Skip to main content
Momentum Actuarial

Methodology

How our vendor analysis works

When a health plan, an employer or an advisor weighs a health-care vendor, most of what it knows comes from the vendor. Our vendor analysis gives an independent read before you commit. Its value is not more data. It is curated evidence, with each fact sourced and dated and the vendor's claims labeled and kept out of our estimate, and written judgment: an actuarial method, applied the same way on each report and tested against judged cases, that decides what the evidence means for you.

01

Why this analysis exists

Most of what a buyer learns about a vendor comes from the vendor, and the public record rarely holds an independent test of its savings claims. The vendor analysis gives buyer and vendor a shared set of facts and says what the independent evidence supports: confidence before you commit. It is step 3 of four, and it is available now.
Where the vendor analysis fits

  1. See the problem

    Condition maps

    With early customers

  2. Find the options

    Options guides and comparisons

    With early customers

  3. Check the vendor

    Vendor analysis

    Live

  4. Measure results

    Results measurement

    Coming soon

02

What a report is, and what it is not

A report analyzes one capability of one vendor through one buyer profile: one of eight generalized profiles, or your own population's figures. No claims data is needed. It has five parts, each answering one question.

A report does not tell you whether to buy, rank the vendor or compare it with others. It shows where to look hardest, and the decision stays with you.

The five parts of a vendor analysis
1The factsWho is this company?
2The evaluationDoes the evidence hold up?
3The scrutiny levelHow closely should we look at this vendor?Light · Moderate · Elevated, with a flag beside it
4The savings rangeWhat could this save a plan like ours?
5The questions to askWhat do we still need to know?

03

Our approach: curated evidence and written judgment

Many people can see a patient's lab results. The value is in knowing which results matter and what they mean for this patient. A vendor's materials and the published studies are largely open to anyone who looks; the work is in deciding which count as evidence and what they mean for you.

We are AI-enabled software that applies the same written method on each report. AI does the reading; scripts do the arithmetic; the method and the checks are actuarial.

Curated evidence and written judgment

Curated evidence

  • Each fact sourced and dated
  • Vendor claims labeled as the vendor's
  • Studies screened for a sound comparison

Written judgment

  • Written rules, applied the same way on each report
  • Tested against judged cases

The vendor analysis

one vendor, one buyer profile

Only the buyer or the vendor knows

Contract terms and fees · How members would enroll · Unpublished results

Listed in the report as questions to ask

Your decision

04

Curation: where the evidence comes from

Evidence comes from the vendor's own materials, labeled vendor-stated; from documents the vendor presents about itself, such as security attestations and filings; and from independent published studies. Each fact is sourced and dated. Each study is labeled by who paid for it, how strong its design is and how closely it matches the program, and earns credit in proportion. A study credits a saving only if it has a sound comparison: a randomized design, or a credible comparison group with its starting point shown.

Curation: where the evidence comes from

Vendor's own materials

Labeled vendor-stated

Claims about results: shown and labeled, kept out of the estimate

shown in the report, not used in our estimate

Public record about the company

Each fact sourced and dated

Independent studies

Graded: who paid, design, match

Sound comparison?

randomized, or a credible comparison group

Yes
Can credit a saving
No
Shape only

shapes the scenarios

The report

Where evidence is missing, the report says so plainly rather than filling the gap.

05

Judgment: how the evaluation works

The evaluation rates a fixed set of clinical, financial and operational questions, each with its reason and its sources. Each claim of effect is read for the usual pitfalls: members who choose to join, high-cost members drifting back toward average, before-and-after comparisons with no comparison group, and engagement that drops off. The scrutiny level, Light, Moderate or Elevated, rises with the number of questions rated adverse, and a serious adverse finding on a question that weighs most for your type of organization sets it to Elevated. A flag beside the level says whether independent evidence tests the vendor's savings; it is a flag, not a veto.

Judgment in the evaluation: the scrutiny level

From the evaluation

  • How many questions are rated adverse
  • A serious adverse finding on a question that weighs most for this type of buyer

Read alongside the level

How large and how reversible is your decision?

  • Light
  • Moderate
  • Elevated

Flag beside the level

Does independent evidence test the vendor's savings?

A plan-wide change that is hard to unwind deserves close scrutiny whatever the level says; the level is set from the evaluation's findings.

A flag is not a veto: one adverse rating on whether savings can be credited to the vendor does not raise the level on its own.

06

Judgment: how the savings range is built

The range starts from how the program would produce a saving. Each figure about you comes from your buyer profile, labeled with its version. The range is a grid: your design choices, from light to aggressive, against how members and providers respond, high, base and low; the headline is the middle package under the base response. Credit comes only from studies with a sound comparison, with each adjustment to your plan stated. Vendor claims are shown, not used. Real savings are kept apart from cost shifted to members, and figures are before fees. Where the design or the evidence doesn't support a number, the report says so in a plain section instead. Each estimate names a result in your own claims that would show it wrong.

The savings range: a grid, not a single number
Rows · Your design choicesColumns · How members and providers respond
HighBaseLow
Lightbroad, voluntary
Standardthe middle packageHeadline
Aggressivenarrower, firmer

Real savings

what the plan stops paying

Cost shifted to members

kept apart, not counted as a saving

Before fees

Each figure states its year and its assumptions.

Each figure names a result in your own claims that would show it was wrong.

07

How we keep it honest

We work for the buyer. No pay to play: we take no money from vendors, with no sponsorships, paid placement, or paid listings or rankings, and who requests or pays for an analysis doesn't change what it says. No figure shown to you is chosen from among runs. On each report, the savings are recomputed from their inputs by separate arithmetic and checked against the evaluation. Changes to the engine are scored against cases judged in advance before they are adopted, so the written judgment stays tested.

08

Limits

The analysis rests on the public record unless you add your own figures. The generalized profiles are illustrative. Savings are before fees. A released report is a snapshot of what was known on its date.

An analysis is a well-founded starting point for your decision, not a certified verdict.

09

Questions we hear

  • Can an analysis built on public information be trusted?

    For the first question a buyer faces, whether a vendor's claims hold up, the important evidence is public. The claims are in the vendor's own materials, and independent evidence on programs like this one, peer-reviewed studies and independent evaluations, is often published even when no study tests this vendor; where none does, the report says so, which a buyer should know. A vendor's private data is usually what its own return-on-investment figures rest on, and independent actuaries have pointed out that employers have good reason not to rely on those alone. What we add is the work of finding that evidence, sorting the independent from the vendor-stated, judging how closely each study matches the program, and applying the same written method to each vendor.

  • What can't public information tell you?

    Some things only the buyer or the vendor knows: contract terms and fees, how the buyer's members would actually enroll, and results the vendor hasn't published. We don't treat those as known. The report lists them as questions to ask; where an estimate needs one, such as how many members would enroll, it uses a stated assumption drawn from evidence on similar programs and names the buyer's own figure that would replace it. Savings are given as a range before fees, with a plain note instead of a number where the evidence doesn't support one. Each estimate names a result in the buyer's own claims that would show it wrong, so it can be tested with the buyer's data later.

Get started

See the method applied to a vendor you're weighing.

Run a vendor analysisBook a walkthrough

No pay to play: we take no money from vendors, and who asks for our work doesn't change what it says.