For a cardholder
Which card earns the most after its fee, for a household like mine?
[ Analytics case · independent project · 2026 ]
Which travel card actually pays?
Premium cards now cost up to $895 a year and promise “$3,000+ in value”. I put 6,666 real U.S. households’ spending on 11 cards (eight classic travel cards, plus Bilt’s three) for a year to see who actually comes out ahead — and why.
- Question
- For a real household, which travel card earns the most after its annual fee — and what decides it?
- What I did
- Turned BLS survey microdata into a year of spending in 8 card categories, encoded 11 cards’ current terms (8 classic + 3 Bilt), and simulated every household under two ways of using points, then stress-tested coupon habits.
- Answer
- Among the 8 classic cards: cash out your points and a no-fee 2% card wins for 95.5% of households. Transfer them well and the $95–$395 cards win. The $795–$895 cards win for about 3% — unless you use 70%+ of their coupons.
- 2026 addendum
- Add Bilt and it becomes the best card for 91% of households that transfer points — mostly because of rent points and Bilt Cash, not everyday earning.
- Skills
- Survey microdata · weighting · imputation · scenario & sensitivity analysis · unit economics · Python / pandas · interactive tool
- real households simulated
- 6,666
- best off with 2% cash back (8 classic cards, cashing out)
- 95.5%
- would pick Platinum if they used every coupon (8 classic cards)
- 69%
- best off with Bilt once it’s added (transferring points)
- 91%
01
The question
Premium cards sell “value”. I wanted to know how much of it real households can actually use.
Card reviews compare earn rates and list every credit at face value. But a $300 Resy credit is worth $0 to someone who never books Resy. I looked at it from both sides: the cardholder (which card pays the most after the fee?) and the issuer (how much of a premium card’s economics rests on value customers never claim — the industry calls it breakage).
For an issuer
How much does the premium proposition depend on coupons people don’t use?
For a product
A calculator that answers the first question in 30 seconds, with honest assumptions.
02
Data & model
Real spending, current card terms, and every assumption out in the open.
The U.S. Bureau of Labor Statistics interviews thousands of households every quarter about what they spend. I kept households interviewed at least twice in 2024, scaled their quarters to a year (so lumpy spend like flights reflects several real quarters), mapped 17 survey variables into 8 card categories, and weighted them to represent U.S. households. Then I encoded each card’s fee, earn rates, caps and credits as of October 2026.
- 016,666HouseholdsBLS CE Interview 2024, ≥2 quarterly interviews, annualised and population-weighted
- 028Card categoriesGroceries, dining, gas, airfare, hotels, transit, streaming, everything else
- 0311Cards8 classic travel cards + 3 Bilt cards: fees, earn rates, caps and credits as of Oct 2026, $0 to $895
- 042Redemption stylesCash out (simple floor) vs. transfer to partners (The Points Guy values)
net value = points × value per point + credits you would really use − annual fee
The average householdCard-payable spend · per year
How credits are countedAssumptions
| Any-travel creditup to your own travel spend | 100% |
| Portal-only creditmust book through the issuer | 80% |
| Luxury-hotel creditonly if you book hotels | 50% |
| Merchant couponsUber, Resy, Saks, streaming… | 35% |
What a point is worthCents per point
| Currency | Cash out | Transfer |
|---|---|---|
| Chase Ultimate Rewards | 1.00 | 2.05 |
| Amex Membership Rewards | 0.60 | 2.00 |
| Capital One Miles | 1.00 | 1.85 |
| Citi ThankYou | 0.75 | 1.90 |
One honest patch: the Interview survey doesn’t collect groceries (they live in the separate Diary survey). I imputed them per income quintile from USDA totals minus each quintile’s measured dining, scaled by family size. The result ($5,575 average) sits a little under BLS’s published $6,224 — conservative for grocery-heavy cards.
03
What the simulation says
How you use points matters more than which card you hold — and premium cards are a coupon decision.
Best card, by household shareWeighted · 6,666 households
A
If you cash points out, a no-fee 2% card is the best choice for 95.5% of households. Transfer points to airlines and hotels and the answer flips: Amex Gold 37.8%, Venture X 23.3%, Strata Premier 19.0%, Venture 12.5%.
Median net value per yearTransfer scenario
B
At realistic coupon use, the $795 Sapphire Reserve nets a median $214 and the $895 Platinum $296 — versus $811–$891 for the $95–$395 cards. Premium cards rarely win on spending alone.
If people used more of the couponsTransfer scenario
C
At 35% coupon use, Platinum is best for 0.7% of households. At 70%, 24.6%; at 100%, 69%. The premium card’s value proposition is built on breakage — value most customers leave on the table. For an issuer that is the business model; for a customer it is the question to ask before paying $895.
Travel decides the tier
- No travel54.6%of households Amex Gold
- < $1k14.8%of households Venture X
- $1–3k12.7%of households Venture X
- $3–8k10.6%of households Venture X
- $8k +7.2%of households Sapphire Reserve
Card with the highest total value for each travel tier (transfer scenario).
When does a premium card win?Average household · transfer
- Venture X
- Amex Gold
- Sapphire Reserve
- Amex Platinum
- Citi Double Cash
D
55% of households report no airfare or hotel spend at all — for them Amex Gold wins on groceries and dining. $1–8k travellers land on Venture X. Only around $10k of travel a year does the Sapphire Reserve pull ahead — that’s about 7% of households.
04
2026 addendum: Bilt
Three more cards and one new input, housing. The results for the eight classic cards above don’t change.
The original eight left out a popular newcomer: Bilt, with three cards (Blue $0, Obsidian $95, Palladium $495). Bilt pays points on rent and mortgage, so I added each household’s housing cost from the same survey: rent plus mortgage interest and principal, $9,225 a year on average. Bilt makes you choose between points on housing (0.5–1.25x, depending on how much you put on the card) and 4% back in Bilt Cash; the model gives each household the better one and values Bilt Cash like a merchant coupon (35%). Bilt points are valued at 2.0¢, below TPG’s 2.2¢, because Hyatt transfers are moving to 1:0.75.
Best card by share of households · 11 cardsWeighted · 6,666 households
E
Cashing out, the 2% card still leads with 56.1%, but Bilt Obsidian and Palladium take 43.2% between them. Transferring points, Bilt wins for 91.4% of households: Palladium 43.0%, Obsidian 42.5%, Blue 5.9%.
Where Bilt’s lead comes fromTransfer · Bilt’s share of households
F
Without rent points and Bilt Cash, Bilt is the best card for only 6.1% of households. On everyday spend alone it’s just another good card. Bilt Cash is the bigger lever: it makes Bilt the pick even for 89.7% of households with no rent or mortgage. If you’d never spend Bilt Cash, rent points alone still win for 49.1%.
If people used more coupons · 11 cardsTransfer · share of households
G
Bilt leads at every level. At 0% it drops to 49.1%, with Strata Premier (19.5%) and Venture X (11.4%) next. At 100%, the Platinum card comes back to 24.2%, but Palladium still takes 57.7%.
Bilt Cash is the big assumption
4% back sounds like 4%, but Bilt Cash only goes to Bilt’s own credits (Lyft, dining, hotels, rent fees) and mostly expires each year. I value it at 35%, the same as merchant coupons. If you’d use all of it, Bilt looks even better.
Why 2.0¢, not 2.2¢
TPG values Bilt points at 2.2¢, the highest of any currency. Hyatt, the partner behind much of that value, is moving to 1:0.75, so I use 2.0¢. That’s my own call, and the calculator uses it too.
Rent points assume one card
The top 1.25x rate needs card spend at least equal to your housing payment. If a household puts all its spending on Bilt, 83% of those with rent or a mortgage get there. Split your spending across cards and the rate drops to 0.5–1x.
05
Try it with your own spending
Same formula as the simulation, now with your numbers.
Set your monthly spend per category and your rent or mortgage (only Bilt uses it), choose how you use points, and say honestly how many merchant coupons you would actually use. All 11 cards, Bilt included, rank live.
Card labOct 2026 terms
$ / month
Points
Net value per yearYearly card spend $30,240
- 1Bilt Palladium$1,210 points + $390 credits + $423 rent − $495 fee$1,528
- 2Bilt Obsidian$880 points + $80 credits + $423 rent − $95 fee$1,288
- 3Venture X$1,119 points + $425 credits − $395 fee$1,149
- 4Strata Premier$1,154 points + $50 credits − $95 fee$1,109
- 5Amex Gold$1,271 points + $148 credits − $325 fee$1,094
- 6Sapphire Preferred$1,071 points + $80 credits − $95 fee$1,056
- 7Bilt Blue$605 points + $0 credits + $423 rent − $0 fee$1,028
- 8Venture$1,119 points + $0 credits − $95 fee$1,024
- 9Amex Platinum$677 points + $1,028 credits − $895 fee$810
- 10Sapphire Reserve$946 points + $655 credits − $795 fee$806
- 11Citi Double Cash$605 points + $0 credits − $0 fee$605
Best: Bilt Palladium · vs. 2% cash back +$923 / yr
Excludes welcome bonuses, lounges and status. Not financial advice — a way to see which assumptions matter.
06
Recommendation
Start from how you redeem and how you travel — not from the advertised value.
- Won’t bother with transfers? A no-fee 2% card beats the classic travel cards for almost everyone.
- Will transfer, travel little? A $95–$325 card that matches your biggest category (groceries and dining → Amex Gold).
- $8k+ of travel a year, or you already use most of the coupons? That is when a $795–$895 card pays.
- Pay rent or a mortgage? Look at Bilt Obsidian or Palladium first — but count only the Bilt Cash you would really spend (section 04).
- Premium value rests on coupon breakage: model it explicitly, because a cohort that learns to use every credit flips the economics.
- The natural premium audience is the ~7% of households with $8k+ travel — target them, not average spenders.
- A transparent calculator like this one builds trust: show the net value at the customer’s real usage, not face value.
- Bilt’s lesson: reward a big expense no one else pays on (housing) and keep the value in your own currency (Bilt Cash). That moved households far more than earn rates did.
07
Limits & next steps
Imputed groceries
Part of Amex Gold’s lead comes from 4x groceries. Next: link the Diary survey to measure groceries directly.
Self-reported spend
Survey spending under-counts some categories, and the card-payable share is approximated.
Steady state only
Welcome bonuses, lounges, status and authorised-user fees are excluded — this is year-two value.
Point values vary
Transfer values are third-party estimates; Bilt’s 2.0¢ is my own call, below TPG’s 2.2¢. The calculator lets you switch to the cash floor.
Data: U.S. Bureau of Labor Statistics, Consumer Expenditure Survey, Interview PUMD 2024; USDA ERS food spending 2024; card terms from issuer sites and card reviews (Oct 2026); point values from The Points Guy (Aug 2026), except Bilt at 2.0¢ (my estimate). Computed by my own code; available on request. Not financial advice.
SAY HI.
Want the notebook, or to talk about how I would turn this into a product feature?