Last updated: October 2026
Most people do not have a rewards strategy. They have a rewards card — one card, used for everything, earning whatever it earns. That is not a criticism. It is the rational default when comparing cards takes an afternoon and the payoff is a few hundred dollars a year.
So we measured the payoff. We built 1,000 modeled U.S. households, ran each one through the same optimizer that powers our credit card rewards calculator, and compared what those households could earn against a single flat-rate card used for every purchase.
The median household leaves $1,376 a year on the table.
The mean gap is $1,548. Even measured against a 2% flat-rate card — the best no-thought single card there is — the median gap is $1,230. And if you strip out every statement credit and count only rewards minus annual fees, the median household still leaves $551 a year on the table against a 1.5% card, or $405 against a 2% card.
Read this first: these are modeled households, not survey respondents. We generated them from the calculator’s own spending presets, scaled by a modeled income distribution, and then optimized them. The point is not to describe a particular family. The point is to measure how much value sits between “one card” and “the right cards” across a realistic spread of spending. The method, the assumptions and every row of data are public — see the methodology or skip to the raw data.
What “left on the table” means here
The phrase is doing specific work, so here is the definition:
Left on the table = (value of the optimized wallet) − (value of one flat-rate card on the same spending).
Both sides are measured the same way, over the same spending:
- Optimized wallet: the calculator’s pick for that household with up to four personal cards, valued at Year 2+ — no signup bonuses, since a bonus is a one-time event and we did not want it to flatter the result.
- Flat baseline: a single no-fee card paying 1%, 1.5% or 2% on everything. No categories, no credits, no annual fee.
The 1.5% case is the headline because it is the most common shape of a “normal” rewards card: a no-fee card that pays a flat rate and asks nothing of you. It is also not a straw man. The CFPB’s 2025 Consumer Credit Card Market Report puts the rewards actually earned on general-purpose rewards cards at about 1.6% of purchase volume — so 1.5% is the realistic status quo. The 2% case is the generous version of it, which makes it the conservative floor for the gap, and the 1% case is what you get with no rewards strategy at all.
One thing worth holding next to the baseline: the gap here is about strategy, not card count. The average U.S. consumer already has about 3.7 credit cards in active use (Experian, June 2025). A household can hold four cards and still put every dollar on one of them — which is exactly what the flat baseline models. The opportunity is not opening more accounts; it is using the ones already in the drawer.
The headline numbers
| Status quo | Median gap | Mean gap | 10th percentile | 90th percentile |
|---|---|---|---|---|
| 1% flat cash back | $1,518 | $1,709 | $820 | $2,839 |
| 1.5% flat cash back | $1,376 | $1,548 | $737 | $2,570 |
| 2% flat cash back | $1,230 | $1,387 | $665 | $2,294 |
The model’s median household earns about $86,350 a year and puts about $2,366 a month — roughly $28,386 a year — on credit cards, after excluding rent and mortgage. Across that spread of income and spending, the gap between “one card” and “the right cards” is remarkably stable: a median of about 4.9% of card spending, and between 4.7% and 4.9% in every income band.
That stability is the most interesting finding. The gap is not a rich-person problem. It scales with spending almost exactly.
Almost nobody is at zero, either. Every modeled household leaves at least $100 a year on the table against a 1.5% card, 99.1% leave at least $500, and 75.8% leave at least $1,000.
Two cards get you 92% of the way
The optimism in every rewards article is that you need a wallet full of cards. The data says otherwise.
| Wallet size | Median optimized value (Year 2+) | What the extra card adds (median) | Median gap vs 1.5% flat |
|---|---|---|---|
| 1 card | $1,318 | — | $894 |
| 2 cards | $1,687 | +$371 | $1,263 |
| 3 cards | $1,774 | +$86 | $1,352 |
| 4 cards | $1,797 | +$21 | $1,376 |
A two-card wallet captures 92% of the four-card gap, and the second card is where almost all of the money is: a median of +$371 a year.
After that the curve flattens hard. The third card adds a median of $86, and for 64% of households it adds less than $100. The fourth card adds a median of $21, and for 67% of households less than $25. That is not a reason to open a fourth card — it is a reason to stop at two or three.
This is consistent with what we found in the full rewards study behind our guides, and it is the single most useful thing in this dataset: get the first two cards right; the rest is small change. That study’s headline second card is bigger (a median $506) because it weights every spending level from $1,000 to $10,000 a month equally. At $2,500 a month, close to this sample’s median of $2,366, it finds a second card worth $387 — in line with the $371 here. The methodology explains the difference.
Where the money actually comes from
Here is the part most rewards studies skip, and it matters.
| Wallet size | Rewards minus fees | Statement credits counted | Year 2+ total |
|---|---|---|---|
| 1 card | $974 | $485 | $1,318 |
| 2 cards | $962 | $725 | $1,687 |
| 3 cards | $1,044 | $725 | $1,774 |
| 4 cards | $972 | $825 | $1,797 |
Look at the middle column. The optimizer’s four-card wallet counts about $825 a year in statement credits at the median — and rewards minus fees barely move from one card to four. Most of the extra value in a multi-card wallet, as our optimizer scores it, comes from credits you have to remember to use, not from earning rates.
That is why we report the credits-free number alongside the headline:
| Status quo | Median gap (credits included) | Median gap (rewards only) |
|---|---|---|
| 1.5% flat cash back | $1,376 | $551 |
| 2% flat cash back | $1,230 | $405 |
The honest headline depends on who you are. If you will actually use the dining credits, the Uber Cash and the lounge access you are paying for, the bigger number is real. If you won’t, the smaller number is the one to trust — and it is still a few hundred dollars a year for doing nothing more than carrying the right second card.
The gap by income
| Modeled household income | Households | Median monthly card spend | Median gap vs 1.5% flat | Gap as share of spend |
|---|---|---|---|---|
| Under $40k | 150 | $1,242 | $718 | 4.9% |
| $40k–$60k | 148 | $1,823 | $1,060 | 4.9% |
| $60k–$80k | 158 | $2,059 | $1,154 | 4.7% |
| $80k–$100k | 120 | $2,405 | $1,391 | 4.7% |
| $100k–$150k | 179 | $2,834 | $1,621 | 4.9% |
| $150k–$250k | 147 | $3,654 | $2,120 | 4.9% |
| $250k and up | 98 | $5,248 | $3,002 | 4.7% |
The dollar gap grows steeply with income, but the percentage gap does not move. Every band leaves roughly the same share of its card spending on the table.
By household type the pattern is the same, driven by spending rather than by demographics:
| Household type | Households | Median monthly card spend | Median gap vs 1.5% flat |
|---|---|---|---|
| Single | 311 | $1,918 | $1,118 |
| Couple | 329 | $2,357 | $1,352 |
| Family with children | 360 | $2,969 | $1,668 |
If you would rather not deal with points
Points are worth what you can get for them, and that requires a transfer partner, an award seat and some patience. Plenty of people would rather have cash.
Restricting the optimizer to cards that pay cash back — 11 of the 95 cards in the catalog — the median optimized wallet is worth $856 a year, and the median gap is $433 against a 1.5% flat card and $289 against a 2% card. Smaller, but not nothing, and it comes with no annual-fee math to do.
The cash-back optimizer picked the same pair for 99% of households: Costco Anywhere Visa + Robinhood Gold. That is a narrower answer than the points case, which is a limitation of the catalog rather than a discovery — there are not many no-fee cash-back cards that beat 2% on enough categories to matter.
The wallets the optimizer actually picked
Across all 1,000 households, with all reward types allowed:
| Allowance | Most common wallet | Share |
|---|---|---|
| 1 card | Capital One Venture X | 53% |
| 2 cards | American Express Gold + Capital One Venture X | 93% |
| 3 cards | Amex Gold + Venture X + Wells Fargo Autograph | 88% |
| 4 cards | Amex Gold + Venture X + Citi Strata Premier + Wells Fargo Autograph | 99.5% |
The one-card answer is a single premium travel card. The two-card answer is a dining-and-groceries earner plus that premium card, which is the two-card structure we would expect to see: one card for the categories where the bonus is big, one card for everything else and for the transfer partners. For more on those specific cards, see our guides to the Amex Gold, the Chase Sapphire Reserve and our per-purchase cheat sheet.
What this study does not say
A few limits worth stating plainly, because a number without them is a marketing claim:
- These are modeled households, not real ones. The income distribution is calibrated to the U.S. household income distribution and the spending shapes come from our calculator’s presets. No row is a person.
- Points are valued at The Points Guy’s October 2026 transfer valuations. In cash terms, points are worth less. The cash-back-only view exists for exactly this reason.
- Statement credits are counted at the calculator’s defaults. We show the credits-free number so you can judge for yourself.
- The optimizer assumes you can get approved and will pay in full. Interest charges dwarf any rewards decision — the Federal Reserve’s 2025 household survey found that 45% of cardholders carried a balance at least once in the previous year. If you carry a balance, the right answer is to pay it down, not to optimize categories.
- Rent and spending abroad are excluded. Most rent payments cannot go on a card without a fee, and foreign transactions have their own fee structure.
- Only 11 catalog cards pay cash back. The cash-back view rests on a small set.
The full method, including every assumption and the exact seed, is on the methodology page.
The raw data
Everything behind this article is public and reproducible:
- All 1,000 households (CSV) — income, spending by category, every baseline and every wallet the optimizer picked
- Medians by income band (CSV)
- Full results (JSON)
- Method notes and the regeneration command
The command is go run ./scripts/cmd/household-study. It uses the same catalog, the same optimizer and the same point valuations as the live calculator, and it runs in under a minute with no network access.
The short version
- The median modeled household leaves $1,376 a year on the table against a typical 1.5% flat rewards card.
- Against the best no-thought flat card (2%), the gap is $1,230.
- Counting rewards only, with no statement credits, the gaps are $551 and $405.
- Two cards capture 92% of the four-card upside.
- The gap is about 4.7% to 4.9% of card spending at every income level.
- If you want cash instead of points, the median cash-back gap is $433.
Ready to see your own number? The rewards calculator runs the same optimizer on your spending in about a minute, and gives you the gap — in dollars — between the card you have and the cards you should.
