Related frameworks

Visibility Bollinger Bands

Category	AI visibility framework, generative engine optimisation
Adapted from	Bollinger Bands (technical analysis)
Originator	Paul Truscott
Year introduced	2026
Components	Middle band (moving average), upper band, lower band
Typical parameters	N = 20 periods, K = 2 standard deviations
Input data	Rolling citation or mention frequency across AI-generated answers
Citation RSI Entity Support and Resistance Visibility Drawdown

Visibility Bollinger Bands is a framework used in generative engine optimisation (GEO) that plots a moving average of an entity's tracked citation or visibility frequency alongside upper and lower bands set at a specified number of standard deviations. Paul Truscott developed the framework in 2026 by adapting Bollinger Bands, a statistical charting method John Bollinger introduced for financial markets in the 1980s.

The framework separates ordinary statistical fluctuation in AI visibility from a genuine, statistically significant shift in an entity's standing. A reading that stays inside the bands represents normal variance. A reading that pushes outside either band signals a movement large enough, relative to the entity's own historical behaviour, to warrant further diagnosis.

Overview

Visibility Bollinger Bands answer a question raw citation counts cannot: is this specific movement in an entity's AI visibility normal, or exceptional? A ten percent week-on-week change in citation frequency can be routine for a high-volatility category and a five-alarm signal for a stable, well-established brand. The bands express volatility relative to the entity's own recent behaviour rather than against a fixed threshold, so the same percentage move is judged differently depending on the entity's history.

Origin and development

Paul Truscott built Visibility Bollinger Bands from the same problem John Bollinger's original indicator solved for price data: a value on its own tells almost nothing about whether a move is normal or exceptional, because normal differs for every instrument and every period. Truscott's claim is that most of what gets reported to clients as a "visibility drop" or a "citation surge" is statistical noise sitting comfortably inside the entity's normal range, and that treating ordinary fluctuation as a crisis, or a normal dip as an emergency, spends budget reacting to nothing.

Visibility Bollinger Bands sits alongside Citation RSI, Entity Support and Resistance, and Visibility Drawdown as one of four original frameworks Truscott built at the intersection of semantic SEO, GEO, and financial technical analysis.

Calculation

Visibility Bollinger Bands follow the same three-line construction as the original indicator: a moving average flanked by two bands set a fixed number of standard deviations away.

Middle Band = N-period moving average of citation frequency
Upper Band  = Middle Band + (K × σ)
Lower Band  = Middle Band − (K × σ)

σ is the N-period standard deviation of the entity's citation frequency over the same rolling window used for the moving average. Typical parameters mirror the original indicator's defaults: N set to 20 periods and K set to 2 standard deviations, though both are adjustable to the reporting cadence and volatility profile of a given entity or category.

Percent V and bandwidth

Percent V shows where a current citation reading sits relative to the bands, expressed on the same 0-to-1 scale as the original indicator's %b measure.

Percent V = (Last Citation Reading − Lower Band) ÷ (Upper Band − Lower Band)

A Percent V of 1 places the reading at the upper band; a Percent V of 0 places it at the lower band. Visibility Bandwidth shows how wide the bands are on a normalised basis.

Visibility Bandwidth = (Upper Band − Lower Band) ÷ Middle Band

Narrow bandwidth indicates a period of low volatility in an entity's citation behaviour. Expanding bandwidth indicates an increase in the rate of change of that entity's visibility.

Interpretation

A citation reading that touches the lower band and returns toward the moving average is read as a bounce: the dip stayed inside normal variance and does not, on its own, warrant intervention. A reading that breaks through the upper or lower band and holds outside it is read as a genuine shift in standing, distinct from ordinary week-on-week noise.

When the bands run close together and track roughly parallel over an extended period, an entity's citation frequency generally oscillates between them as though moving inside a channel, matching the behaviour the original indicator describes for price during low-volatility periods.

The visibility squeeze

A period of narrow bandwidth, a visibility squeeze, often precedes a period of expanding bandwidth. An entity whose citation frequency has compressed into a tight range for an extended period carries a higher likelihood of a coming breakout in visibility, in either direction, once the reduced range resolves. This mirrors the volatility-contraction pattern traders watch for on a Bollinger Band chart before a price breakout.

Application in generative engine optimisation

Paul tracks an entity's citation frequency across AI-generated answers over a rolling period, calculates the moving average, and sets bands at a specified number of standard deviations above and below it. A reading that stays inside the bands is treated as normal variance and requires no intervention. A reading that pushes outside the upper band gets checked against Citation RSI to determine whether the surge is supported by genuine corroboration or represents an overextension at risk of correcting. A reading that pushes outside the lower band triggers a Visibility Drawdown assessment to establish the severity and likely cause of the decline.

This keeps client reporting grounded in statistically real movement rather than reacting to every week-on-week wobble in citation data.

Limitations

Citation and mention frequency data have no confirmed statistical distribution and can carry fat tails relative to a normal distribution, matching the same caveat researchers have raised about security price returns. A 20-period sample, the typical default, can prove too small for statistical techniques relying on the central limit theorem to produce reliable conclusions, particularly for entities with sparse or seasonal citation activity.

Citation events are commonly correlated in sequence rather than independent observations, since one AI-generated citation often follows directly from the same underlying corroboration event as the citation before it. This serial correlation limits how cleanly the bands separate genuine signal from a run of related mentions counted as separate data points.

Related frameworks

Framework	Function
Citation RSI	Measures whether an entity's citation frequency has outrun its corroboration base
Entity Support and Resistance	Identifies the visibility levels an entity holds against or breaks through
Visibility Drawdown	Measures the depth and duration of a visibility decline from a prior peak

See also

• Bollinger Bands
• Citation RSI
• Standard Deviation
• Entity Support and Resistance
• GEO (Generative Engine Optimisation)
• Semantic SEO

References