Steps where decisions are based on "guessing" (inadequate information) in contemporary drug discovery and development

Picking targets
Validating targets in a disease pathway
Differentiating from other therapeutic targets
Identifying targets with a big signal on the disease pathway with the human disease and predict clinical response for the target
Rapidly testing targets in humans
Choosing chemical class and best compound in class
Potency on target
Dose-response relationship
Off-target effects
Selecting animal models
Characterizing pathophysiologic basis of disease in preclinical models
Selecting preclinical models that share underlying pathogenesis with human disease
 Identifying preclinical models that do not predict clinical response
Establishing therapeutic proof-of-conceptDetermining whether target has effect on pathway in whole organism
Inexpensively comparing different targets in humans
Rapidly determining if agent has effect on target pathway in humans
Generating confidence to move forward with an agent in humans
Identifying the right patients
Identifying pathogenic subsets
Establishing treatment-responsive and –unresponsive subsets
Understanding patient variability of response and effect size
Determining dose-response relationships in subsets
Establishing the role of personalization in dosing and monitoring
Finding the best dose and regimen
Determining optimal study design, dose and regimen for Phase II studies
Making use of intensive studies with predictive measures
Generating accurate power calculations and statistics for Phase III studies
Establishing variability in patient
Determining within subject variation of pathway and modulators
Establishing between-subject variability and normal ranges in populations
Understanding effect size and variability of therapeutic response
Establishing interactions with approved agents in combination therapies
Generating experimental basis for powering Phase II and III clinical trials          
Anticipatingand confirming  toxicities
Identifying at-risk pathways for the compound or class
 Establishing dose-response relationships for therapeutic effect
Selecting predictive approaches beyond preclinical model
Translating preclinical toxicity findings rapidly into humans
Personalization to improve response
Determining if patient selection by pathogenesis increases response signal
Adjusting dose based on therapeutic effect on target pathway
Testing adaptive design clinical trial strategies
Deciding whether to get out early
If target is not hit by the compound or the class
If hitting the target does not impact the intended pathway
If altering the pathway does not alter the disease
If toxicities are too risky for the benefits observed