Every single non-direct source of traffic on mobile is modified in some way by an optimization algorithm -- that is, there is some set of historical data that is being used as guidance for targeting your campaign based on the features that are available to that traffic source. Unless you're buying traffic directly from an exchange or DSP, your campaign is being targeted in this way to the subset of impressions that the traffic source you are using can access. This is true for Facebook and Google, which self-attribute, as well as the SDK networks like Unity, Applovin, etc., although they have far less data about the end-users to whom the impressions are being served than Google and Facebook.
Whenever you make a change to your campaign's targeting, the optimization algorithm needs to re-calculate (or at least update) the parameters it uses to target traffic because the data it had for your campaign is no longer relevant. If you are targeting very broadly, this can happen quickly: the campaign will show your ad to lots of people, observe conversions, and quickly come to targeting parameters that work within whatever constraints you have set for the campaign (like CPI, CPE, budget, etc.).
And, of course, the opposite is true if you are targeting very narrowly: your campaign will not be able to experiment as easily as with broad targeting and so it will take longer to accumulate data from conversions and build a targeting model.
This is the process that that is happening when ad network account representatives say that a campaign is "scaling up" or "tuning," and it's why Facebook doesn't allow campaigns to be set for CPI bids until enough installs have taken place on a CPM basis (and why you can't use the VO strategy until the campaign has seen enough revenue events).
So the answer to your question is dependent on context on the overall budget, how broadly you are targeting, and the channel you are using. But in general, especially on Facebook, it's not a good idea to radically alter a campaign's targeting: unless you are worried about audience overlap, creating a new campaign with the new targeting parameters and scaling down (or pausing completely) the old campaign is more efficient.