Insights

Open Insights -- in the sidebar under Markets.

Insights is a question picker. The left pane lists 41 prepared questions, grouped by the job you are doing, with a Search insights... box; the right pane runs the selected ranking and shows the answer, why it matters, and a short narrative of what stands out.

Every question is a real query against the Medicare public files, not a saved screenshot. Changing scope re-runs it.

Question categories

Question kinds

Behind the groups, questions run in seven different ways:

• Ranking -- a straight sort on one field. "Who writes the most prescriptions overall."
• Ratio -- a computed share, markup, per-patient or prevalence figure over the highest-volume providers, with a minimum denominator so tiny panels do not top the list. "Brand reliance" counts prescribers with at least 2,000 claims; "charge aggression" needs at least 100,000 USD in allowed amount. The chronic-condition questions rank by estimated patients (panel prevalence applied to panel size) and count individual clinicians only, leaving out labs, pathology, radiology and other non-treating billers.
• Composite -- a weighted multi-factor score. The diabetes launch targets, for example, combine diabetes prevalence (weight 3), brand spend (weight 2) and patient complexity (weight 1). Use Tune the weights to re-rank to move any factor between 0x and 5x; Reset restores the defaults. Custom weights are kept in the page link, so a copied link opens with the same weights.
• Savings -- generic-substitution sizing. Pick a brand; Qope prices it against its generic equivalent in NADAC and ranks prescribers by the dollars a switch would recover. The header shows the brand and generic price per unit and the total recoverable across the ranking.
• Shortage exposure -- every drug FDA currently lists as in shortage (discontinuation notices excluded) is matched to its Part D names, and prescribers are ranked by their combined claims for those drugs over the 500 highest-claim prescriber lines. Each row names the shortage drugs that prescriber writes most.
• Patent cliff switch targets -- the same prescriber scoring as the switch-target board on Patent Cliff: Part D spend on brands whose protection ends in the window, weighted by how open the generic market is and how soon it opens. Shown as a score where 100 is the #1 account; each row names the brands and when the first one opens. National only.
• New clinicians by state -- individual NPIs first issued in the last six months of NPPES data, counted per state. Each row opens the Providers table filtered to that state and window. National only.

Scope controls

Questions over provider data show a Peer group bar:

• Drug -- drug questions only. When you open one without a drug, Insights starts on the brand behind the largest single-prescriber Part D spend lines and says so; pick any drug to re-rank. The generic-switch question still asks you to pick a brand. Choose a drug opens a searchable list with a Brand / Generic toggle. The generic-switch question takes brand names only.
• Specialty -- All specialties or one specialty. The list is fetched live from the dataset the question runs against.
• State -- All states or one state.
• Patient panel -- Any patient panel, or providers whose panel carries a condition, with a threshold. See below.
• Clear resets state, specialty and patient panel.

Facility, laboratory, ACO, patent cliff and new-clinician questions have no scope bar.

If your active cohort has exactly one state or exactly one specialty, Insights applies it when you open a question. Clearing the scope turns the cohort off for that view.

Patient panel filter

The patient panel filter lets you ask "among providers whose panel is at least 30 percent diabetic, who spends the most on brand drugs?" It is available on questions built on the Medicare provider summary, not on Ask-by-Drug questions.

Conditions: diabetes, cancer, ischemic heart disease, heart failure, atrial fibrillation, chronic kidney disease, COPD, asthma, Alzheimer and dementia, depression, anxiety, schizophrenia, osteoporosis, arthritis, stroke, hypertension, hyperlipidemia, Parkinson, tobacco use, alcohol and drug use.

Thresholds: 10, 20, 30, 40 or 50 percent of the panel ("30%+ of panel"). Default is 20.

Prevalence-based questions require a panel of at least 50 beneficiaries, so a provider with a handful of patients cannot show 100 percent prevalence.

If the panel filter cannot be applied, Insights says so and shows no ranking, rather than an unfiltered one.

Results

The header line says what you are looking at -- how many rows are shown, out of how many providers in scope, and what they are ranked by -- and, where a peer sample is available, the Typical peer value: the median of a sample of providers in the same scope.

Each row shows rank, full name, state, specialty and value, and where available:

• Percentile -- where the row sits among everyone in scope, shown as "Top N%".
• Typical-peer multiple -- how many times the typical peer the value is ("Nx typical peer"), shown at 1.5x and above.
• Phone -- the practice phone from the NPI registry, on provider questions other than the scored target lists.
• Peers -- reruns the same question for that row's specialty in that row's state, so you can see who else looks like this account.

Above the rows, a short summary calls out what stands out: how far #1 leads #2, how much of the list the top 3 hold, the state and specialty that dominate, and ratio outliers worth checking before outreach. On scored target lists it shows only while the weights are at their defaults.

Each row links to the record page (provider, facility or ACO). The default result limit is 20 rows.

Below the ranking:

• Export -- CSV of the rows shown: rank, name, NPI, state, specialty, phone, value, percentile, typical-peer multiple and, for composite questions, each factor score.
• Save as lead list -- saves the ranking as a new lead list and opens it. The dialog proposes a name from the question and scope and can share the list with your team. Available on provider, facility and ACO questions; these come from Pro datasets, so saving needs a Pro or Teams plan.
• View full table -- opens the source table with the same state, specialty, drug and sort. Shortage and patent cliff questions link to Open Shortages and Open Patent Cliff instead. A patient panel filter is not carried over, and the link says "(panel filter not applied)".
• Open in Map -- when a state is set, opens the Map on the Providers dataset with that state selected (Pro and Teams).

If part of the candidate pool could not load, a Partial ranking notice warns that some providers may be missing.

Limitations

• Medicare public files are aggregated and lagged. They describe Medicare fee-for-service and Part D activity, not total practice activity, and not commercial payers.
• The Providers and Prov Summary tables use different specialty vocabularies (for example CHIROPRACTOR versus CHIROPRACTIC). A specialty chosen in one context may not exist in the other. Insights always fetches its specialty list from the dataset it will actually query, but be careful when carrying a specialty value between pages by hand.
• Ratio and composite questions rank a pool of the highest-volume providers, not every provider in scope.
• Opening a question uses one credit. See Usage and Limits.