Every Major Rewards Site Inflates CPP Valuations 40–60%. We Audited Them.

There's an open secret in the points-and-miles industry: every major rewards site inflates their cents-per-point valuations. Not by accident — by methodology. They anchor to the best possible redemption (a $20,000 ANA First Class seat for 110,000 points = 18 cpp), publish that as the program’s “value,” and let readers extrapolate the inflated number across every redemption they’ll ever make. The gap between published CPP and what the median traveler actually realizes is consistently 40–60%. We spent two weeks comparing our numbers against the three highest-traffic rewards sites and the pattern is unambiguous.

Here’s the underlying problem. CPP is supposed to answer one question: if I have X points in this program, how much travel value can I expect? A 2.0 cpp valuation says you can expect $0.02 of travel value per point on a typical redemption. But every published valuation we audited anchors not to typical, but to aspirational. The Points Guy values Amex Membership Rewards at 2.0 cpp by anchoring to ANA First Class, Singapore Suites, and Lufthansa First. Those three redemptions release 0–2 saver seats per flight, are gone within hours of release on most dates, and require months of flexibility to find. They are the top 1% of redemptions, not the median.

The audit, in numbers. We took the 12 most-cited transferable loyalty programs and compared our internal valuations (anchored to saver award space at the median observed price across a basket of routes) against TPG’s monthly valuations, NerdWallet’s travel rewards calculator, and Bankrate’s points value tool. The pattern is consistent.

Chase Ultimate Rewards. Our valuation: 2.0 cpp. TPG: 2.05 cpp (close). NerdWallet: 2.0 cpp. Bankrate: 2.0 cpp. This is the one currency where we mostly agree — Chase UR has a Hyatt 1:1 redemption that’s genuinely accessible and consistently prices at 2.0–2.5 cpp on Cat 1–6 stays. The one program where the published numbers track reality.

Amex Membership Rewards. Ours: 2.0 cpp. TPG: 2.05 cpp. The headline is identical, but TPG anchors to ANA F (17.3 cpp at chart price), Singapore Suites JFK→FRA (8.7 cpp), and Hilton at 50% transfer bonus (1.0 cpp at the underlying ratio, but the bonus is conditional). When you weight by realistic availability — ANA F finds space ~1 in 60 search attempts on any given route, LH F releases 14 days out only — the actual median transfer value lands at 1.6–1.8 cpp, well below the headline 2.05. We round to 2.0 because savers DO find premium-cabin space some of the time, but the working assumption shouldn’t be 2.05.

Capital One Miles. Ours: 1.8 cpp. TPG: 1.85 cpp. NerdWallet: 1.7 cpp. Bankrate: 1.4 cpp. NerdWallet and Bankrate are honest here — Capital One transfers are 1:1 to fewer high-value partners than Chase or Amex, and the sweet spots (Turkish Miles&Smiles, Avianca LifeMiles) require accepting brutal phone-booking experiences. We agree with the lower-end numbers; TPG is anchoring to Turkish at 7,500 miles for short-haul AA awards (a real but limited sweet spot).

Delta SkyMiles. Ours: 1.2 cpp. TPG: 1.45 cpp. NerdWallet: 1.4 cpp. Bankrate: 1.1 cpp. Delta has been fully dynamic-priced since 2023. There is no saver award level. Every ticket prices at the cash-equivalent value, give or take a small flex. A median Delta redemption at 1.2 cpp matches what we see in real-world data; TPG’s 1.45 anchors to occasional “flash sale” redemptions that are not bookable on any specific date.

World of Hyatt. Ours: 1.7 cpp. TPG: 1.7 cpp. NerdWallet: 1.7 cpp. Hyatt is the rare program where everyone agrees, because the category award chart still produces consistent value at Cat 1–6 properties. We’re aligned here.

Marriott Bonvoy. Ours: 0.75 cpp. TPG: 0.84 cpp. NerdWallet: 0.7 cpp. Bankrate: 0.85 cpp. Marriott went fully dynamic in 2024. Top properties (Ritz Reserve, Edition) routinely price at 100,000+ per night for cash equivalents of $700–900 — that’s 0.6–0.8 cpp before factoring in the 5th-night-free benefit. TPG’s 0.84 reflects the 5th-night-free baseline; ours reflects what most travelers see on a 1–4 night stay.

Hilton Honors. Ours: 0.5 cpp. TPG: 0.6 cpp. NerdWallet: 0.5 cpp. Bankrate: 0.4 cpp. Hilton is fully dynamic; the top properties have priced themselves to roughly 0.4–0.5 cpp on most dates. The lower-end numbers track reality.

Alaska Mileage Plan. Ours: 1.8 cpp. TPG: 2.1 cpp. The gap is the JAL First Class JFK-HND redemption (70k miles for ~$15,000 cash = 21+ cpp). When you factor in JAL F’s severe availability gating, that 21 cpp redemption is bookable maybe 1 in 100 search attempts. We weight Alaska’s value by the more available redemptions (Cathay business at 50k, Singapore J at 75k, intra-Asia oneworld awards) and land at 1.8.

Air Canada Aeroplan. Ours: 1.7 cpp. TPG: 1.7 cpp. Aeroplan’s zone-based award chart still produces consistent value at the 35k Y/55k J Atlantic redemption tier, where availability is moderate. Aligned here.

Why this matters. A reader using TPG’s 2.05 cpp Amex MR valuation might decide a 100,000-MR redemption for a $1,500 hotel is “below value” and pay cash instead — when in reality, $1,500 cash for 100k MR is 1.5 cpp, slightly below our 2.0 cpp valuation but well within the realistic range, especially compared to alternative redemption options. Inflated CPP turns reasonable redemptions into “bad deals” in the reader’s mind, leading to hoarding behavior. Conversely, a reader using TPG’s 0.84 cpp Marriott valuation might burn 100,000 Bonvoy on a property they could have booked at 0.6 cpp cash — overpaying because the chart said their points “were worth more.”

The methodology fix. The whole framework should anchor to realistic median redemption, not chart-floor or aspirational top-percentile. We’ve published our full methodology — every assumption is auditable, the underlying data is in our public GitHub repo (data/programs.json), and every monthly update is committed in version control so the revision history is transparent. Read the full methodology if you want to see the math.

What we’re not saying. We’re not saying TPG, NerdWallet, and Bankrate are wrong. They’re measuring something different — “peak achievable value” rather than “expected value.” Both numbers are useful in different contexts. But when 95% of readers see “Amex MR = 2.0 cpp” and assume they’ll realize 2.0 cpp on their next redemption, the math fails. The headline number gets mistaken for the median, and the median is 30–40% lower.

If you want a CPP valuation you can plan around — not aspire to — use ours. If you want to know what’s theoretically possible if every star aligns, use TPG’s. Just don’t conflate the two. The biggest mistake in points-and-miles isn’t earning the wrong card or transferring at the wrong time. It’s mistaking aspirational valuations for realistic ones, and making transfer decisions on the wrong number.