You Agreed to 25,000 Words. Reading Them Takes 102 Minutes.
Four budgeting apps' privacy policies, measured. Empower's runs 11,352 words at a postgraduate reading level. Reading all four takes an hour and 42 minutes — and that is the optimistic estimate.
Of four measured in September 2026: Empower 11,352 words, YNAB 7,161, Monarch Money 5,890, Goodbudget 1,176. Together 25,579 words, which takes about 102 minutes to read at 250 words per minute.
Get new posts in your inbox
No spam, unsubscribe anytime.
About the Author
Sujit KarkiFinance Researcher & Market Analyst
Independent finance researcher and market analyst with expertise in macroeconomics, equity markets, and personal finance. I help regular investors make better-informed decisions through rigorous, data-driven analysis.
Four budgeting-app privacy policies total 25,579 words — about 102 minutes of reading at a brisk pace.
Empower's runs 11,352 words at Flesch-Kincaid grade 16.6, past a bachelor's degree.
The four average grade 14.8, above the reading level of roughly half of US adults.
Goodbudget's is the outlier at 1,176 words — proof the length is a choice, not a legal necessity.
This measures length and reading level only. It makes no claim about what the policies permit.
A budgeting app wants read access to every transaction you make. In exchange it
shows you a chart. Before that happens you click a box confirming you have read a
document.
Nobody reads it. This is a measurement of what "it" actually is.
What you agreed to
Measured 7 September 2026
App
Words
Reading time
Reading ease
Grade level
Empower
11,352
45.4 min
22.7
16.6
YNAB
7,161
28.6 min
29.6
15.7
Monarch Money
5,890
23.6 min
31.9
14.2
Goodbudget
1,176
4.7 min
40.6
12.9
25,579 words. 102 minutes. That is the reading you consented to having done,
across four apps that each do roughly the same job.
And 102 minutes is the optimistic number. It assumes 250 words a minute, a
normal pace for ordinary prose. Legal text is slower — you re-read clauses, you
stop at defined terms. The real figure is higher, and I have used the flattering
one deliberately so the argument does not depend on a generous assumption.
Empower's policy alone is 45 minutes. That is a feature-length reading
commitment attached to a checkbox.
The reading level is the real barrier
Length is the obvious problem. Difficulty is the more serious one.
The four average grade 14.8 — a university sophomore. Empower's requires 16.6,
past a completed bachelor's degree, and scores 22.7 on reading ease, which is
harder than every crypto whitepaper I measured last
week and harder than the US
Constitution.
Roughly half of US adults read below the level these documents demand. For those
readers the policy is not merely long — it is not readable at all, in the plain
sense that the words will not resolve into meaning at any reasonable speed.
Pro Tip
Look at Goodbudget: 1,176 words, grade 12.9, under five minutes. It is a
budgeting app doing the same category of thing as the others, and its policy is a
tenth the length of Empower's. The length is a choice, not a legal
requirement. That is the most useful single fact in this post, because it means
a short policy is possible and most companies are choosing otherwise.
What I deliberately did not measure
This post does not tell you which app is more invasive, what data each shares, or
who they sell to. That would be the more useful article, and I am not writing it
from this data, because I do not have a method I trust.
The previous experiment on this site
tried to score qualitative text by keyword matching. It produced confident,
precise-looking numbers that were wrong in both directions, and I only caught it
by reading the raw output. Detecting "does this policy permit selling your data"
by searching for phrases has exactly the same failure mode: a policy can permit
it without the word "sell" appearing anywhere, and can contain "we do not sell
your personal information" three clauses before an exception that swallows it.
Word counts and readability formulas are arithmetic over counts. They cannot be
wrong in that way. So this post reports only what the arithmetic supports, and
the more interesting claim goes unmade rather than made badly.
What to actually do
Given that reading them is not realistic, three things that are:
Check what the app connects to, not what it says. The permissions you grant
at the bank-linking step are the actual disclosure. Read-only access to
transactions is different from access to move money.
Prefer the shorter policy, all else equal. Not because short means safe,
but because a company that wrote 1,200 words has made a decision about
comprehensibility that a company writing 11,000 has not.
Disconnect the ones you stopped using. Access persists after you stop
opening the app. Revoke it at your bank, not just in the app.
The honest summary is that the consent model does not work at this length, and no
amount of diligence on your part fixes that. What you can control is how many of
these you are signed up to at once.
Documents. Four privacy policies fetched from each company's own domain on 7
September 2026 — YNAB, Monarch Money, Goodbudget as HTML, Empower as the PDF its
privacy centre links to.
A correction caught before publishing. Empower's /privacy page is a privacy
centre — a hub that links to the real document. Scoring it directly gave 629
words, which would have understated the policy by a factor of eighteen. The
script now targets the linked PDF. If a figure in a post like this looks
implausibly small, it usually means the extraction found a landing page.
Measures. Word count, Flesch Reading Ease, Flesch-Kincaid Grade Level, and
reading time at 250 words per minute — a common benchmark for adult reading of
non-technical prose, and optimistic for legal text.
Excluded, and why. PocketGuard blocks automated requests; EveryDollar and
Copilot's privacy URLs return 404; Mint and Simplifi did not resolve reliably.
They are listed in the data as exclusions rather than quietly omitted, since an
unmentioned gap is how a sample gets flattering. Four apps is a small sample and
this does not claim to characterise the category.
Limits. Readability formulas measure sentence and word length, not
conceptual difficulty. They are a floor: these documents are at least this hard.
Disclosure. The commentary was drafted with AI assistance; the document
selection, the method and the conclusions are mine.
Sujit Karki||5 min read