Pharma Brand Performance & KPI Drivers

Most brand reports are slow because teams have to pull data from many sources (CRM, marketing, access), fix mismatches, join them by hand, and then build charts and slide stories manually. Each cycle turns into three analysts reconciling numbers in Excel instead of spending time on what the numbers mean. This delay is why you often understand a 20% swing in TRx only weeks after it happens, when there’s little time left to react.

‍

An analytics platform with a shared semantic layer can automate the data joins, keep KPI definitions consistent across brands, and refresh core views on a regular schedule. On top of that, AI narrative generation can draft the first version of the commentary (what changed, where, and by how much) so teams focus on interpretation and decisions instead of production work.

‍

Tellius can combine structured metrics with unstructured context from field reports in Google Drive, Gong call summaries highlighting customer feedback themes, and competitive intelligence documents, generating comprehensive narratives that explain not just what changed but why, based on actual field intelligence rather than just numerical analysis.

Global brands have to deal with very different metrics and data standards across markets. U.S. teams track TRx and NBRx, European teams may use DDD (Defined Daily Dose), and emerging markets often report only units sold. Some countries provide data monthly, others quarterly, and the lag times vary. Platforms with semantic normalization can map all these local metrics into a common set of global KPIs while still preserving the original local definitions for in-country teams. This lets global leaders compare performance across 15 countries on a like-for-like basis without losing important local nuance.

Gross-to-net (GTN) calculations are complex because they have to model dozens of contract types, rebate tiers, and channel dynamics that all affect how much revenue you actually realize. Many organizations keep these GTN models in separate spreadsheets or offline tools, which leads to constant reconciliation headaches with the main BI environment.

‍

Tellius semantic layer can encode this GTN business logic directly in the platform, so net price realization can be analyzed in real time without relying on external spreadsheets. This makes GTN analysis more transparent, easier to maintain, and better aligned with the rest of your commercial analytics.

A semantic layer is a translation layer that turns business language into data queries. It’s the dictionary between English and SQL. In pharma, it stores shared definitions so everyone calculates metrics the same way: for example, NBRx excludes samples, adherence uses PDC (Proportion of Days Covered), and HCP deciles are defined by territory.

‍

When someone asks, “Show me share for new customers”, the semantic layer knows this means NBRx divided by class NBRx. This prevents teams from using different formulas for the same KPI and also enables natural language queries. Tellius semantic layer can be configured with pharma-specific business logic, hierarchies, and synonyms, so it understands that “scripts” means prescriptions and can return accurate, business-relevant answers.

Quantitative metrics show what is happening, but HCP feedback helps explain why and can act as a leading indicator. Track measures like Net Promoter Score (NPS), post-event feedback, and intent-to-prescribe surveys alongside sales data. Territories with higher HCP satisfaction scores often show better NBRx growth over time. Tellius can analyze free-text feedback collected by reps to surface common themes, including feedback captured in Gong call transcripts, forms, and documents, which can be ingested as unstructured data and analyzed alongside structured metrics in Tellius. If sentiment trends negative while sales still look stable, it’s an early warning sign of future decline and a trigger for proactive action.

‍

Comparing your performance to competitors or to the industry usually requires external data. Many teams use syndicated sources to see class-level totals and competitor market share.

‍

For example, you might learn that your brand has 12% TRx share, while Competitor A has 40% and Competitor B has 30%. This tells you your true position in the market, not just whether your own numbers went up or down.

‍

You can also benchmark quality metrics like adherence using published studies. If the industry average adherence is 50% at one year and your brand is 60%, it is a real advantage.

‍

To make benchmarks usable, show them directly in your reporting as reference points (for example, “Our adherence: 55% vs industry: 50%”) so teams can quickly see whether a 20% share means leadership, parity, or clear room to grow.

• An awareness problem usually means most HCPs don’t know or think about the brand at all. You see low brand recall in surveys, low digital or field engagement, and a very small group of prescribers doing almost all of the writing.
• An adoption problem shows up when HCPs know the brand but don’t use it much. In research or rep feedback, they might say they have concerns about efficacy, safety, dosing complexity, or they simply prefer familiar options.
• An access problem appears when HCPs are trying to prescribe but patients are not getting on or staying on therapy. You see many prescription attempts but low fill rates, higher prior auth denials, more reversals at the pharmacy, and patterns where prescribing survives only in regions or plans with better coverage.

‍

Tellius can line up promotional reach, prescription attempts, and actual fills to help you see which pattern dominates (for example, “high reach + low trials” pointing to adoption, or “high trials + low fills” pointing to access) so you know whether to fix awareness, clinical education, or payer strategy.

Conversational analytics lets a brand manager type or ask questions in plain language instead of writing SQL or waiting for a dashboard request. For example, they can ask “Why did prescriptions fall in this region last month?” and the system can automatically break results down by payer, specialty, or segment. They can then ask follow-up questions like “Show me the payer mix for that region” or “Compare this to last quarter” in the same flow.

‍

Because the system is built on a pharma semantic layer, it can interpret terms like TRx, NBRx, and prior auth using the company’s approved definitions—so managers don’t need to know database field names or ask analysts to translate their questions.

Non-technical users need three main things: a common business language, an easy way to ask questions, and guided paths for common analyses.

• A semantic layer encodes metrics like TRx, NBRx, share, and pull-through so brand managers do not have to know how tables join or which filters to apply.
• Natural language query lets them ask questions in their own words and refine the analysis with follow-ups instead of rebuilding reports.
• Guided, reusable workflows for typical brand questions (such as source-of-business, launch tracking, or access impact) provide a structured way to explore the data.

Tellius brings these ideas together so brand teams can answer “why” questions themselves while analysts focus on deeper modeling instead of repetitive ad-hoc cuts.

This drives brand analysts insane. You build a quarterly forecast using September syndicated Rx data, then in October the vendor "updates their methodology" and suddenly last month's TRx numbers are 8% different. Your forecast is now wrong, your leadership deck shows contradictory numbers, and nobody knows which version to trust.

‍

The problem is that syndicated data providers use proprietary projection algorithms that change without warning when they add or lose pharmacy partners. Without a platform that tracks vendor methodology changes, flags when historical data gets restated, and creates audit trails showing which version of the data was used for each analysis, brand teams waste days every quarter reconciling reports that were accurate last month but wrong today.

• Marketing says NBRx excludes samples and refills
• Sales Operations includes 30-day lookback
• Finance uses a 90-day new patient definition

‍

All three claim they're using "the standard NBRx metric", but they're calculating completely different numbers. This happens because most companies don't have a centralized business logic layer that enforces consistent metric definitions across teams. Everyone pulls their own data, applies their own filters, and creates their own version of truth.

‍

Analytics platforms with semantic layers solve this by storing a single, governed definition of every metric so "NBRx" means the same thing whether Marketing, Sales, or Finance runs the report.

Formulary restrictions mean brand teams really only have control over roughly 40% of prescriptions, but they’re still judged on 100% of the results. When access tightens, even great execution can look bad because formulary changes hide the true impact of the team’s work.

‍

Analytics that split prescriptions into “accessible” (where the brand is covered and can be influenced) versus “restricted” (where coverage blocks use) make this clear. This way, teams can focus on the battles they can actually win and set realistic expectations with leadership about what performance is truly achievable.

Traditional dashboards can show trends and period comparisons, but they usually can’t tell you what event actually caused a change like a formulary win, a competitor launch, or a new safety communication. Real root cause analysis means drilling several levels deeper to find the specific payer changes, physician segments, or geographic pockets that are driving the variance.

‍

AI-generated insights can improve this by quantifying how much each factor contributed (for example, X% from a formulary loss, Y% from fewer calls) and suggesting concrete interventions, turning variance analysis more actionable.

A real 360° brand view connects data from early development all the way through commercial performance. This includes:

‍

• clinical pipeline status and trial milestones
• regulatory timelines and label expansions
• patent expiry and loss-of-exclusivity dates
• market access coverage by payer
• prior authorization and step-edit rules

‍

It also needs to factor in Inflation Reduction Act (IRA) and other pricing policy impacts, plus the core commercial metrics like NBRx, TRx, and market share by segment. This unified dashboard also acts as an early warning system so teams can adjust strategy early.

Multi-touch attribution models look at all touchpoints to separate the impact of field and digital. You pull data on rep calls, emails, webinar attendance, and digital impressions, then analyze how they correlate with prescription lift while controlling for overlap.

‍

Advanced analytics might show that physicians with 3+ rep visits plus digital exposure write 20% more prescriptions than those with digital exposure alone, revealing channel synergy. Some teams also run A/B tests by region to isolate the effect of different channel mixes. The result is clear guidance like: “In Q3, growth was 60% field-driven, 30% digital, 10% other factors”, which you can use to adjust budget and resource allocation.

• TRx reflects the overall base of treated patients. NBRx is a better signal of how many new patients are starting on your treatment.
• If TRx is steady but NBRx is rising, it often means you are winning new patients but also losing some existing ones.
• If NBRx is flat while TRx declines, the issue may be adherence and persistence rather than demand at the top of the funnel.

‍

Industry research also shows that NBRx can lead TRx by several months, so watching NBRx closely gives brand teams an earlier warning of where TRx will go next.

On a weekly basis, it is more useful to watch leading indicators than to chase every small move in TRx. For example,

• NBRx trends
• the number of new writers
• activity with high-value HCPs
• early signals from key payers

‍

On a monthly basis, you can add deeper KPIs like

• market share
• source of business (new vs switch vs add-on)
• key access metrics (approval rates or pull-through on recent formulary wins)

‍

You also want to track measures of sustainability like persistence and refill behavior, even if they move more slowly. The key is to limit the dashboard to a small set of KPIs that clearly link to decisions, instead of overwhelming the team with dozens of charts that no one acts on.

Prescription performance usually moves because of a mix of four forces: promotion, access, competition, and the patient experience.

• Promotion covers things like call activity, digital campaigns, and peer programs that influence awareness and consideration.
• Access covers formulary status, prior authorization rules, and patient out-of-pocket costs, which decide whether prescriptions can be filled and kept.
• Competition includes new entrants, label expansions, or safety events that shift physician confidence and habit.
• The patient experience includes effectiveness, side effects, dosing convenience, and follow-up support, which affect whether patients stay on therapy.

‍

Key driver analysis in Tellius helps brands separate which of these forces is most strongly linked to recent changes so they can focus their response where it matters.

When TRx (total prescriptions) drops suddenly but reps are still working hard, the cause is usually outside the rep’s control. Common reasons include:

• Insurance changes: a big plan may move your drug to a worse tier, add prior auth, or start preferring a competitor, so fewer scripts get filled.
  
• Generic launches: once a cheaper generic is available, pharmacies often switch patients at the counter.
  
• Safety worries: new safety signals, warnings, or even rumors can make doctors pause or slow down prescribing.
  
• Supply or distribution problems: if wholesalers, specialty pharmacies, or local pharmacies can’t get stock, prescriptions don’t turn into fills.

‍

Analytics helps by showing where the drop is happening (by payer, region, or pharmacy type), so you can quickly see which of these is the most likely cause.

When your brand moves up or down, you want to know: is it the doctors, or the payers? If it’s mainly HCP behavior, you’ll see changes across many payers at once:

• more or fewer HCPs writing
• changes in how many patients they start
• shifts in depth of use, no matter which plan the patient is on.

‍

If it’s mainly payer decisions, the drop is more targeted:

• volume falls mostly in certain plans or benefit types
• you see more denials, reversals, or abandoned fills in those payer segments

‍

By comparing prescription attempts, fill rates, rejection reasons, and patient out-of-pocket costs across payers, you can see whether doctors still want to use the brand but patients can’t get on therapy.

Traditional BI tools mostly answer questions that users already know to ask, through fixed dashboards and scheduled reports. An agentic analytics platform watches the data continuously and looks for unusual changes on its own, then drills through geography, specialty, payer, and competitive dimensions to find likely causes.

‍

In a brand review, this means the system can show up with a ready explanation of last month’s share shift by market or segment. Agentic workflows can also chain these steps together:

• from detecting a change
• breaking it down
• checking key drivers and history
• summarising the findings in plain language

‍

In addition, they also learn over time which patterns your teams care about most and prioritising those alerts.

Formulary decisions are often made months before they actually go live, but most teams only react after they see TRx change. Predictive analytics can change this by using signals like P&T (Pharmacy & Therapeutics) committee meeting schedules, competitor contract timing, and historical decision patterns to forecast likely formulary wins or losses. With these forecasts, teams can train the field force early and adjust inventory before access changes hit prescriptions, instead of scrambling after the fact.

Building predictive models in-house requires data scientists, infrastructure, and ongoing maintenance, which many organizations don’t have the capacity to support. Buying a commercial solution is usually faster to deploy, but out-of-the-box models may not fully reflect your specific market dynamics or connect easily to your proprietary data.

‍

Tellius offers pre-built pharmaceutical forecasting models that can be customized with your own data, giving you a balance between speed to value and company-specific accuracy.

Yes. To see which prescribers are switching away from your brand, you need to track their behavior over time, not just look at aggregate reports.

‍

Longitudinal, patient-level analysis shows when a physician stops starting new patients on your drug or begins moving existing patients to a competitor. Most organizations only see roll-up trends, so they miss these early warning signs. Analytical platforms with patient-journey analytics can flag prescribers whose behavior is changing and mark them as “at risk”, so the field team can intervene before the switching pattern accelerates.

Yes. Leading indicator analysis looks at early signals (such as samples requested, speaker program attendance, and formulary discussions) to forecast future prescription trends and KPI performance. Traditional statistical models often struggle with the non-linear relationships and many interactions in pharmaceutical markets, so they miss patterns that show up before a KPI slips.

‍

Machine learning models in platforms like Tellius can learn complex relationships between these leading indicators and lagging outcomes (for example, TRx 60-90 days later) and then forecast which KPIs are likely to miss target, with confidence intervals around the predictions. Tellius Kaiya can automatically build these models, detect anomalies when leading indicators start to behave differently from patterns that usually lead to hitting targets, and run what-if scenarios to show how different actions could change the outcome. Together, this gives brand teams 60-90 days of advance warning so they can adjust tactics before KPIs actually miss their goals.

Link your formulary data with your prescription data at the payer/plan level.

• Formulary data tells you which plans cover your drug, what tier it’s on, and what restrictions apply (prior auth, step edits).
• Prescription data tells you whether that coverage is actually turning into NBRx.

‍

Don’t just count covered lives. Measure conversion, using a metric like “NBRx per 1,000 covered lives” for each payer. This shows which formulary wins create real volume versus wins that look good on paper but don’t move prescriptions.

‍

After the win, track that payer’s NBRx share month by month. If there’s little improvement after about 3 months, it usually means uptake is blocked, often because physicians don’t know about the access change or the plan’s restrictions are still causing friction. This is where targeted pull-through programs should focus.

TRx usually moves last; by the time it drops, earlier signals have already changed.

• To see trouble sooner, brand teams watch new patient flow first: NBRx, starter program enrollments, and sample use. If those start to slow, TRx will often follow.
  
  
• They also track access signals like prior auth approval rates, how long approvals take, and changes in patient out-of-pocket cost, because rising friction here often leads to future drops in fills.
  
  
• On the interest side, falling email open rates, fewer site visits, or less webinar attendance can show that HCP attention is shifting or a competitor is gaining mindshare.

‍

When you monitor these signals together, you can spot a possible decline early and investigate what’s going wrong, instead of waiting for the TRx curve to finally show the damage.

AI-based anomaly detection learns what “normal” looks like for your brand across regions, channels, and segments, taking into account known seasonality and typical fluctuations. It then flags patterns that do not fit this history, such as a sharper-than-usual drop in a specific payer segment or an unexpected spike in a certain specialty.

‍

Because the model can watch many combinations of dimensions at once, it can spot issues that might be missed in high-level dashboards. These alerts can trigger deeper analysis to understand whether the anomaly is tied to access, competition, or execution. This turns monitoring from a manual, retrospective exercise into something more proactive.

Yes. Driver analysis can show which factors have the biggest impact on brand performance.

‍

Traditional “driver trees” are often static PowerPoint diagrams. They might link calls → awareness → TRx, but they become outdated as soon as the market shifts.

‍

AI-based driver analysis is stronger because it can test many drivers at the same time such as promotion, access, competition, patient mix, and channel mix. It can also check how they interact. It does not look at one variable in isolation.

‍

The hard part is connecting the full chain: operational activity (calls, reach) → intermediate outcomes (awareness, trial) → business results (TRx, revenue).

‍

More advanced approaches go a step further. They estimate what happens if you change one driver while holding others steady. This gets closer to the real question brand teams ask: “If we adjust this lever, what changes?”

‍

In practice, this turns the driver tree into a living model that updates as new data arrives, especially when a semantic layer keeps metric definitions and relationships consistent without manual rework.

AI can scan across many slices of the business to find segments where there is a clear gap between potential and actual performance.

• For HCPs, this might be physicians who see many eligible patients but use the brand less than peers with similar access.
  
  
• For payers, it could be plans where coverage is relatively favorable but pull-through is weak compared to other plans with similar benefits.

‍

Models can rank these opportunities by expected impact, so teams see where an extra push is most likely to generate new starts or share gain.

Leadership wants to know why the forecast was wrong, but most BI tools can't decompose forecast error into root causes. Was the miss because

• your syndicated prescription data coverage shifted mid-quarter (data quality), or
• the team assumed 12% market share growth when the category only grew 3% (bad assumptions), or
• a competitor launched mid-quarter (market changes)?

‍

Traditional dashboards show you missed by 40% but can't tell you which of the 15 forecast inputs was wrong. Platforms with forecast variance analysis can automatically compare your assumptions against actual outcomes across market share, pricing, formulary access, and competitive dynamics, showing you which assumptions broke and by how much. Without this, brand teams argue for weeks about whose fault it was instead of fixing the forecasting model.