Modeling lifetime value for a prescription brand

Two patients start the same brand at the same point in their disease. One starts in year three of the exclusivity window, the other in year eight. A subscription model prices them identically. In practice the second is worth a fraction of the first, because her therapy runs into loss of exclusivity and, at the next refill, a pharmacist substitutes an AB-rated generic without calling anyone. Every LTVLTVLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → number you quote for a prescription brand answers two questions at once: how long does the patient stay on therapy, and how long does the molecule stay yours?

This lesson builds that number from four inputs: duration of therapy, refill decay, net price after rebates, and the hard stop at patent expiry.

Why the subscription formula fails here

The formula most marketers learn:

LTV = Average Revenue per User × Gross MarginGross MarginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition → % × Average Customer Lifespan

It assumes smooth exponential churn, a stable price, and a lifespan that ends when the customer decides to leave. Prescription brands break all three. Persistence falls in steps tied to clinical and access events, the shape the benchmarking lesson lays out by therapeutic area. Revenue arrives at refills, so a 28-day specialty injectable and a 90-day mail-order oral with identical annual value produce very different cash timing and very different exposure to a mid-year drop-off. And the tail gets cut by a date on a patent, which no amount of adherence support moves.

Step 1: duration of therapy as a curve, not an average

Persistence data comes from longitudinal pharmacy claims (IQVIA and Symphony Health both sell the data under discussion here) and is read as a Kaplan-Meier survival curve. Use the 12-month persistence your therapeutic area actually supports rather than a portfolio average; the benchmarking lesson explains why two chronic categories can sit an order of magnitude apart.

Model time-bucketed survival instead of a single lifespan:

One thing a pooled curve hides: the January effect. Deductibles reset, co-pay assistance accumulators restart, and abandonment at the pharmacy counter spikes in the first weeks of the calendar year. A cohort that started in October carries that shock at month three; a February cohort carries it at month eleven. Pool them and you get a smeared curve that misprices both. If volume allows, build the curve by start month.

Step 2: attach revenue to the refill cycle, not the calendar month

Revenue hits at each fill, not continuously. For a monthly-refill specialty biologic with net revenue of $1,800 per fill (illustrative post-rebate estimate; real net prices are confidential and vary hugely by payer contract), multiply revenue by the probability the patient is still on therapy at each fill point.

Simplified worked example:

Assume:

• Net revenue per fill: $1,800
• Refill cycle: monthly
• Persistence curve as above (linear interpolation between points)

LTV over 12 months ≈ sum of (net revenue per fill × persistence probability at that fill)

• Months 1 to 3, average persistence ~89%: 3 × $1,800 × 0.89 ≈ $4,806
• Months 4 to 6, average persistence ~71%: 3 × $1,800 × 0.71 ≈ $3,834
• Months 7 to 12, average persistence ~58%: 6 × $1,800 × 0.58 ≈ $6,264

12-month LTV ≈ $14,900 (before gross-to-net adjustments, co-pay assistance costs, or program costs)

The naive version, $1,800 × 12 = $21,600, overstates by 31%. On a brand adding 4,000 starts a year, that gap is roughly $27M of imaginary value sitting inside a budget request.

Step 3: layer in PSP retention as a controllable variable

Patient support programs (manufacturer-run nurse support, co-pay assistance and refill coaching) move the persistence curve rather than sit beside it. Enrolled patients often show meaningfully higher persistence than non-enrolled patients on the same drug, sometimes cited as a 10 to 20 percentage point improvement at 12 months (estimate, varies with program design and disease state). Selection bias inflates that gap: patients who enroll are already the ones who answer the phone. Any lift you put in a model should come from a matched comparison, not a raw enrolled-versus-not split.

Baseline 12-mo persistence:        52%  →  LTV ≈ $14,900
With PSP (+10 pts persistence):    62%  →  LTV ≈ $17,600  (illustrative)

PSP cost per enrolled patient (estimate): $600–$1,200/year
Incremental LTV from PSP: ~$2,700

At $900 per patient for ~$2,700 of incremental LTV, program cost belongs on the same line as the acquisition spend the sibling lesson prices per new patient start, not in an overhead bucket.

For persistence and adherence methodology, the CDC's guidance on medication adherence measures and claims-based persistence studies indexed on PubMed are solid starting references.

Step 4: adjust for gross-to-net, and remember net price decays

Net revenue per fill is never list. Gross-to-net (GTN) describes the erosion from wholesale acquisition costacquisition costCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → down to what the manufacturer keeps after PBM rebates, payer discounts, co-pay assistance, 340B and Medicaid pricing, and distribution fees. For US branded specialty drugs, GTN erosion commonly runs 40% to 50% or more off list (widely cited industry estimate, varies by class and negotiating position).

The trap is treating GTN as a constant. Rebates deepen as competitors enter a class and as a brand ages into contracts it needs to defend, so a molecule in year seven can net materially less per fill than the same molecule at the same list price in year two. Under the Inflation Reduction Act, Medicare negotiation adds a second scheduled step-down well before patent expiry, and it lands earlier for small molecules than for biologics (negotiated prices take effect at nine years from approval versus thirteen). A brand with heavy Medicare exposure should carry a declining net price by year in the model.

In Europe the mechanics differ: direct negotiation or reference pricing with national systems (Germany's AMNOG, France's CEPS) sets a net price up front rather than eroding it through rebates. Less volatility, lower absolute per-patient revenue.

Step 5: the hard stop, patent expiry and substitution

Duration of therapy is bounded by the patient. Duration of revenue is bounded by the calendar. For small molecules the two rarely meet gently. Pfizer's Lipitor lost US exclusivity in November 2011 and most of the prescription volume moved to generic atorvastatin inside a year; Teva and the other large generic makers do not need to persuade a single physician to take that volume, because state substitution laws let the pharmacist swap an AB-rated generic at the counter unless the prescriber writes dispense-as-written. Years of HCP preference give you almost no protection there.

Practical consequence for the model: set the horizon T to the smaller of expected therapy duration and months remaining to loss of exclusivity, then value post-LOE fills at near zero for a small molecule. A 30-month therapy relationship worth roughly $30,000 collapses to about $14,900 if LOE falls at month twelve. That makes the LTV of a new patient start a function of the launch year, and it should change what you are willing to spend on acquisition in the last two years of exclusivity.

Biologics decay more slowly. Biosimilar substitution is not automatic unless the product is designated interchangeable and state law permits it, so erosion tends to be a multi-year slide rather than a cliff. That creates room for franchise moves: Alexion converted Soliris patients to Ultomiris, dosed every eight weeks instead of every two, well before the first US Soliris biosimilars were approved in 2024. A Soliris patient's LTV in 2021 therefore included option value that brand-level modeling would have missed, so model the franchise, not the pack. Merck has been public for years about preparing for Keytruda's late-decade US exclusivity loss, including a subcutaneous formulation, which is the same arithmetic on a much larger base.

Putting it together: the pharma LTV formula

$$

LTV = \sum_{t=1}^{T} (\text{Net Revenue per Fill}_t \times \text{Persistence}_t) - \text{PSP and Access Support Costs}

$$

where persistence comes from real claims data, net revenue per fill declines over the brand's life rather than holding flat, and T is capped by months remaining to loss of exclusivity.

🎬 [VIDEO: "How Pharma Companies Use Real-World Data" - youtube.com - search for IQVIA or Definitive Healthcare explainer content on claims-based patient journey analytics, illustrating how persistence data is sourced in practice]

Key Takeaways

• Never use flat subscription LTV formulas for chronic therapies. Persistence declines in steps tied to clinical and benefit-design events, and pooled curves hide the January deductible reset.
• Anchor value to refill cycles and net price. GTN erosion of 40%+ for US specialty brands (estimate) makes list-price modeling useless, and net price itself decays with class competition and Medicare negotiation.
• Cap the horizon at loss of exclusivity. For a small molecule, post-LOE fills are worth close to nothing once pharmacists can substitute at the counter, which makes a new start's value depend on the year it happens.
• Treat patient support as an LTV lever with a measured lift, using a matched comparison rather than the raw enrolled-versus-not gap.
• Model the franchise where a successor product exists: Alexion's Soliris-to-Ultomiris conversion changed the value of a patient long before biosimilars arrived.