PROVEPT-PRV-9507

Imperfect Stat Stack

A run of several uneven, oddly exact figures reported together, with the weak ones left in, so the set reads as counted rather than chosen.

Definition

What it does

Instead of leading with one flattering number, the copy reports a whole set of results at once, each with its own irregular figure, and keeps the low ones in the list. A 17 sits next to a 93. A modest result follows a big one. The reader stops reading a claim and starts reading a record. The spread across the numbers, not the size of any single one, is what carries the proof.

Why it works

Round, evenly spaced numbers read as authored. Uneven ones read as measured. A single figure invites the question of what got left out, and a stack answers that question before it is asked by showing the range, including the results nobody would pick for a headline. Leaving a poor number in costs the writer something, and people treat a costly admission as more honest than a clean one. The set also gives the eye something to inspect, which feels like checking evidence rather than being handed a conclusion.

Where it appears

Formatssales letters, magalogs, landing pages, product pages, video sales letters, print ads, video ads
Position in the copyproof, bullets, body copy, headlines
Industriesinvesting, supplements, B2B SaaS, fintech, nonprofit, and 3 more industries
In the Taxonomy24 examples from 21 brands

* Most frequent first, based on materials selected for the Persuasion Taxonomy corpus.

Examples of Imperfect Stat Stack

2000s·CXS Penny Stock Fortunes / James Boric

“you could have gained 62.3%, 31.6%, 93.6%, 28.7%, 54.1%, 17.5%, and 52.4%”

A penny stock advisory reports what its rating system averaged per trade and then lists the individual results behind that average.

Why it’s this techniqueSeven returns arrive at one decimal place apiece, '62.3%, 31.6%, 93.6%, 28.7%, 54.1%, 17.5%, and 52.4%', with the average stated as '33.8%' and the compounded figure carried to the cent at '$3,677.83'. The uneven precision does the work: numbers this ragged read as transcribed from a record instead of composed for a headline. The tell is the mix. A deliberately small '17.5%' sits beside the '93.6%' in unsorted order, so the run declines the flattery a single round number would carry. The claim rests on the texture of the digits, not on the named method behind them.

Classification

Primary technique
PT-PRV-9507
Classification confidence
0.90
Source
CXS Penny Stock Fortunes, 2000s direct mail
Also an example of
2020s·AG1

“✅85% of participants felt more energy† ✅70% noticed less gas and bloating† ✅84% felt less tired in the afternoon†”

A greens powder ad reports self-reported outcomes from its own three month study and spells out the study design in a footnote.

Why it’s this techniqueThree benefit claims arrive as a run of odd figures: '85% of participants felt more energy', '70% noticed less gas and bloating', and '84% felt less tired in the afternoon'. None rounds to a clean number and none reaches 100, so the ragged spread reads as measurement rather than marketing. The structural tell is that the weakest figure stays in the stack instead of being cut, and the order refuses to climb, which is what separates this from a straight proof list. The study footnote supplies backing, but the belief is carried by the shape of the numbers.

Classification

Primary technique
PT-PRV-9507
Classification confidence
0.85
Source
AG1, 2020s ad
Also an example of
2000s·WaveStrength Trader

“generated an accuracy rate of 81% and total gains of 1,164% on its closed picks! Folks who followed the 'WaveStrength 2004 All Stars' picks realized gains of 155%... 159%... 71%...28%...27%...and more!”

A stock alert email reviews the prior year's published picks and the gains each of them closed at.

Why it’s this techniqueThe numbers arrive rough and out of order. An '81%' hit rate sits beside total gains of '1,164%', then individual winners drop in unsorted: '155%... 159%... 71%...28%...27%...and more!' The descent is uneven, 159% lands after 155%, and small results stay in the run rather than being pruned to the highlights. That raggedness is the tell. A polished claim would round, sort, and stop at the biggest figure; here the ugly precision and the modest tail do the persuading, because a list nobody bothered to clean up reads as a ledger rather than a pitch. The date stamp only certifies the ledger.

Classification

Primary technique
PT-PRV-9507
Classification confidence
0.88
Source
WaveStrength Trader, 2000s email
2010s·Quentin

“79% Don't understand Cyber Security 80% Don't use data protection 91% Don't use endpoint mobile security 83% Don't have a plan for cyber security 70% Don't have an adequate backup system”

A cyber security pitch opens with a block of survey findings about how unprepared small businesses are.

Why it’s this techniqueFive figures land in a column and none of them rounds off: '79%', '91%', '83%', values ragged enough to read as counted rather than asserted. Each attaches to a separate exposure, 'Don't use data protection', 'Don't have a plan for cyber security', so the reader cannot wave off one number without facing four more. The tell is repetition of an identical frame across every line with no source, no explanation, and no single figure asked to carry the argument. Accumulated bulk persuades. Agitation of risk rides along, but the unrounded values and the stacking are what the copy is built on.

Classification

Primary technique
PT-PRV-9507
Classification confidence
0.82
Source
Quentin, 2010s web page
2000s·Agora / Dr. Jonathan V. Wright

“Several studies show that it can: Cut TOTAL cholesterol as much as 17%... Slash LDL cholesterol by 25%... Raise HDL ('good') cholesterol up to 29%... Drop dangerous triglycerides as much as 18%...”

A health newsletter promotion lists what published trials report a plant extract does to four separate blood markers.

Why it’s this techniqueFour claims arrive as a run of odd, unrounded figures, '17%', '25%', '29%', '18%', each capped by a ceiling phrase, 'as much as' and 'up to', so the numbers read as measured results rather than sales rounding. The tell is the spread: the values refuse a pattern, they are not multiples of five, and they do not climb in order, which is what a real study table looks like and what invented numbers do not. A cited frame, 'Several studies show', hands the arithmetic to outside record keepers, so the reader audits the digits instead of the promise.

Classification

Primary technique
PT-PRV-9507
Classification confidence
0.80
Source
Agora, 2000s direct mail
Also an example of

See whether your own copy uses Imperfect Stat Stack, and what else it is doing: analyze your copy.

Boundary Conditions

When it lands

  • The figures come from one real measured set, a portfolio, a trial, a customer survey, so the range between them is genuine
  • At least one number in the run is ordinary or disappointing and is left exactly where it fell
  • Each figure carries its own unit or window, a percent, a count, a number of days, so a reader could go check it
  • The claim the numbers support is modest enough that the spread backs it up instead of undercutting it

When it dilutes

  • Every figure in the run is a winner, which turns the stack back into a highlight reel
  • The numbers are rounded or evenly spaced, such as 20, 40 and 60 percent, which reads as written rather than counted
  • The denominator is missing, so a reader cannot tell how many results were left out of the list
  • The run keeps going well past six or seven items and the eye stops reading figures and starts skipping them

Taxonomic Relationships

Provenance

Introduced in v1.0Last revised 2026-09-16MethodologyErrata