Users leave an information patch only when its yield falls below the environment’s average

What it is

An information patch is a source processed continuously: a document, result list, knowledge-base section, or chat history. Users usually keep extracting after initial difficulty; they leave when the estimated yield per unit time of staying falls below the expected yield of switching to another source in the environment.

Why it happens

Leaving has switching costs—new search, loading, context reconstruction, and source verification—so a slowing patch may be retained while relevant content remains extractable. Diminishing returns, repetition, advertising density, rising cognitive load, or a new cue indicating a better source lowers the staying threshold. The environment average consists of sources already seen and available alternatives.

Studying it

Segment browsing logs into patches and measure entry, dwell, scrolling, clicks, backtrack, and leave time; combine think-aloud records of extracted facts and leave reasons. Manipulate current-page quality, next-candidate scent, and switching cost to observe changes in departure point. Query reformulations in search sessions can also be analyzed.

Where it stops holding

Estimates of yield and average are biased by titles, social proof, and sunk cost. Obligatory tasks may force staying; skilled users may decide quickly. Not every departure is failure—it may be a reasonable transfer.

Applying it

• Provide key points, table of contents, and “applies to / does not apply to” early in long documents and result lists.
• When current content cannot satisfy, recommend adjacent patches, synonyms, related tasks, or human support.
• Reduce pointless scrolling by collapsing outdated content and labeling date and version.
• Use scroll depth and dwell before departure to locate the yield drop rather than relying only on bounce rate.

Related

• Same group: B4.09.1 Users follow paths according to information scent · B4.09.2 Weak scent leads to abandonment rather than deeper search · B4.09.3 Link and label wording directly determines scent strength · B4.09.4 The theory borrows optimal-foraging models and explains path choice by benefit-to-cost ratio · B4.09.6 Scent is a subjective judgment of cues and may mismatch true relevance · B4.09.7 Increasing cue density and shortening path length are different interventions with noninterchangeable effects · B4.09.8 The theory explains abandonment and retreat, not why the goal arose
• Nearby: Q4 Research Methods and Evaluation · T1 Interface Copy and Content
• Search terms: information patch · patch leave · engagement depth