What Does "Spent" Actually Mean?

One of the harder parts of building a budgeting app is a question that sounds almost too simple:

How much did I spend this month?

You can't just add up everything leaving your accounts.

If you spend $2,000 on a credit card in August and pay the $2,000 balance from checking in September, another $2,000 left an account. You didn't spend $4,000. You spent $2,000, then moved money to settle the bill.

That sounds obvious on paper. Get it wrong in software and the totals still look plausible enough to trust, which is worse.

We've been working through that problem in Watchtower.

Spending is not the same as money moving

A transaction leaving an account might be groceries, a utility bill, a credit-card payment, a transfer to savings, or money sent to an investment account.

They all move money. They are not all spending.

Here's the distinction we use:

Spending is household consumption. It belongs in monthly spend, category budgets, merchant totals, and spending trends.

Money-moving transactions (credit-card payments, transfers) stay visible. They're real. They just aren't new spending.

Hidden transactions are different. Hide something and Watchtower leaves it out of your monthly spend and budgets altogether.

A useful rule:

A transaction can matter without being spending, and it can be visible without affecting the budget.

Credit-card payments are the clean example. The purchases already counted when they happened. Paying the card later adds $0 to spending.

Not August spending. Not September spending. Just $0.

Refunds work the other way

Refunds are the flip side.

Spend $100 on Shopping, get a $40 refund, and net spending is $60. Treating that refund like paycheck income is as wrong as treating a card payment like another purchase.

Refunds on a spending category reduce spending in that category.

Shopping purchase      $100
Shopping refund         -$40

Net Shopping spend      $60

If the refund lands in a later month, it normally affects the month it arrives. The household can already move the date when they want a different close, so we don't need fancy purchase-refund matching in the background.

What matters is using the same accounting definition everywhere.

Uncategorized doesn't mean unspent

Another way to make spend look better than it is: ignore transactions until someone categorizes them.

We don't buy that.

$500 of uncategorized purchases is still $500 spent. Watchtower just doesn't know where it belongs yet.

Uncategorized outflows count toward total household spending. They stay outside any category budget until they're cleaned up.

Later, if that charge turns out to be a credit-card payment, move it to the transfer category and it drops out of spending.

The numbers improve as the information does.

Pending money counts too

We first leaned toward posted-only math for budgets. Pending can change, vanish, or post at a different amount.

All true. We still changed our mind.

If a $180 grocery charge is pending on the card, pretending that $180 is still free in the grocery budget isn't more accurate. The bank just hasn't finalized it yet.

So pending spending counts.

A restaurant authorization for $82 that posts at $101 after the tip should show $82 first, then update the month by $19 when the tip lands.

If the hold disappears, the money drops back out of the calculation.

That isn't Watchtower being inconsistent. It's Watchtower updating when the facts change.

The line we care about:

Pending changes certainty, not what the transaction means.

A pending grocery purchase is spending. A pending credit-card payment still isn't. A pending hidden transaction stays out.

Same rules either way.

When it helps, we can say quietly that part of a total is still pending:

Spent so far      $2,940
Includes $126 pending

The headline is the best picture we have. The secondary line is how much of it hasn't settled.

One definition has to power everything

This became one of the more important rules in Watchtower.

We shouldn't define spending one way on Overview, another in Budgets, and another later for merchant trends or unusual spending.

If a grocery purchase counts as spending, it counts everywhere we measure spending. If a credit-card payment doesn't count, it shouldn't show up again in a chart or insight.

Nobody should have to reverse-engineer why two screens disagree.

Same discipline for remaining.

Say you've budgeted $5,000 and spent $3,000 in those budgeted categories, plus $700 in categories with no budget.

The household spent $3,700. You did not use $3,700 of a $5,000 budget.

The useful view:

Budgeted                 $5,000
Spent against budgets    $3,000
Remaining                $2,000
Other spending             $700
Total spending           $3,700

A little more care underneath makes the numbers easier on the surface.

Budget progress needs context, not judgment

We've also been bringing back a small idea we used to have: a tick on each budget bar for how far through the month we are.

Mint did something similar. I always liked it. Spending suddenly had a sense of time.

The catch: spending isn't linear.

A mortgage can be fully spent on day one. Groceries land throughout the month. Subscriptions hit on fixed dates. Travel might be almost all in one week.

So a calendar marker should never mean:

You should have spent exactly this much by now.

The time context is still useful. We're bringing the tick back with a second reference point.

The bar can show three things:

fill = what you've spent
│    = how far through the month we are
●    = how much you typically have spent by this point

The calendar tick orients you. The historical mark is your own household's pattern.

A mortgage budget might show that we're only a third of the way through the month, and that being 100% spent already is normal.

Groceries might look different: 39% through the month, usually around 55% of grocery spending by now, currently at 70%.

Watchtower doesn't need to say you're "ahead of pace" or paint the bar red. It can just show the information.

Theme we keep landing on:

Context is useful. Judgment usually isn't.

For a new user, only the calendar marker shows. With enough history, we can add the household mark when the pattern is solid enough to trust.

Some categories may never get one.

That's fine.

Boring software should still be thoughtful

None of this is flashy.

We're deciding whether pending counts, whether a refund reduces spend, whether paying a credit card moves the month, and what a tiny mark on a progress bar means.

Those are the decisions that make someone look at a finance app and think:

Yep. That adds up.

Or:

Wait, where did that number come from?

We want the first reaction.

Show the best picture of the household we have right now. Be clear when some of it can still change. Keep one definition of spending under everything.

Simple budgets. Less clutter. Numbers you can trust.

Before Watchtower can say anything interesting about your money, it has to get this part right.