature in a car has a ‘base value’, which is essentially derived from the resources and time/power/research required to make that component, plus a normal profit margin. That gives a value of say $100 for the feature, and if we include that feature in our car, we will get $100 for it when it sells, assuming we sell cars at a ‘normal’ rate.

If we are wildly profitable, more companies will enter the market and put general downwards pressure on prices, squeezing the profit margin thats considered normal, and forcing us to reduce prices. The reverse is also true.

The value of a feature decreases inversely with its ‘rarity’ as more and more competitors make that feature available. Some features (most) can eventually be considered ‘universal’. There is a multiplier applied to the final value of each feature based on how rare or common it is.

Finally, ‘customers’ appear on a regular basis and shop for cars. They have a probability based approach to buying. So a cheap car is likely to sell immediately, an expensive one will get passed over again and again until eventually a customer buys it.

So that is the current system as it stands, in build 1.22 of the game. Its not bad, it achieves a lot of my objectives. Competition acts a a balancing factor, research is incentivised, and mass market production means you need to lower your prices to compete. However, its not perfect because it assumes that basically there is just one model of car and one type of customer. This needs to be fixed.

My current thinking is that I need to reflect a number of new characteristics in the model:

• Demand for your cars need to be scalable with marketing (not implemented yet)
• There needs to be different overall demand for each body type (sedan, compact, SUV etc), which I can control with events.
• The ‘rarity’ of features needs to be associated with final price. Everyone expects sat-nav in a $50k car. Not so much in a $10k car…
• All of these factors need to be clearly presented to the player so that they understand how the market is working and WHY cars are selling or not selling.

So in other words, if our car showroom has a dozen SUVS where we set a 20% price premium, all of which have the rare and modern climate control, but none of which have electric windows, and competition from other car makers is currently extremely high, we need all of that to be displayed in a way that shows the player why its taking so long to shift those SUVs, and what they can do about it. The current system does none of this.

Also, because the game is in early access, I can’t just take months off to faff around, I need to make steady progress on this stuff. So with that in mind:

The current sales(showroom) GUI does nothing to show the level of price competition with other car manufacturers. This is already in the game, but not being shown. Maybe a pie chart or bar chart showing our current market share would be a good indicator to add to the showroom?

The number of customers who visit the car showroom that do (or do not) buy a car is also currently modelled but not shown. I should probably get some GUI added to reflect this, even if its just (for now) a percentage indicator showing the number of people who did/did not buy a car. (Plus I could also show the average length of time each car was on the showroom floor?

Once that is in the game, I can start thinking about modelling those per-body-style markets. This could be a multiplier that reflects the needs of each customer, so instead of being open-minded, they could now be SUV buyers or Sedan Buyers. This would allow me to show a breakdown of the visits to the showroom by each customer, and also breakdown showroom-times and sales percentages for each body style (or each car model) in an additional stats tab in the showroom window?

And then phase III would be modelling the different expectations of features on a car-price based model (electric windows expected on cars above price $X…). That would require a fair bit of internal code wrangling PLUS some fancy GUI changes.

Anyway, I’m interested to hear peoples views on all this, does this sound like its going in the right direction?

So, when watching a lets play for my new game Production Line, I encountered the point that ‘surely compact cars should sell for less than an SUV? given the same options…?’ Which is of course absolutely true. As a result I immediately leapt into my code and made it so. There is now a price modifier for each body style, with compact cars being cheaper, sedan being the default and SUV being more expensive. So far…so good.

But in the current game, producing all 3 cars takes the same time and resources, so why would anybody ever build a compact car? Suddenly I have introduced yet another dodgy piece of game balance. Argggh. In the back of my mind, I have always planned to simulate multiple ‘markets’ for cars (sop one sort of customer is looking for a compact car, another for an SUV etc), so the strategy of purely producing SUVs would be a bad one, but there still remains the problem that I have hard coded higher profitability into the larger vehicles. Plus this means that because it only applies to the base car features (wheels, doors, roof etc), there is an additional incentive to sell ‘basic’ SUVS, whereas we all know that cars make a lot more profit on the extras…

So to quote Tolstoy, ‘what is to be done’?

The obvious solution here would be to reflect the extra effort in building a bigger (or smaller) vehicle in the game mechanics, and have everything balance out. Presumably making a door for an SUV is harder work and more expensive than making one for a mini. Also in theory I guess you need bigger stronger robots exerting more power to lift heavier doors, yada yada. The amount of steel in the roof of an SUV is likely noticeably bigger than for a compact car, and so on.

The trouble is, I have boxed myself into a design corner in my game by making resource units ‘discrete’ (ie: not fractional). Components use up ‘1’ steel or maybe ‘2’ steel, but never ‘1.2’ steel. To change this would not only be a huge endeavour, I suspect it would lead to confusion, as individual items of resources are represented graphically and fall into neat slots. As a result of this, pretty fixed, design decision, I don’t think realistically I can change the resource quantities needed to make different car body types.

So the alternative that presents itself is to instead vary the time taken to assemble them, which is fractional, and could be relatively easily adjusted. A set of robots at the ‘fit doors’ slot could easily take 20% more time on an SUV than a sedan. I could probably code that in 30 minutes. The only problem there is how to represent that to the player. The window that shows the slot status for something like this does show how long the task will take, but there is no further breakdown. Hovering the cursor over some upgrades will tell you how much time they have added or subtracted to the total, but that involves some maths by the player to work out the ‘base’ process time.

Maybe this is something I just have to accept, and possibly just provide a little ‘i’ icon to hover over which shows a breakdown saying that the task takes 2 minutes 10 seconds, plus 2.3 seconds because we are fitting an alarm as well, minus 1.2 seconds because its a compact car, – 0.65 seconds because we have extra robots…etc…

Too clunky? or a reasonable compromise? I haven’t decided yet.