If one must use the term data (much compromised by the brilliant Star Trek character), please substitute "data-informed." The problem with letting the data take the wheel is that they have an uncomfortable habit of not turning out to show (or even be) what they purported on first glance to resemble. There is a bent towards the third person in the discourse of science, a pretense to objectivity as a matter of creed which hopes to remove the observer effect by leaving her out of the grammar altogether. But practicing scientists know how fragile and ambiguous our perception, not to mention our recording, of what we observe can be. They take pains to be careful but mostly recognize the ease of failure. This is part of the reason for the insistence on repeated results in experimental science.
The practitioners of the so-called social sciences are deservedly looked down upon by their "hard" counterparts, since so much of what they claim to be studying scientifically is not capable of rigorous experimental investigation. There is a pretense to scientism in the very labeling of the field, and most practitioners would probably admit their misgivings in small groups, though perhaps not for the record. Education professionals have recently been hoodwinked by a cadre (one might even call it a cabal) of pretenders to authority who have shoddily fudged a lot of their so-called data to make a case for the reckless destruction of our public education system in order to privatize and profit from that system.
We must vigorously resist this "deform" movement. The imposition of a regime of testing is not a neutral data-gathering effort but part -- if not the root -- of the problem. Assessment of learning is different from testing. Good teachers are constantly assessing--to be able to notice and take advantage of the teachable moments which arise with regularity as they interact with students about their learning, they must be on perpetual alert for the level of current understanding being displayed by the student and the best way to help her reach her next step into her zone of proximal development. This interaction cannot be "driven" by any form of data I've ever seen. It's an ineffable process, challenging to teach to others, certainly capable of being informed by many kinds of data, but not a moment I would ever surrender to the direction of a spreadsheet.