There are a number of possible reasons. For instance, the credit inquiries probably weren't at all three CRAs (although some creditors will pull all three). Also, you most likely previously had a different number of inquiries at different times for each CRA. So the impact of a new inquiry is going to be somewhat different at each CRA.
Similar things could happen with the balance increase. Even if it reported to all three CRAs, one might code it differently from the others. Or, again, what was already there might be different for each CRA, so impact of the change will also be different.
And it could be something completely unrelated to the events you know about. For instance, something could have aged in such a way at one CRA that made a change.
There's also something else. Although the goal is for FICO scoring models to produce essentially the same result at each CRA (assuming the exact same information being reported), it doesn't always work that way in practice. Think of it like a computer gaming console. XBox and Playstation are designed to do basically the same thing, and they are essentially the same technology. But there are also some obvious (and some not-so-obvious) differences. When a game-maker publishes a game for both platforms, their goal is for the game to play the same way on both consoles. However, because there are differences in the consoles, games don't always play exactly the same. They're (usually) close, but there are always going to be some fundamental differences.
That's how the CRAs are. All three are designed to do essentially the same thing. But there are differences in the way they handle and process the information (because each one developed their system on their own). The FICO model is like the game that gets played on the CRA console. It has to be tweaked for each CRA in order for it to work. And although the goal was to produce the same score (given the same input), there are bound to be some differences.
