The Persistence-Portability Gap: Why AI Citation Authority Persists Over Time but Fragments Across Platforms

LLM Authority Index Measurement Framework | AI Citation Persistence | Cross-Platform Portability

A website can be consistently important to AI systems without being universally important across AI systems.

That distinction is the foundation of the Persistence-Portability Gap.

LLM Authority Index uses the term Persistence-Portability Gap to describe the difference between:

• persistence, or how consistently a source remains visible in AI citations over time; and
• portability, or how consistently that citation authority transfers across different AI platforms.

The framework emerged from the LLM Authority Index high-stakes consumer citation panel, which analyzed 278,499 visible citation events from 60,881 deduplicated AI responses across six AI platform families and 53 source vertical datasets.

The result was striking.

Eighty-eight domains remained in the overall monthly Top 100 across July, August, and September 2026. Yet only 15 domains appeared in the Top 100 for all six AI platform families.

The leading source layer was highly persistent over time, but much less portable across platforms.

That means a publisher can have durable AI citation authority without having universal AI citation authority.

Persistence asks: Does the source keep showing up over time? Portability asks: Does that authority carry across AI platforms?

Those are separate properties, and they should be measured separately.

Answer Capsule

What is the Persistence-Portability Gap in AI search?

The Persistence-Portability Gap is the difference between how stable a citation source set remains across time and how similar that source set is across AI platforms. In the LLM Authority Index high-stakes consumer panel, the median pairwise Jaccard similarity between monthly Top 100 source lists was 85.2%, while the median pairwise Jaccard similarity between platform Top 100 lists was only 36.1%. Using a like-for-like median Jaccard measurement, that produces a 49.1 percentage-point Persistence-Portability Gap in this dataset. The finding means citation authority was far more stable across months than it was transferable across platforms.

Questions This Article Answers

• What exactly is the Persistence-Portability Gap?
• How should persistence and portability be measured without mixing incompatible metrics?
• Why does the gap matter to publishers, CMOs, PR teams, and AI search practitioners?

The following sections answer each question directly.

Why This Research Belongs in Daily AI Search Marketing Decisions

Questions This Section Answers

• Why does a publisher's durable AI citation authority not automatically transfer across platforms?
• What should CMOs and PR teams do differently when persistence is much higher than portability?
• How should teams use the gap without collapsing it into one misleading authority score?

The Persistence-Portability Gap turns a research finding into an operating rule for AI search strategy:

A source can be durable without being universal.

In the overall high-stakes consumer panel, median monthly Top 100 overlap was 85.2%, while median cross-platform Top 100 overlap was only 36.1%. Using the same median pairwise Jaccard statistic on both dimensions produces a 49.1 percentage-point Persistence-Portability Gap.

That means the broad source ecosystem was far more stable across time than it was transferable across ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Gemini and Microsoft Copilot.

For publishers, this changes how AI citation authority should be packaged and sold. A publisher should not rely only on one aggregate rank or total citation count. A stronger profile shows how often the domain remains visible month to month, which AI platforms actually cite it, which commercial decision categories produce the strongest coverage, whether visibility is broad or engine-specific, and whether the same prompts retain the source over time.

For CMOs, PR teams and AI-search practitioners, the implication is equally practical: there is no single universal publisher list for AI visibility. A source that is strategically important for ChatGPT may be less important for Gemini or Copilot. A source that is highly portable across platforms may still be weak for a specific product category. A source that is highly persistent at the portfolio level may still move in and out of individual prompt-level citation sets.

The four high-stakes decision families now make that pattern more concrete:

These category gaps are study-specific and exploratory. They come from the published category source studies and their platform child studies, not from one newly rebuilt controlled platform-by-month experiment. The financial family also carries a versioning note: the frozen parent reports 28.2% portability, while recomputation from the final six child Top 100 lists is about 27.4%. The published family PPG continues to use the frozen 28.2% value until the family is rerun through one consolidated pipeline.

The operating model is therefore three-layered:

1. Cross-platform core: sources that recur across several engines and can support broad AI visibility programs.
2. Platform-specific layer: sources that matter disproportionately to one engine and should be prioritized when that engine matters commercially.
3. Vertical-specific specialist layer: sources that may be unusually important for one buyer journey even if they are not broad category leaders.

The framework should not be used to create one composite "AI authority score." Persistence, portability, category strength, response citation coverage, prompt-level citation stability, brand mentions, recommendations and conversion outcomes answer different questions and should remain separate.

Commercial Action Matrix: What Publishers and CMOs Can Do With the Findings

Want This Research for Your Company? LLM Authority Index provides vertical-specific AI visibility research, tracking and reporting for brands and publishers, including citations, mentions, recommendations, competitors, source-market changes and platform differences.

Paid engagements do not influence our rankings, inclusion decisions, methodology or research findings.

What Is the Persistence-Portability Gap?

The Persistence-Portability Gap describes a structural feature of AI citation authority.

A publisher can remain visible month after month while its visibility varies substantially between ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, and Microsoft Copilot.

That creates two different dimensions of authority.

Persistence

Persistence measures whether citation visibility survives across repeated measurement periods.

Examples of persistence include:

• remaining in a Top 100 source list for several months;
• maintaining similar rank positions over time;
• preserving response-level citation coverage across repeated observations;
• or producing high overlap between source sets measured in different periods.

Persistence is fundamentally a time dimension.

Portability

Portability measures whether citation authority transfers across AI platforms.

Examples of portability include:

• appearing in the Top 100 on several AI platforms;
• ranking highly across multiple platform-specific source lists;
• producing high pairwise overlap between platform source sets;
• or maintaining broad platform coverage within a defined query population.

Portability is fundamentally a cross-platform dimension.

The gap

The Persistence-Portability Gap appears when those dimensions diverge.

A large positive gap means the source ecosystem is much more stable over time than it is consistent across platforms.

A small gap means temporal stability and cross-platform similarity are closer together.

A negative gap is theoretically possible if source sets are more consistent across platforms than they are over time.

The important point is that persistence and portability should not be collapsed into one undifferentiated idea of "AI authority."

The Finding That Led to the Framework

The framework came out of a simple contrast inside the LLM Authority Index high-stakes consumer citation data.

The monthly source layer was highly persistent

Across the overall Top 100 source lists:

Across all three months, 88 domains appeared in the Top 100 every month.

The median pairwise monthly Jaccard similarity was 85.2%.

That is strong portfolio-level persistence.

The platform source layer was much less portable

Across the six platform-specific Top 100 lists:

• 247 distinct domains appeared in at least one platform Top 100;
• 117 domains, or 47.4%, appeared in only one platform Top 100;
• 37 domains, or 15.0%, appeared in exactly two platform Top 100s;
• only 15 domains, or 6.1% of the 247-domain union, appeared in all six platform Top 100s.

The median pairwise platform Jaccard similarity was 36.1%.

The most similar pair, Google AI Overviews and Google AI Mode, had 69.5% Jaccard similarity.

Even the closest platform pair did not produce an identical source ecosystem.

The platform-overlap study documents the complete distribution:

• Do AI Platforms Cite the Same Websites for High-Stakes Consumer Decisions?

A Like-for-Like Way to Measure the Gap

The intuitive 88 persistent domains versus 15 universal domains result is useful, but those counts should not be subtracted from each other and presented as a formal score.

They answer different questions and involve a different number of sets.

A stronger measurement approach compares the same statistic on both dimensions.

Jaccard-based operational definition

For a fixed query population, source eligibility rule, domain-normalization method, and ranking depth:

Temporal Persistence Jaccard

Persistence = median pairwise Jaccard similarity across time-period source sets

Cross-Platform Portability Jaccard

Portability = median pairwise Jaccard similarity across platform source sets

Persistence-Portability Gap

Gap = Persistence - Portability

For the LLM Authority Index high-stakes Top 100 panel:

Persistence = 85.2%
Portability = 36.1%
Persistence-Portability Gap = 49.1 percentage points

This 49.1 percentage-point gap is a study-specific operationalization, not a universal constant.

It describes this query population, these six platform families, this time period, this ranking depth, and this domain-normalization methodology.

Why median pairwise Jaccard works well here

Using median pairwise Jaccard has several advantages.

It:

• compares source-set overlap using the same unit on both dimensions;
• reduces the influence of an unusually high or low platform pair;
• avoids treating raw citation volume as equivalent across platforms;
• works even when exact rank positions shift inside a stable source set;
• and makes the temporal-versus-platform contrast directly interpretable.

The measure is not the only possible implementation of the Persistence-Portability Gap.

It is one transparent way to operationalize it.

Measurement Rules Matter

The gap is only meaningful when persistence and portability are calculated under comparable conditions.

A valid measurement should hold the following constant wherever possible:

1. Query population
2. Ranking depth
3. Source eligibility rules
4. Domain normalization
5. Response-level versus citation-event counting
6. Geographic and session conditions
7. Observation window
8. Treatment of failures and missing observations

If persistence is calculated from a Top 100 response-level ranking while portability is calculated from raw citation events across all domains, the resulting gap is not a like-for-like comparison.

The two measures may still be individually useful, but subtracting them would create a misleading score.

The Persistence-Portability Matrix

The framework becomes more useful when persistence and portability are treated as two independent axes.

High persistence + high portability

A source remains visible over time and appears broadly across AI platforms.

This is the strongest evidence of durable cross-platform citation authority.

It does not prove causation or recommendation influence, but it indicates that the source is repeatedly present across both time and platform dimensions.

High persistence + low portability

A source remains visible over time but is concentrated on particular AI platforms.

This may be the most strategically important quadrant because an aggregate leaderboard can make these sources look universally authoritative when their strength is actually platform-specific.

Low persistence + high portability

A source appears across several platforms but does not maintain stable visibility over time.

This can happen when a source is broadly useful for a period but lacks durable ranking persistence.

Low persistence + low portability

A source has narrow platform reach and unstable visibility.

There is limited evidence of durable citation authority under the measured conditions.

Why the Gap Matters to Publishers

Publishers increasingly want to know whether AI systems are using their content.

A single citation count or aggregate rank does not fully answer that question.

A publisher may rank highly overall because it is exceptionally strong on a subset of AI platforms.

Another publisher may rank lower overall but appear consistently across all six.

Those are different forms of authority.

The Persistence-Portability framework gives publishers a more precise way to describe their visibility.

Instead of saying:

"We are authoritative in AI."

A publisher can make a narrower and more defensible statement such as:

"Our citation visibility has persisted across three measurement periods and reaches five of the six measured AI platforms."

That is more measurable, more auditable, and more useful to advertisers or brand partners.

Why the Gap Matters to CMOs and PR Teams

A CMO trying to improve AI visibility cannot safely assume that one publisher strategy works across every AI platform.

If the source ecosystem were highly portable, a single prioritized publisher list could be a reasonable approximation.

The LLM Authority Index data suggests that this assumption is weak in high-stakes consumer decisions.

Nearly half of the combined six-platform Top 100 universe appeared in only one platform's Top 100.

Nearly two-thirds appeared in no more than two platform Top 100 lists.

That means a brand targeting ChatGPT, Google AI Overviews, Gemini, Perplexity, and Copilot is not dealing with one publisher ecosystem.

It is dealing with several overlapping publisher ecosystems.

The operational consequence is straightforward:

Build publisher strategy by platform and decision category, then look for sources that provide useful cross-platform overlap.

Why the Gap Matters to AI Search and GEO Practitioners

AI search optimization is often discussed as though "AI visibility" were one channel.

The Persistence-Portability Gap shows why that abstraction can break down.

A domain can have:

• strong long-term visibility;
• high category authority;
• excellent performance on one platform;
• and weak presence on another platform.

A single citation score can hide all four facts.

That suggests AI search measurement should move toward a multidimensional model.

At minimum, a citation authority profile can include:

1. Persistence
2. Portability
3. Category strength
4. Response citation coverage
5. Raw citation-event volume
6. Platform-specific rank
7. Prompt-level citation stability
8. Recommendation or entity coupling, when separately measured

The Persistence-Portability Gap is therefore not intended to replace other metrics.

It helps identify one structural difference that those metrics can otherwise obscure.

Persistence Is Not the Same as Prompt-Level Stability

One of the most important measurement distinctions is between portfolio persistence and same-prompt citation stability.

The overall source leaderboard can remain highly persistent while the exact sources attached to an individual prompt change substantially.

The LLM Authority Index citation-drift study found:

• 6,601 strict same-prompt, same-source-surface July-to-September pairs;
• 4,309 matched pairs with at least one cited domain in either month;
• only 11.3% of those non-empty pairs returned the exact same cited-domain set;
• the mean same-prompt Jaccard similarity was 26.9%;
• and 32.6% of non-empty matched pairs shared no cited domain at all.

Read the full study:

• How Stable Are AI Citations? Measuring Citation Drift From July to September 2026

This creates a useful hierarchy of measurement.

Portfolio persistence

Are the same publishers still prominent in the broader citation ecosystem?

Prompt-level stability

Does the same prompt keep receiving the same source set?

Platform portability

Does authority transfer across AI platforms?

All three can produce different answers at the same time.

A publisher can be persistent at the portfolio level, unstable at the individual-prompt level, and only moderately portable across platforms.

That is not contradictory.

It reflects different layers of the citation system.

The 15 Universal Top 100 Domains

Only 15 registrable domains appeared in the Top 100 for all six platform families in the high-stakes consumer panel:

1. amazon.com
2. bankrate.com
3. businessinsider.com
4. credible.com
5. experian.com
6. finder.com
7. forbes.com
8. goodrx.com
9. healthline.com
10. hearingtracker.com
11. lendingtree.com
12. money.com
13. seniorliving.org
14. squaremouth.com
15. stockbrokers.com

These domains are not identical in business model or editorial role.

The group includes:

• financial publishers;
• marketplaces;
• comparison resources;
• business media;
• health publishers;
• and specialized vertical review sites.

Their common feature in this study is not source type.

It is cross-platform Top 100 portability.

That makes the 15-domain set particularly useful for studying what durable cross-platform source authority looks like.

The Gap Can Exist at Multiple Ranking Depths

The Persistence-Portability Gap does not have to be measured only at the Top 100 level.

Researchers can apply the same framework at:

• Top 10;
• Top 25;
• Top 50;
• Top 100;
• Top 250;
• or another predefined evidence threshold.

The ranking depth should be declared before interpretation.

A Top 10 gap answers a different question from a Top 250 gap.

The Top 10 may reveal the stability and portability of dominant source authorities.

A deeper list may reveal how fragmented the broader citation ecosystem becomes.

This creates a useful research direction:

Does the Persistence-Portability Gap widen or narrow as ranking depth increases?

That question requires its own measurement and should not be inferred from the Top 100 result alone.

The Gap Can Also Be Measured by Category

The category studies now provide direct evidence that the Persistence-Portability Gap is not uniform across high-stakes consumer decisions.

Questions This Section Answers

• Which measured category has the largest Persistence-Portability Gap?
• Which category source market is the most portable across platforms?
• Why should category PPGs be treated as decision-family measurements rather than universal constants?

Insurance has the largest measured category-level gap at approximately 51.2 percentage points. Its leading source layer was very persistent over time, but the final platform-specific Top 100 lists overlapped far less strongly across engines.

Credit, Debt, Banking and Lending is the most portable of the four measured category source markets at 44.9%. Even there, however, more than half of the typical pairwise Top 100 union differs between platforms.

Health and Medical has the lowest measured portability at 25.8% and the lowest Top 100 concentration at 44.9%. Its final six platform source studies contain 290 distinct Top 100 domains, only 20 of which appear in all six platform Top 100s, while 161 appear on only one platform.

Financial Decisions combines moderate persistence with low portability. The frozen parent study reports 66.7% persistence and 28.2% portability, producing an approximately 38.5-point gap. Recomputing portability from the final six child-study Top 100 lists produces about 27.4%. The framework preserves the frozen 28.2% parent value until the family is rerun through a single reconciled pipeline.

Credit has the smallest category gap, not a small gap. Persistence of 77.0% versus portability of 44.9% still produces an approximately 32.1-point difference, and the final six Credit platform studies contain 222 distinct Top 100 domains, with 26 appearing in all six and 97 appearing on only one platform.

These differences matter because the same PPG value cannot be generalized across buyer journeys. A source strategy for Insurance should expect greater engine fragmentation than a source strategy for Credit. A Health strategy should expect an especially wide specialist long tail. A financial strategy should preserve product-level specialization across Mortgage, Investing, Crypto, Retirement and related decisions.

The category values are best understood as decision-family measurements, not fixed properties of AI systems. A different query population, time window, ranking depth, geography, platform configuration or source-normalization rule can produce a different persistence, portability or gap.

What the Gap Does Not Prove

The Persistence-Portability Gap is descriptive.

It does not prove:

• that a citation caused a company to be mentioned;
• that a citation caused a company to be recommended;
• that a highly portable publisher is higher quality;
• that one platform's source-selection system is better than another;
• that a persistent source will remain persistent indefinitely;
• that a publisher appearing across all six platforms will influence the same companies or recommendations;
• or that earning coverage from a cited publisher will automatically increase AI visibility.

Citation-recommendation coupling, company mentions, ranking position, sentiment, and causal influence require separate analysis.

The framework measures where source authority persists and where it transfers.

Measurement Protocol

A reproducible Persistence-Portability study should specify the following before calculating the gap.

1. Define the query population

State the exact decision categories, prompt types, geography, and user conditions.

2. Define the platforms

Treat distinct AI surfaces separately unless there is a documented reason to group them.

3. Define the time periods

Use comparable measurement windows.

4. Define the source unit

For this research, sources are consolidated to registrable domains.

5. Define the response unit

State whether a domain counts once per response or once per citation event.

6. Fix the ranking depth

Use the same Top N threshold for temporal and platform comparisons.

7. Build temporal source sets

Create one ranked source set per time period using the same rules.

8. Build platform source sets

Create one ranked source set per platform using the same rules.

9. Calculate pairwise overlap

For Jaccard-based measurement:

Jaccard = size of intersection / size of union

10. Summarize persistence and portability

Use a declared summary statistic, such as median pairwise Jaccard, for both dimensions.

11. Calculate the gap

Persistence-Portability Gap = temporal persistence metric - cross-platform portability metric

12. Report the components, not only the gap

A gap of 40 percentage points can arise from very different underlying structures.

Always publish:

• the persistence value;
• the portability value;
• the gap;
• the number of periods;
• the number of platforms;
• the ranking depth;
• and the underlying methodology.

Worked Example From the High-Stakes Consumer Panel

Using the Top 100 source sets from the LLM Authority Index flagship research:

Temporal persistence

Monthly pairwise Jaccard values:

• July vs. August: 85.2%
• July vs. September: 83.5%
• August vs. September: 88.7%

Median temporal persistence:

85.2%

Cross-platform portability

The platform study calculated all 15 pairwise combinations across six platform Top 100 lists.

Median cross-platform portability:

36.1%

Persistence-Portability Gap

85.2% - 36.1% = 49.1 percentage points

Interpretation:

In this high-stakes consumer panel, the Top 100 citation source ecosystem was much more stable across months than it was similar across AI platforms.

That is the Persistence-Portability Gap.

Why a Single AI Authority Score Can Be Misleading

A single score is attractive because it is easy to rank.

But a one-dimensional score can hide whether a source is:

• durable but platform-specific;
• broad but temporary;
• highly visible because of one platform;
• moderately ranked but present everywhere;
• stable in aggregate but volatile at the prompt level.

Two publishers with the same aggregate score could therefore have completely different AI citation profiles.

One could be strong on ChatGPT and Perplexity for three consecutive months.

Another could be moderately visible across all six platforms but change rank frequently.

Calling both publishers "equally authoritative in AI" loses useful information.

A better approach is to preserve the dimensions before combining them.

Recommended Reporting Format

For publishers, brands, and AI search measurement platforms, a compact citation authority profile can report:

This preserves information that one aggregate score would otherwise compress away.

Practical Takeaway

The Persistence-Portability Gap changes how AI citation authority should be interpreted.

For publishers:

Durability does not automatically mean universality.

For CMOs and PR teams:

A publisher strategy that works on one AI platform should not be assumed to transfer to another.

For AI search practitioners:

Measure the source ecosystem as a set of overlapping platform networks, not as one global citation leaderboard.

For researchers:

Separate time stability, platform transferability, and prompt-level citation drift before drawing conclusions about "AI authority."

Related LLM Authority Index Research

This measurement framework is the connective layer across the full 66-article High-Stakes Consumer Decisions research corpus.

Foundation studies

• 01 - The 2026 AI Citation Authority Study
• 02 - Overall Reddit Citation Decline
• 03 - Cross-Platform Citation Overlap
• 04 - AI Citation Drift
• Persistence-Portability Gap Glossary Definition

Overall platform source studies

• 09 - Google AI Overviews Top 100
• 10 - Google AI Mode Top 100
• 11 - ChatGPT Top 100
• 12 - Perplexity Top 100
• 13 - Gemini Top 100
• 14 - Microsoft Copilot Top 100

Credit, Debt, Banking and Lending branch

• 05 - All-Platform Credit Citation Authority Study
• 15 - Google AI Overviews Credit Top 100
• 16 - Google AI Mode Credit Top 100
• 17 - ChatGPT Credit Top 100
• 31 - Perplexity Credit Top 100
• 32 - Gemini Credit Top 100
• 33 - Microsoft Copilot Credit Top 100
• 27 - Credit Reddit Decline Parent Study
• 43 - Google AI Overviews Credit Reddit Trend
• 44 - Google AI Mode Credit Reddit Trend
• 45 - ChatGPT Credit Reddit Trend
• 55 - Perplexity Credit Reddit Trend
• 56 - Gemini Credit Reddit Trend
• 57 - Microsoft Copilot Credit Reddit Trend

Insurance branch

• 06 - All-Platform Insurance Citation Authority Study
• 18 - Google AI Overviews Insurance Top 100
• 19 - Google AI Mode Insurance Top 100
• 20 - ChatGPT Insurance Top 100
• 34 - Perplexity Insurance Top 100
• 35 - Gemini Insurance Top 100
• 36 - Microsoft Copilot Insurance Top 100
• 28 - Insurance Reddit Decline Parent Study
• 46 - Google AI Overviews Insurance Reddit Trend
• 47 - Google AI Mode Insurance Reddit Trend
• 48 - ChatGPT Insurance Reddit Trend
• 58 - Perplexity Insurance Reddit Trend
• 59 - Gemini Insurance Reddit Trend
• 60 - Microsoft Copilot Insurance Reddit Trend

Investing, Retirement, Mortgage and Financial Decisions branch

• 07 - All-Platform Financial Citation Authority Study
• 21 - Google AI Overviews Financial Top 100
• 22 - Google AI Mode Financial Top 100
• 23 - ChatGPT Financial Top 100
• 37 - Perplexity Financial Top 100
• 38 - Gemini Financial Top 100
• 39 - Microsoft Copilot Financial Top 100
• 29 - Financial Reddit Decline Parent Study
• 49 - Google AI Overviews Financial Reddit Trend
• 50 - Google AI Mode Financial Reddit Trend
• 51 - ChatGPT Financial Reddit Trend
• 61 - Perplexity Financial Reddit Trend
• 62 - Gemini Financial Reddit Trend
• 63 - Microsoft Copilot Financial Reddit Trend

Health and Medical branch

• 08 - All-Platform Health and Medical Citation Authority Study
• 24 - Google AI Overviews Health Top 100
• 25 - Google AI Mode Health Top 100
• 26 - ChatGPT Health Top 100
• 40 - Perplexity Health Top 100
• 41 - Gemini Health Top 100
• 42 - Microsoft Copilot Health Top 100
• 30 - Health and Medical Reddit Decline Parent Study
• 52 - Google AI Overviews Health Reddit Trend
• 53 - Google AI Mode Health Reddit Trend
• 54 - ChatGPT Health Reddit Trend
• 64 - Perplexity Health Reddit Trend
• 65 - Gemini Health Reddit Trend
• 66 - Microsoft Copilot Health Reddit Trend

Short Definition

Persistence-Portability Gap: The difference between how consistently an AI citation source set remains stable over time and how consistently that source authority transfers across different AI platforms.

Conclusion

AI citation authority has at least two distinct properties: durability and transferability.

A source can remain important month after month without carrying the same authority across ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, and Microsoft Copilot.

The LLM Authority Index high-stakes consumer panel made that distinction visible.

The median pairwise overlap between monthly Top 100 source lists was 85.2%.

The median pairwise overlap between platform Top 100 source lists was 36.1%.

Under a like-for-like median Jaccard implementation, the measured Persistence-Portability Gap was therefore 49.1 percentage points.

That does not mean 49.1 percentage points is a universal property of AI search.

It means that, in this measured query population, source authority persisted far more strongly across time than it transferred across platforms.

That is precisely why AI citation measurement should not stop at a single leaderboard.

It should ask two separate questions:

Does the authority persist?

Does the authority travel?