Confidential · Prepared for Insperity leadership

The revenue is intact.
The cost signal failed.

Insperity is still a $6.8B-revenue PEO with 99% monthly client retention. What collapsed isn't the business — it's the ability to see a benefits-cost trend forming before it detonates the margin. That is a data problem. It is the problem Talon's healthcare Data Lake was built to solve.

Annual revenue
$6.8B

Flat-to-up. Client base steady at 309,000 worksite employees, 99% monthly retention.

vs
Market capitalization
~$1.4B

Down from a $5–6B peak. The market repriced an earnings collapse, not a shrinking company.

362M
Rows
76
Sources
132
Datasets
9.5M
Providers
6 yr
Cost depth
02 — The Diagnosis

A margin event, not a market-share event.

Insperity pools 309,000 worksite employees and their dependents into one risk-bearing health plan and prices that benefits risk to clients in advance. When real claims outran the forecast, Insperity absorbed the gap — and 2025 closed in the red, a full-year net loss for a company that earned $171M just two years earlier.

9.1%

Q3 2025 benefits cost trend — against a budgeted full-year range of just 4.5–6%.

−51%

Full-year 2025 adjusted EBITDA, down to $131M. The year closed at a $7M net loss — against $171M of net income in 2023.

+1%

Worksite employees, up to 309,327. The customers stayed. Retention held at 99%/month.

200–400

Basis points the industry claim trend ran above start-of-year estimates — an unexpected surge that emerged mid-year.

The cost drivers — named by their own CFO and CEO

01
Prescription drugs & new high-cost treatments. Increased pharmacy use, including the introduction of new higher-cost drugs.
02
Outpatient & inpatient utilization. Higher than expected, outpacing the favorable demographic shifts they were counting on.
03
Large-claim frequency. A significant increase, both sequentially and year-over-year.
04
Provider AI tools. Newly flagged — AI in diagnosis, documentation, coding, and preauthorization is inflating billed cost trends.

The common thread isn't any single driver. It's that the trend emerged during the year and they didn't see it coming. They expected it to taper in the second half. It didn't.

03 — The Blind Spot

They are downstream of their own carrier's data — on a lag.

Insperity's view of where healthcare costs are heading comes from its carrier and outside advisers, surfacing in claims data after the money is spent. There is no independent, forward-looking line of sight into the drug, procedure, and provider trends building in the broader market. By the time the signal reaches them, the margin damage is already booked.

The tell is in their own filings: Insperity carries an independent actuary for workers' compensation — but forecasts health claims internally, as a management estimate built on carrier-fed data, flagged in its own 10-K as a critical audit matter because it is "subjective and judgmental." Management called the 2025 claims runoff "unprecedented" and conceded that "predictability has been affected." They are structurally dependent on the one party whose data they most need to challenge.

The drivers that hit them — drug adoption curves, procedure-volume shifts, the provider-billing patterns behind large claims — are externally observable at NPI, drug, and procedure resolution long before they clear as commercial claims. That observable layer is the Data Lake.

04 — What Talon Sees

Four capabilities, mapped to the four drivers.

An independent healthcare-cost intelligence layer that sits outside the carrier relationship — forecasting the trends that blindsided them, at a resolution no commercial database publishes.

Driver 01 · Drugs

Drug-cost wave forecasting

Six years of Medicare Part D at NPI × drug × year — 153.8M rows. The adoption and cost curve of every drug class, visible as it forms. FDA approval pipelines flag the next wave before it lands in plan costs.

Maps to → prescription-drug surge
Driver 02 · Procedures

Utilization-trend tracking

Medicare Part B at NPI × HCPCS × year — 58.85M rows. Outpatient and procedure volume shifts tracked as they accelerate, including the procedure categories that drive large claims.

Maps to → outpatient / inpatient utilization
Driver 03 · Pricing

Commercial-rate benchmarking

Hospital Price Transparency files — actual commercial negotiated rates by hospital, payer, and procedure, joined to a Medicare baseline. Validates whether a carrier's negotiated rates in a market are competitive, and where to steer.

Maps to → large-claim cost basis
Driver 04 · Billing

Anomalous-billing detection

Six-year longitudinal provider billing profiles surface upcoding and outlier utilization at NPI level — directly addressing the provider-AI coding inflation Insperity just flagged and has no tool for. The Peterson Health Technology Institute ties AI-driven coding intensity to approximately $2.3B in added healthcare spend.

Maps to → provider AI / coding inflation
Proof — the GLP-1 wave, already quantified
405×
Medicare Part D cost ramp · Ozempic · 2018 → 2023
$21M
'18
$404M
'19
$1.2B
'20
$2.2B
'21
$4.0B
'22
$8.5B
'23

A single drug, traced from $21M to $8.5B in Medicare spend at single-prescriber resolution. This is the exact class of cost wave that drove Insperity's pharmacy trend — and the lake saw the curve forming years before it peaked. The same method forecasts the next wave: new obesity agents, Alzheimer's drugs, cell and gene therapies.

05 — "But we just fixed this with UnitedHealthcare"

The November 2025 deal helps. It does not close the gap.

Insperity extended UHC through 2028 and dropped the large-claim pooling level to $500K from January 2026. Real progress — and three reasons the intelligence layer still matters more, not less.

Risk transfer isn't free

UHC priced that $500K pooling cap. Insperity's margin now hinges on whether the transferred tail risk was fairly priced — and they have no independent data to validate it against.

They still own everything under $500K

The 9.1% trend is overwhelmingly high-frequency, mid-cost claims — pharmacy, outpatient, GLP-1s — and the pooling cap barely touches it. At a $500K cap every large claim they can predict and reprice is now pure margin. Prediction is the lever; they have no engine for it.

Forward pricing still failed them

A reactive, carrier-lagged feed is exactly what broke. Pricing clients forward — accurately — needs a forward-looking, independent signal. That is the gap the deal does not fill.

06 — Prevent / Recover

How the lake answers the two questions that matter.

The 2026 recovery is already underway — but it rests on a bet that repricing holds and the trend behaves. An independent cost signal turns that bet into a calculation, and makes the recovery durable instead of one favorable year.

Could it have prevented the collapse?
  • The collapse was a forecasting failure — they were blindsided by a trend the market was already showing.
  • An external signal tracking pharmacy, outpatient, and large-claim drivers at NPI / drug / procedure resolution surfaces an accelerating trend earlier than carrier-lagged claims.
  • That lead time is the difference between repricing once and absorbing a full year of compounding margin damage.
Can it help them recover?
  • Accurate forward pricing — anchored to observed market trend, not last year's claims.
  • Smarter plan design and member steerage, informed by commercial-rate benchmarking.
  • Independent validation that the UHC risk-transfer deal is fairly priced.
  • Early warning on the next cost wave before it reaches the plan.
07 — The Asset

A self-built healthcare data lake. Not a reseller feed.

362 million rows across 76 active sources and 132 datasets — CMS Medicare Part B & D, NPPES, Open Payments, FDA, HRSA, and a commercial price-transparency layer — unified and queryable at provider, drug, and procedure resolution. Six years of longitudinal Medicare depth means the lake doesn't just show today's costs; it shows the trajectory.

82M
Mfr payments
153.8M
Part D rows
58.8M
Part B rows
2018–23
Cost series
Honest scope

The lake is built on Medicare data (65+ / disabled) plus a commercial price layer. Drug and procedure adoption curves and provider behavior transfer as leading indicators of commercial trend; absolute commercial price levels come from the price-transparency layer, which is expanded per engagement. We position this as independent cost-trend intelligence alongside Insperity's actuarial data — not a replacement for it. That distinction is the credibility.

08 — Engage

The blind spot is observable.
We already see it.

A focused engagement: map the cost drivers behind Insperity's 2025 trend against the lake, benchmark the UHC deal, and stand up a forward cost-trend signal for the 2027 pricing cycle.

Start the conversation →
Talon AI · matt@runtalon.ai · runtalon.ai