I try to zero everything out to start.
My initial goals are, of course, to satisfy domestic demand [based on any number of indication parameters, such as DAR, and Immigration, and acceptable tax revenues,...]. If investment seems warranted, then I start investing in the base commodities first.
Remembering that Electricity costs depend on fuel costs, first I would try to get fuel costs down somewhat [using WM prices as a benchmark, but not necessarily a target], then try to affect electricity itself. This can be tricky if you are also exporting coal or oil, as your efficiency investment is a factor of total production - so if you bump up oil production, investment increases [and it is apparently non-linear, and also affected by inflation]. Rule of thumb, don't invest deeply in commodities that you expect to overproduce.
Getting power costs under control greatly affects the costs of finished goods, especially consumer goods, so stabilize this first. Do investment changes on one commodity at a time, and validate them over at least a week of game-time - look for 'percolation' effects.
Another odd rule of thumb - if you are running commodities in demand mode, and this is usually the case, try setting the demand target to 98% instead of 100%. You will generally end up producing a small surplus anyway, and in the long run it will seriously reduce the impact of investment outflows. Much micromanagement went into validating this recommendation, but I would urge you to validate it for yourself -don't take my word for it.
Another clue - changing your investment rate also impacts your effective markups, in an odd way. If you reduce investment, you need to also reduce markup, as it is a percentage of cost: reducing investment raises consumer costs. There is no real linear relationship here - that is, dropping investment a tick AND dropping markup two ticks at the same time might still result in a domestic price hike - so be very careful to monitor domestic prices when markup and investment are active and changing.
Once markups are stabilized, you might think that investment could help to increase margin on domestic sales - and it will, except that in very extreme or liquid inflation situations, it can create a lot of havoc. Get markup, demand and inflation under control, and use investment judiciously when all else fails.
It's all particularly difficult in the early years of a phase, as you are probably also upgrading or adjusting capacity. Running a bunch of low-efficiency facilities has real costs - so as you replace or upgrade these, you will have to also adjust investment. To simplify the upgrade process it is [IMO] easier to not have investment active if possible [as you can save some time by avoiding cost stabilization when new facilities come online].
Colorless green ideas sleep furiously [but otherwise, they do not worry and are happy].