Cannibalism in business used to be a simple thing. It used to be, if you had a successful product that competed in, say the automobile market and you release a similar product, sales of your initial product will start to slump. It's called cannibalism because sales from one of your brands are eating into the sales from another one of your brands.
This can be good - if you sell 100 units of a product and introduce a new product, your sales of the old product might drop to 75, but the new product will be selling 50 units as well, so you still end up selling more even if the numbers shift. Assuming an active and competitive marketplace, even the illusion of competition can be beneficial, both internally to the company and in the market as a whole. But problems come when the market fluctuates, and markets unfortunately tend to do that.
Think of it like this: when any given market shrinks, it doesn't deflate evenly like water going down a drain, more like the ripples left in a pond after you throw a stone in it. Richer and poorer sectors of the market will shrink less while the middle class will drop radically. If your target demographic includes those upper or lower echelons you'll lose money at close to the expected rate, but in that middle class (where, pound for pound, all the money lives) will sink like a stone. But, unlike the rest of your competition, you'll be doubly screwed because you're in the sector twice. That's why forceful brand differentiation is a great idea.
It also used to be, when you had one brand eating into another, it was because of publicly noticeable and translucent business practices - you introduce Diet Coke to the Coca-Cola line, and two things happen: sales of regular dip and then rise again (dip because of the new product, and rise because of the improved state of overall brand awareness that comes with a successful new product launch) and sales of the new product average out to what its standard selling ratio will be.
That was the way it used to be. Now, things are very different, slightly trickier and much more ethically strange. Now, because of the effect of over-arching parent companies, the people who buy your products have no idea that you make other products in the same market. The people you're selling to matter, and complete brand differentiation can be the right thing to do if you're fairly certain that their markets won't overlap, but if they do...well, for instance:
How many people knew that Amstel Light is a brand owned by Heineken? Amstel Light was developed for idle drinkers and women, because research showed that everything about Heineken that made people love it (it's hard-edged German pronunciation, bad ass symbolism and the utter, well, manliness of it) was turning off more social, less aggressive consumers (ie women and responsible adults. No, that wasn't a shot.) so a new product was introduced to bring them into the family. Fine. They're not competing at all, really, just increasing market share. Never a bad thing.
But then something weird happens: Hard-core drinkers discover that they like the taste of Amstel Light better and that, more importantly, they can drink a ton more of it without getting full. The market shifted. It happens.
But you're stuck in a tricky situation - you can't change Heineken without losing some of your customers to outside brands like Budweiser, but you can't scale back production of Amstel for the same reasons. And you're losing customers, because light beer drinkers rarely trade back up to your full-flavored beer, they find other light beers that you can't control.
So what do you do? It's counterintuitive as hell but so amazingly cool: you introduce a third beer. Heineken Premium Light. If you target it juuuuuust right, you'll pull Amstel drinkers back to your "hard-edged but so not wussy" image, and your "I love this Heineken, but I'd look like a pansy ordering a light beer" image, and probably suck in some of the "I'm sick of feeling like a pansy with this other beer, but original Heineken gives me a headache" crowd. Your little corner of the market stabilizes and all is right with your world.
...and if you're wrong, your new beer devours the market of your other two beers and you find yourself back selling one kind of beer but spending the money to produce three.
THAT'S why it's called Cannibalism.