I spent the largest part of my professional career thus far at the @WorldBank. During that time, I was involved in formulating policy for a wide variety of nations, both democratic, and less so. So FWIW, here are some thoughts on the policymaking process in a democratic society. 
For good or ill, decisions made by the majority (or representatives of that majority) apply to all citizens in the country. One may individually disagree. If so, the time to debate a policy’s merits is prior to enactment. But once policy becomes law, all are compelled to follow.
It would thus be profoundly undemocratic (and discriminatory) to apply a given policy selectively to a segment of the population, unless of course the segment were the explicit target of the policy (think maternity leave for mothers only, or min wages for the low-income only).
This doesn’t rule out limited policy trials or A/B testing, but this works well when there are opt-in (or out) clauses. Still, policymakers must always be mindful of the risks of involuntary exclusion/inclusion, which can result in perceptions of unfair or biased treatment.
Such bias doesn’t just offend our sense of fairness. It can also invalidate any inferences we draw about the efficacy of a policy, because--unlike atoms or chemicals—humans possess agency & can modify their behavior in response to a policy when they are aware of how it’s applied.
The role of parliament is to debate the most desirable policies to adopt. Ideally, MPs consider both the interests of their constituents and the country as a whole when they choose to support or oppose a proposed policy. Policies are evaluated on their broad merits and demerits.
The devil, of course, is in the details. Who should furnish such details? The policy’s proposers, surely, but up to a point. Broad considerations like the potential impact on the fiscal budget, effects on the overall economy & society, winners & losers should be well understood.
The finer details should be left with a capable and apolitical bureaucracy. These are nonpartisan professionals, expected to translate the general into the specific and practical. In some democracies, only top appointments in the civil service change with govt, to align the two.
Because social science is imprecise and country- and situation-specific, all policy should be evaluated. The gold standard of evaluation is a randomized trial (like for medicine), but doing so for national policy is often impractical, since it treats some people as guinea pigs.
Thankfully, the science of statistics has advanced sufficiently that we can assess a policy’s impact with tools that are less discriminatory. A modern civil service should have professionals that are trained to use such tools, in service of nation.
Ideally, independent assessments (by academics, say) are also undertaken (which requires data sharing when in the public interest). Assessments of efficacy should then be fed back to parliament to determine if policies are worthwhile, and where they may have failed.
