Anatomy of a Finance Dashboard: What Each Section Shows

The dashboard this tool builds is a single offline HTML file with eleven linked sections, and one decision runs through all of them: every transaction is typed before anything is counted. Money you move between your own accounts, and the payment you make to your own credit card, never count as spending. That single rule is why the numbers here are worth trusting, and why they usually come out lower and more honest than a budgeting app's. Here is what is in the report, section by section, and the methodology that makes each figure defensible.

If you want to click through it while you read, the live demo is loaded with fictional sample data: 7 accounts, 7,110 transactions, 85 months.

Why you can trust the numbers

Before it draws a single chart, the pipeline does four unglamorous things. They are the whole reason the report is more than a pretty spreadsheet.

• Flow-typing. Every transaction is tagged with one of eleven flow types: spend, income, internal transfer, card payment, person-to-person transfer, investment, interest, fee, cash, refund, or foreign exchange. The consequence is the important part. An internal transfer between your own accounts and a credit-card bill payment are never counted as spending. Most tools quietly count both, which is how people end up thinking they spent twice what they did. Get this wrong and every downstream number is wrong.
• Reconciliation. The parser ties each account back to the totals printed on the statement, so the figures match the source instead of drifting from it.
• De-duplication. If you export overlapping date ranges, the duplicates are removed by matching date, description, and amount, so nothing is double-counted or dropped.
• True run-rate. Subscription costs use a trailing-twelve-month run-rate that trims one-off lumps and is cadence-aware, so a charge that bills once a year shows its real per-month cost, not a spike.
• Recurrence detection. A curated catalog of known services plus an algorithm that looks for the same merchant on a regular cadence at a stable amount finds the recurring charges, including the obscure ones.

None of that is glamorous. All of it is why you can read the rest of the report without arguing with it.

Overview and cash flow: money in, money out, and what you actually spent

The Overview opens with the vitals: money in, identified income, total spending, money out, active subscriptions, the digital-subscription run-rate, the all-recurring run-rate, and the time span. The first insight is usually a gap. "Money in" is every dollar that landed in your accounts; "identified income" is only the part that is actually income. The difference is transfers, refunds, and reimbursements. The same gap sits between "money out" and "total spending," because real spending strips out the internal moves. In the demo, that is the difference between roughly $1.5 million of gross flow and about $346k of real spending across seven years.

Cash flow turns that into a diverging monthly chart: income above the line, spending below it, a net line threading through. Toggle it to quarters or years and the multi-year shape appears, with the surplus and deficit months and the seasonal pattern you never notice one statement at a time.

Income and spending: where the money actually goes

Income is stacked by source with an all-time donut, so you can see how concentrated or diversified your income is and how stable it has been. Person-to-person transfers are deliberately kept out of this view. A $5,000 transfer from a friend is not a paycheck, and counting it as income would flatter the picture.

Spending is where most people linger. Monthly stacked bars show the category mix, a full-width treemap lets you drill from a category straight into the individual merchants inside it, and a category-by-period heatmap shows which categories crept up over the years. The treemap is the one that answers "where did it all go," because it goes all the way down to the merchant.

Patterns: the trends you can't see one month at a time

The Patterns section is built for behavior, not budgeting. A GitHub-contributions-style calendar heatmap shades every day by how much you spent, so streaks, paydays, and blowout days jump out. A month-by-year heatmap shows seasonality at a glance, and quarterly bars show the longer arc. This is where the slow, silent creep of a category becomes visible.

Subscriptions: every recurring charge, with its real monthly cost

The Subscriptions section is the one that pays for the afternoon. Group cards summarize your recurring spend, a chart shows how that recurring cost has grown over time, and a sortable table lists every service with a sparkline, an active-or-lapsed badge, and its true monthly run-rate. Because the run-rate is trailing-twelve-month, an annual renewal shows up as its honest per-month figure instead of hiding. In the demo, digital subscriptions alone run about $178 a month and all recurring charges come to roughly $574 a month. Seeing those two numbers together is the moment most people start cancelling. If that is what you are here for, the deeper guide is how to find the subscriptions you forgot about.

Transfers: follow the money

Transfers answers where money moves and who it moves with. A Sankey diagram maps the flow from income sources, through your accounts, out to where it ends up, with internal moves correctly separated from spending. A counterparty chart ranks the people and companies you send to and receive from, and clicking any one of them draws their full in-and-out timeline plus a ledger. That is how you answer "when did this person start paying me so much," which is a question a normal banking app cannot touch.

Accounts and explorer: down to the receipt

Accounts gives you a card per account with money in, money out, net, the record count, and the date range, so you can sanity-check each one and see which account does the heavy lifting. Explorer is the ground truth: a searchable, filterable, paginated table over every single transaction. Every aggregate number elsewhere in the report can be traced back to rows here. Nothing in the dashboard is a black box, which matters when the subject is your own money.

Audit and report: what an examiner would flag

The Audit section is a defensive read: if you were audited, what would draw attention? It produces a risk score and a set of severity-ranked flags, each with the finding, the evidence in real numbers, the exposure, and what to do about it. It looks at large unexplained inbound transfers, gaps between income and outflow, cash intensity, foreign transactions, and round-number transfers. It is informational, not tax advice, but it is the kind of self-check most people never run until someone else runs it for them. Finally, the Report section writes a short plain-language read of the key findings, straight from the data.

See it on your own statements

The fastest way to understand the report is to generate one. Open the live demo to click through every section on safe fictional data, then grab the one-shot prompt and run it on a folder of your own exports. It runs locally and nothing is uploaded; if you want the mechanics first, see how to analyze bank statements with AI. The tool is free and open-source, built by Space & Story.

FAQ

How does it avoid double-counting transfers?

Every transaction is flow-typed before anything is summed. Internal transfers between your own accounts and credit-card bill payments are tagged as such and excluded from spending. Only real consumption, like card purchases and direct debits for rent or utilities, counts as spend, so moving money around or paying off a card never inflates your totals.

Is the audit section tax advice?

No. It is an informational, defensive self-assessment that flags patterns an examiner might question, computed from your own numbers. It is meant to help you find and document things before anyone else asks. For an actual filing decision, talk to an accountant.

How is a subscription's monthly cost calculated?

With a trailing-twelve-month run-rate. The tool sums everything recurring over the last year, trims obvious one-off lumps that are not really subscriptions, and accounts for cadence, so a yearly or quarterly charge is shown as its honest per-month figure rather than a single spike.

Can I trust the totals?

The parser reconciles each account to the totals printed on its statement, and the Explorer section exposes every underlying transaction, so any aggregate can be traced to its rows. The numbers are meant to tie out to your source documents, not approximate them.

Do I have to upload my statements anywhere?

No. The whole pipeline runs on your machine and the finished dashboard is an offline file. The prompt instructs the model in writing never to transmit your data. There is no account to link and no server in the loop.