Every rewards card is a bet by the issuer that you will not use it optimally. They pay four percent on dining because they know most of your spending is not dining, and they know you will put your groceries on it anyway out of habit.

Using cards well is not complicated. It is two decisions: figure out where your money actually goes, and carry the right card for the two or three biggest buckets. This takes an hour once and then pays out for years.

Step one: find your real categories #

Not what you think you spend on. What you actually spend on.

Pull twelve months of statements and add up the totals by category. Most card issuers and banks will categorize this for you automatically, and the categorization is imperfect but close enough. You are looking for the ranked list.

For most people in an expensive city, the ranking looks something like:

1. Rent or mortgage, usually not chargeable, or chargeable at a fee that eats the reward
2. Groceries
3. Dining and takeout
4. Transportation: gas, transit, ride-share
5. Utilities and phone
6. Travel
7. Everything else

The shape varies a lot. Someone who cooks every night and does not travel has a completely different optimal setup than someone who eats out five nights a week and flies monthly. Which is the whole reason to do this from your own data rather than from a “best cards of the year” list.

Step two: know what a percentage point is worth #

This is the step people skip, and it is why they end up with four cards for no reason.

An extra one percent on a category is worth one dollar per hundred dollars spent. So:

Look at that first row. If you spend a hundred dollars a month on something, the difference between a mediocre card and a perfect card for that category is twelve dollars a year. It is not worth opening a card, tracking a category, and remembering which one to pull out for twelve dollars.

So only optimize categories where you spend more than about three hundred dollars a month. For most people that is two or three categories, which means two or three cards, which is a system a human can actually run.

Step three: the standard setups #

The one-card setup is a single flat-rate card paying two percent on everything. Zero thinking, zero tracking, and it captures roughly seventy percent of what a heavily optimized setup would earn. If you do not want a hobby, this is a completely respectable answer and I would not talk anyone out of it.

The two-card setup is a flat two percent card for everything, plus one category card covering your single biggest bucket. If you spend eight hundred a month on groceries, a card paying four or five percent there is worth roughly two hundred dollars a year over the flat card. This is where most of the available gain lives.

The three-card setup adds your second category. Diminishing returns start here but it is still worth doing if the categories are large.

Beyond three cards you are optimizing for hundreds of dollars while adding real cognitive overhead and real risk of missing a payment. There are people who run twelve cards profitably. They are running a hobby, and they know it.

The thing that breaks category cards: merchant category codes #

Your card does not know what you bought. It knows the merchant category code the merchant is registered under, and it pays the bonus rate based on that code. This produces results that feel arbitrary until you understand the mechanism.

Warehouse clubs and superstores usually do not count as grocery stores. Costco, Walmart, and Target are typically coded as warehouse clubs or discount stores, not supermarkets, so your five percent grocery card often pays one percent there. This surprises everyone, every time. Meanwhile groceries bought inside a Whole Foods that is coded as a supermarket do count. Same food, different code, different reward.

Delivery apps vary, with some coding as restaurants, some as grocery, and some as something else entirely. A restaurant inside a hotel may code as travel rather than dining. Gas stations that are mostly convenience stores sometimes code as convenience stores.

You cannot look this up reliably in advance because merchants change their coding. What you can do is check your statement after the fact and see which rate you earned. Do this once for each place you shop regularly, and then you know.

Rotating categories and caps #

Some cards pay five percent in categories that rotate quarterly, with an activation requirement and a spending cap, commonly fifteen hundred dollars per quarter.

You have to activate, every quarter. If you do not, you earn the base rate. Card issuers are counting on this, and it is the entire business model of the rotating category card. Set a recurring calendar reminder for the first week of January, April, July, and October.

The cap is real. Five percent on fifteen hundred dollars is seventy five dollars per quarter, three hundred a year if you max every quarter. That is the ceiling and it is a decent ceiling. But spending beyond the cap earns the base one percent, so the card is not a general-purpose card even during its good quarter.

Cards with quarterly caps and activations are the ones most likely to sit unused. If you are the kind of person who will not activate, do not get one, and there is no shame in that.

Points versus cash back #

Cash back is worth a cent per point, always, no thinking required.

Transferable points from the major issuers can be worth more than a cent when transferred to airline and hotel partners and used for expensive redemptions. Business or first class international flights are where the outsized values live. They can also be worth less than a cent if you redeem them for a gift card or merchandise.

If you will not spend time on award redemptions, take the cash back. The theoretical two-cents-per-point valuations you read about require finding award availability, understanding transfer partners, and being flexible on dates. That is a real skill and a real time investment. If you have it, points win comfortably. If you do not, points are cash back with extra steps and a worse exchange rate.

Keeping the system running #

Here is the failure mode of every category setup ever devised. You do the analysis, you open the cards, and for three weeks you pull the right card at the right register. Then you stop thinking about it, and everything goes on whichever card is in the front slot of your wallet.

The optimization only pays if it executes at the point of purchase, in the two seconds you are standing at a terminal. Memory does not do this reliably across three cards and six categories.

I ended up building Credit Card Central for this. You enter your cards once, and when you are about to pay for something it tells you which card earns the most in that category. It also tracks annual fee renewal dates so a fee never posts as a surprise. It is free, there are no accounts, and everything stays on the device, which matters for an app that knows which credit cards you hold. The build story is in how I built a credit card tracker.

A spreadsheet works too. A sticker on the back of each card with its best category works, and I know people who do exactly that. The point is that the system has to survive contact with a checkout line.

A realistic result #

For someone spending thirty five thousand a year on cards, a flat 2% card returns about $700. A two-card setup done properly returns $950 to $1,100. A three-card setup executed consistently returns $1,100 to $1,400.

So the whole exercise is worth roughly four hundred to seven hundred dollars a year over the lazy option. That is real, and it is also small compared to a single decision about where you live or how much you save. Keep it in proportion.

Next, the question of whether a card with a fee is worth carrying at all, which is in the annual fee math. And the rules that keep any of this from going wrong are in credit card rewards without getting burned.