Pith. sign in

REVIEW 3 major objections 6 minor 48 references

Answering Without Referring: How AI Search Rewrites the Web's Economic Bargain

T0 review · 3 major / 6 minor · reviewed 2026-07-09 · glm-5.2

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read AI search sends outbound clicks in only 5.2% of sessions vs 31.1% for Google

desk verdict Solid empirical paper documenting AI search's effect on web traffic routing; deserves serious review read the letter →

arxiv 2607.07652 v1 pith:FCO2JHOV submitted 2026-07-08 cs.CY cs.AIecon.GNq-fin.EC

Qiaoni Shi , Kai Zhu , Kai Gu This is my paper · ORCID
classification cs.CYcs.AIecon.GNq-fin.EC
keywords AIsearchwebtrafficdigitalintermediationattentioneconomyadvertisingclickstreamreferralratioChatGPT
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that AI search is structurally changing how attention flows on the web. For two decades, search engines routed users from queries to destination websites, and those visits were the economic lifeblood of online content—creating ad impressions, subscription conversions, and audience relationships. The authors use URL-level U.S. desktop clickstream data to show that ChatGPT resolves most information needs inside its own interface: only 5.2% of ChatGPT conversation sessions produce a clean outbound click, compared with 31.1% of Google queries. The residual traffic that does exit ChatGPT is not a scaled-down version of Google's referral stream—it skews toward specialized, non-profit, and developer destinations and away from ad-supported sites. Using three natural experiments in which ChatGPT Search access expanded to new user groups, the authors estimate that wider access reduces traditional search queries by 9.4% on average and 17.0% after twenty weeks, with the largest losses concentrated in informational content categories like academic research, reference, and news. The paper frames this as a shift from routed visits to residual referrals: the intermediary now answers the question itself rather than sending the user onward, which weakens the implicit bargain that has linked search, traffic, and content production on the open web.

What carries the argument

The paper uses three empirical components. First, a within-household, within-week comparison of ChatGPT and Google referral ratios using Comscore U.S. desktop clickstream data (October 2024–July 2025), with household and week fixed effects to absorb persistent differences. Second, a stacked difference-in-differences design exploiting three ChatGPT Search access expansions (paid subscribers October 31 2024, free logged-in users December 16 2024, anonymous browsers February 5 2025), with each treated cohort matched to a reweighted control of households with no pre-expansion ChatGPT or Claude activity. Third, a domain classification system (4,266 destinations labeled by content type and monetiz

What would settle it

If traditional search queries did not decline after ChatGPT Search access expanded—i.e., if the event-study coefficients were zero or positive post-treatment—the displacement claim would fail. Alternatively, if the referral-ratio gap between ChatGPT and Google disappeared under a different session definition or attribution window, the absorption claim would weaken substantially.

Watch

Extended reading notes

Core claim

The central object the paper identifies is the referral ratio—the share of information-seeking occasions at an intermediary that produce at least one clean outbound click to a third-party website. By measuring this symmetrically for ChatGPT conversation sessions (5.2%) and Google queries (31.1%), the authors establish that AI search absorbs roughly six times more information needs internally than traditional search does. This is not a welfare claim about whether users are better served; it is a traffic-allocation claim about where observable attention ends. The paper then connects this absorption to downstream substitution: when households gain ChatGPT Search access, their traditional search

Load-bearing premise

The parallel-trends assumption: that households gaining ChatGPT Search access and households without it would have followed similar search-query trajectories absent the expansion. The preferred control group consists of households with no pre-expansion ChatGPT or Claude activity, who may differ from treated households in unobserved ways related to search trends. The gap between the preferred estimate (17.0% long-run displacement) and a cleaner comparison between two groups of

Editorial extensions

If this is right

  • If the referral-ratio gap persists, ad-supported informational websites that depend on routed search traffic face a structural decline in the visits they can monetize, even if AI search continues to use their content as source material.
  • The finding that residual ChatGPT referrals avoid ad-supported sites by 27.6 percentage points suggests that the websites most dependent on search-driven attention are precisely those most bypassed by AI search's smaller click-out stream.
  • Search advertising inventory contracts as traditional queries fall by 9–17%, which would affect pricing and channel allocation in search-ad markets before any equilibrium adjustment in bids or budgets.
  • The category-specific pattern—academic research referrals down 32.8%, reference/knowledge down 26.5%—identifies which content producers face the most acute exposure in licensing and attribution negotiations with AI intermediaries.
  • The robots.txt finding that 79% of classified domains block at least one AI training crawler, yet blocking does not reduce runtime referrals, means that existing opt-out mechanisms do not protect content producers from the traffic reallocation the paper documents.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If AI search absorption rates hold or increase as models improve, the web's traffic-based attribution system becomes increasingly incomplete: a growing share of information needs are satisfied without any observable visit that a website can count, monetize, or convert. This creates a measurement problem for the entire digital advertising and publishing ecosystem, not just for search engines.
  • The gap between the preferred estimate (17.0% displacement at 20 weeks) and the cleaner within-adopter comparison (8.2%) suggests the true causal effect may lie between these bounds. If so, the paper's headline displacement figure may overstate the effect for policy purposes while still confirming the direction.
  • The finding that different households reach different specialty destinations through ChatGPT (aggregate dispersion) while individual households concentrate on few destinations implies that AI search may fragment web audiences in ways that undermine the network effects large destination sites have relied on, without necessarily benefiting smaller sites in aggregate.
  • If content producers respond to declining routed traffic by reducing investment in informational content—the categories most affected—the quality of information available to AI search systems themselves may degrade, creating a feedback loop the paper identifies as its central open question.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. This paper uses URL-level Comscore U.S. desktop clickstream data (October 2024–July 2025) to study how AI search changes web traffic allocation. It makes three contributions: (1) a descriptive comparison showing ChatGPT produces clean outbound referrals in only 5.2% of conversation sessions versus 31.1% of Google queries, with the gap persisting within household-week; (2) an analysis of the composition of ChatGPT's residual referral traffic, showing it skews toward specialized, non-ad-supported destinations; and (3) a stacked difference-in-differences design exploiting three ChatGPT Search access expansions (October 31, 2024; December 16, 2024; February 5, 2025) estimating that wider access reduces traditional search queries by 9.4% on average and 17.0% after twenty weeks, with losses concentrated in informational categories. The paper is transparent about the descriptive nature of the referral-ratio comparison and about the gap between its preferred displacement estimate and a cleaner within-adopter comparison.

Significance. The paper addresses a timely and economically important question: whether AI search structurally reallocates web attention away from the routed-visit economy. Its strengths include a within-household-week comparison design that controls for persistent household differences and common calendar shocks; a stacked DiD with externally timed treatment (access expansions) rather than endogenous adoption; formal parallel-trends tests that pass (joint Wald p=0.161 and p=0.743); ACS reweighting for population representativeness; a blind human validation of the domain classifier (Cohen's κ = 0.76 content type, 0.82 monetization); extensive robustness sweeps across session gaps, window lengths, contamination rules, dominance thresholds, classifier confidence, and control definitions; reconciliation with industry benchmarks; and an estimand-ladder framework (Lundberg et al. 2021) linking theoretical target to regression. The L-vs-A comparison that differences out selection into ChatGPT use is a valuable falsifiability check. The robots.txt analysis showing that opt-out does not bind the referral margin is a useful supplementary finding.

major comments (3)
  1. Section 3.3, Table 4: The gap between the preferred three-event estimate (−17.0% at w≥20) and the cleaner December-only L-vs-A comparison (−8.2% at w≥20) is substantial—roughly a factor of two. The paper attributes this to residual selection in the Nitt control and reports both estimates, which is commendable. However, the abstract and headline claims feature only the preferred (larger) estimate. Given that the L-vs-A design is explicitly described as the 'cleanest single comparison' (Section 3.3) and differences out the selection concern the paper itself raises, the authors should clarify in the main text why the preferred estimate is the primary headline figure rather than the more conservative L-vs-A estimate, or at minimum present both estimates with equal prominence in the abstract. This is load-bearing because the magnitude of displacement is central to the paper's economic-bargain
  2. Section 3.1 and the economic-bargain framing: The paper acknowledges that the 5.2% vs 31.1% referral-ratio comparison 'does not separate differences in the tasks brought to each intermediary from differences created by the intermediaries' (Section 3.1). The surrounding-context fixed effect check (Table OA.2.4) moves the coefficient from −0.302 to −0.300, which the paper interprets as evidence that broad task context does not explain the gap. However, as the paper itself notes, this control captures only the dominant content category of nearby browsing—a coarse proxy that cannot distinguish, e.g., a quick factual lookup from a multi-step research task within the same 'reference/knowledge' bucket. The +23.1pp 'solo' share for ChatGPT is consistent with users bringing harder, more self-contained tasks to ChatGPT. The paper frames the result correctly as descriptive, but the abstract and the
  3. Section 3.3, Figure 8: The category-level displacement results (e.g., −32.8% for academic research, −26.5% for reference/knowledge) are described as 'descriptive heterogeneity within the preferred design, not a separately identified mechanism.' This is appropriate, but the Discussion (Section 4) leans on these category-level patterns to connect retention inside ChatGPT to downstream losses in routed traffic. The connection between the descriptive referral-ratio findings and the causal displacement estimates is suggestive but not formally tested. The authors should clarify that the category-level displacement results are correlational with the intent patterns, not evidence that retention causes displacement within those categories.
minor comments (6)
  1. The abstract states 'ChatGPT produces outbound clicks in only 5.2% of conversation sessions' without specifying the denominator unit (conversation session vs. query). Adding 'of conversation sessions' would improve precision.
  2. Figure 6: The repeated domain labels (e.g., 'python.org' appearing 17 times) make the figure difficult to read. Consider using point markers without text labels, or labeling only the most extreme domains.
  3. Section OA.1.3, footnote 5: The f/conversation endpoint is described as 'not separately documented in public reverse-engineering sources.' The paper should note the risk that this endpoint's behavior may change, potentially affecting the July 2025 message-level referral ratio (Table OA.2.2).
  4. Table OA.6.4: The $150–200K income cell has a post-rake SMD of 0.016, which the paper notes is the binding constraint on ESS. It would help to state the effective sample size (NESS/N=0.81) in the main text rather than only in the table notes, as it bears on the precision of the displacement estimates.
  5. The paper uses 'Nitt' and 'Nitt' interchangeably (e.g., Section 2 vs. Appendix OA.6.1). Standardize the notation.
  6. Section 3.2: The statement 'ChatGPT's referral ratio is highest in developer/technical contexts (13.4%)' could note that this is still well below Google's per-query ratio in the same context (68.1% from Table OA.4.1), to give readers a sense of the within-context gap.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive report. All three major comments are well-taken and will be addressed in the revised manuscript. Two require revisions to the abstract and framing (Comments 1 and 2); one requires a clarifying statement in the Discussion (Comment 3). No standing objections remain.

read point-by-point responses

  1. Referee: Section 3.3, Table 4: The gap between the preferred three-event estimate (−17.0% at w≥20) and the cleaner December-only L-vs-A comparison (−8.2% at w≥20) is substantial—roughly a factor of two. The paper attributes this to residual selection in the Nitt control and reports both estimates, which is commendable. However, the abstract and headline claims feature only the preferred (larger) estimate. Given that the L-vs-A design is explicitly described as the 'cleanest single comparison' (Section 3.3) and differences out the selection concern the paper itself raises, the authors should clarify in the main text why the preferred estimate is the primary headline figure rather than the more conservative L-vs-A estimate, or at minimum present both estimates with equal prominence in the abstract.

    Authors: The referee is correct that the abstract features only the preferred (larger) estimate and that this choice is load-bearing for the paper's economic-bargain framing. We will revise the abstract to present both estimates with equal prominence. Specifically, the abstract will report that wider access cuts search use by 9.4% on average (17.0% after twenty weeks) under the preferred three-event design, and note that the cleaner December-only L-vs-A comparison—which differences out selection into ChatGPT use—yields a smaller but directionally consistent 8.2% decline after twenty weeks. We will also add a brief sentence in Section 3.3 explaining why the preferred design serves as the primary specification: it pools three externally timed access expansions, providing substantially greater statistical power and a longer common-support window, while the L-vs-A comparison trades power for cleaner identification. The paper already reports both estimates transparently in Table 4 and discusses the selection concern; the revision ensures the abstract and headline framing reflect that transparency. revision: yes

  2. Referee: Section 3.1 and the economic-bargain framing: The paper acknowledges that the 5.2% vs 31.1% referral-ratio comparison 'does not separate differences in the tasks brought to each intermediary from differences created by the intermediaries' (Section 3.1). The surrounding-context fixed effect check (Table OA.2.4) moves the coefficient from −0.302 to −0.300, which the paper interprets as evidence that broad task context does not explain the gap. However, as the paper itself notes, this control captures only the dominant content category of nearby browsing—a coarse proxy that cannot distinguish, e.g., a quick factual lookup from a multi-step research task within the same 'reference/knowledge' bucket. The +23.1pp 'solo' share for ChatGPT is consistent with users bringing harder, more self-contained tasks to ChatGPT. The paper frames the result correctly as descriptive, but the abstract and the

    Authors: The referee's point is well taken. The surrounding-context fixed effect is indeed a coarse proxy: it controls for the dominant content category of nearby browsing but cannot distinguish task complexity within a category. The +23.1pp solo share is consistent with the interpretation that users bring harder, more self-contained tasks to ChatGPT, and the current framing does not adequately flag this as a specific, uncontrolled form of task selection. We will make two changes. First, we will add a sentence to Section 3.1 explicitly acknowledging that the solo-share pattern is consistent with task-complexity selection that the surrounding-context control cannot address, and that this is a specific limitation of the descriptive comparison. Second, we will adjust the abstract to qualify the referral-ratio comparison as descriptive and note that task-selection differences—including the possibility that users bring more self-contained tasks to ChatGPT—cannot be fully ruled out. The paper already states that the result 'does not separate differences in the tasks brought to each intermediary from differences created by the intermediaries'; the revision makes this caveat more specific and ensures the abstract carries it. revision: yes

  3. Referee: Section 3.3, Figure 8: The category-level displacement results (e.g., −32.8% for academic research, −26.5% for reference/knowledge) are described as 'descriptive heterogeneity within the preferred design, not a separately identified mechanism.' This is appropriate, but the Discussion (Section 4) leans on these category-level patterns to connect retention inside ChatGPT to downstream losses in routed traffic. The connection between the descriptive referral-ratio findings and the causal displacement estimates is suggestive but not formally tested. The authors should clarify that the category-level displacement results are correlational with the intent patterns, not evidence that retention causes displacement within those categories.

    Authors: The referee is correct. The Discussion (Section 4) draws a suggestive connection between the descriptive finding that informational tasks are prevalent in ChatGPT use and often retained, and the causal finding that informational categories show the largest search-referral losses. This connection is not formally tested: we do not estimate a mediation or mechanism model linking retention to displacement within categories. The category-level displacement results are heterogeneity within the preferred DiD design, and their alignment with the intent patterns is correlational. We will add an explicit clarifying statement to Section 4 stating that the category-level patterns are consistent with—but do not formally identify—a mechanism by which retention inside ChatGPT causes downstream referral losses in the same categories. We will also adjust the relevant sentence in Section 3.3 to make clear that the heterogeneity is correlational with the intent patterns rather than evidence of a causal link. revision: yes

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: empirical paper with externally timed treatment and descriptive measurements not derived from fitted parameters.

full rationale

This is an empirical paper whose central claims rest on externally generated variation (OpenAI's announced access expansions on three specific dates) and descriptive clickstream measurements (referral ratios computed from observed HTTP referrers). The referral-ratio comparison (5.2% vs 31.1%) is a direct descriptive measurement from clickstream data, not a quantity derived from a fitted parameter that is then 'predicted.' The displacement estimate (9.4%) comes from a stacked difference-in-differences design where treatment timing is determined by platform policy dates, not by the authors' model or fitted values. The domain classification uses GPT-4o as a measurement instrument with blind human validation (Cohen's κ = 0.76/0.82), which is an external check on the labeling, not a self-citation chain. The paper does cite Padilla et al. (2025) for the individual-adoption design implementation, but that is a methodological reference for a robustness check, not the load-bearing argument. The preferred design's identification comes from the access expansions themselves. No step in the derivation chain reduces to its inputs by construction: the referral ratio is a count-based proportion, the DiD coefficient is estimated from panel variation with externally timed treatment, and the composition results are descriptive shares. The paper's limitations (parallel-trends assumption, task-selection in the descriptive comparison) are identification concerns, not circularity. The one minor self-citation is Zhu (2026) on robots.txt adverse selection, which supports a supplementary finding (Section OA.5.5) and is not load-bearing for the central claims.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

No new entities are postulated. The paper measures observable browsing behavior using established data infrastructure and methods.

free parameters (7)
  • Session gap threshold = 3600 seconds (1 hour)
    Defines when same-UUID loads open a new conversation session. Sensitivity-tested across 0.25h–5h (Table OA.1.4); estimates move little.
  • Panel foreground threshold (τ_fg) = 4
    Minimum foreground records per month for balanced sub-panel. Sensitivity-tested from 1 to 50 (Table OA.1.1).
  • Dominance threshold (τ) = 0.50
    Minimum share for a content type to dominate surrounding context. Swept from 0.3 to 0.6 (Table OA.4.1).
  • Classifier confidence cutoff = 0.90
    Restricts content-type analyses to high-confidence domain labels. Relaxation to 0.60 shifts magnitudes ≤1.5pp (Section OA.3.3).
  • Cohort dominance threshold (τ_dom) = 0.70
    Minimum API/anon share for cohort classification. Sensitivity-tested from 0.60 to 0.95 (Table OA.6.3).
  • Domain rank cutoff (K) = 2500
    Top-K per side for matched-support domain universe. Joint Pareto optimum covering ~70% of referrals (Table OA.3.2).
  • Surrounding-context window = ±15 minutes
    Window for browsing context around anchors. Also tested at ±30 minutes (Section OA.4.2).
assumptions (5)
  • domain assumption Parallel trends: absent ChatGPT Search access, treated and control households would have followed parallel search-query paths.
    Identifying assumption for the stacked DiD (Section OA.6.3, Equation OA.6.1). Tested via joint Wald test on pre-treatment leads (Table OA.6.6).
  • domain assumption Comscore desktop panel is a valid measurement instrument for U.S. browsing behavior.
    Underlies all measurements. ACS reweighting (Section OA.6.2) addresses demographic representativeness but not the desktop-only scope.
  • domain assumption HTTP referrers reliably identify the source intermediary for outbound clicks.
    Referral measurement depends on referrer headers. Paper notes browsers may strip AI referrers and argues the clean-referral filter addresses this (Section OA.2.1, Table OA.2.1).
  • domain assumption ChatGPT endpoint semantics inferred from community reverse-engineering are stable and accurate.
    Conversation sessions, search-enabled proxies, and cohort classification depend on endpoint URL patterns documented in public reverse-engineering sources (Section OA.1.3).
  • domain assumption GPT-4o domain classification produces labels comparable to human coding.
    Validated against blind human coding with κ=0.76 (content type) and 0.82 (monetization) on a 100-domain stratified sample (Section OA.3.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Answering Without Referring: How AI Search Rewrites the Web's Economic Bargain." pith.science (2026). https://pith.science/paper/FCO2JHOV

@misc{pith2026260707652,
  author       = {Pith},
  title        = {Pith review of: Answering Without Referring: How AI Search Rewrites the Web's Economic Bargain},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FCO2JHOV}},
  note         = {Machine review of arXiv:2607.07652}
}
read the original abstract

Search engines have long allocated attention on the web by routing users from queries to websites. AI search changes this arrangement because information needs can be resolved inside the intermediary. Using URL-level Comscore U.S. desktop clickstream, we compare ChatGPT and Google information-seeking occasions and exploit ChatGPT Search access expansions to estimate traditional search displacement. ChatGPT produces outbound clicks in only 5.2% of conversation sessions, far below Google's referral ratio. The remaining clicks are not a scaled-down Google stream: they skew toward specialized destinations and away from ad-supported sites. Wider access cuts search use by 9.4%, with search-referral losses largest for informational categories. Our findings identify a central economic shift in digital intermediation: AI search might satisfy information needs inside the intermediary while weakening the referral bargain that has linked search, traffic, and content production on the open web.

Figures

Figures reproduced from arXiv: 2607.07652 by the authors.

Figure 3
Figure 3. ChatGPT routes far fewer information-seeking occasions than Google under every session definition. Paid Logged-inNo Login 1 h (primary) 0.5 h 5 h 0.0% 2.0% 4.0% 6.0% 2024-11 2025-01 2025-03 2025-05 2025-07 Referral ratio (a) ChatGPT conversation sessions. Paid Logged-in No Login Bing Google Yahoo 0% 10% 20% 30% 40% 2024-11 2025-01 2025-03 2025-05 2025-07 Referral ratio (b) Search-engine queries. Notes: Points report… view at source ↗
Figure 7
Figure 7. Wider ChatGPT Search access reduces traditional search queries. -10 -5 0 5 -11 -7 -3 1 5 9 13 17 21 25 29 Weeks relative to ChatGPT-Search rollout ATT per user-week Notes: The outcome is weekly Google, Bing, and Yahoo query loads per household. The stacked difference-in-differences design pools cohorts gaining access on October 31, December 16, and February 5 and compares each with the ACS-reweighted Nitt control. E… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages

  1. [1]

    Athey S, Ellison G (2011) Position auctions with consumer search. Quart. J. Econom. 126(3):1213--1270, ://dx.doi.org/10.1093/qje/qjr028

    work page doi:10.1093/qje/qjr028 2011

  2. [2]

    Management Sci

    Bakos JY (1997) Reducing buyer search costs: Implications for electronic marketplaces. Management Sci. 43(12):1676--1692, ://dx.doi.org/10.1287/mnsc.43.12.1676

    work page doi:10.1287/mnsc.43.12.1676 1997

  3. [4]

    Brynjolfsson E, Li D, Raymond LR (2025) Generative AI at work. Quart. J. Econom. 140(2):889--942, ://dx.doi.org/10.1093/qje/qjae044

    work page doi:10.1093/qje/qjae044 2025

  4. [5]

    Burtch G, Lee D, Chen Z (2024) The consequences of generative AI for online knowledge communities. Sci. Rep. 14(1):10413, ://dx.doi.org/10.1038/s41598-024-61221-0

    work page doi:10.1038/s41598-024-61221-0 2024

  5. [6]

    Callaway B, Sant'Anna PHC (2021) Difference-in-differences with multiple time periods. J. Econometrics 225(2):200--230, ://dx.doi.org/10.1016/j.jeconom.2020.12.001

    work page doi:10.1016/j.jeconom.2020.12.001 2021

  6. [7]

    Cengiz D, Dube A, Lindner A, Zipperer B (2019) The effect of minimum wages on low-wage jobs. Quart. J. Econom. 134(3):1405--1454, ://dx.doi.org/10.1093/qje/qjz014

    work page doi:10.1093/qje/qjz014 2019

  7. [9]

    Chiou L, Tucker C (2017) Content aggregation by platforms: The case of the news media. J. Econom. Management Strategy 26(4):782--805, ://dx.doi.org/10.1111/jems.12207

    work page doi:10.1111/jems.12207 2017

  8. [10]

    PNAS Nexus 3(9):pgae400, ://dx.doi.org/10.1093/pnasnexus/pgae400

    del Rio-Chanona RM, Laurentsyeva N, Wachs J (2024) Large language models reduce public knowledge sharing on online Q&A platforms. PNAS Nexus 3(9):pgae400, ://dx.doi.org/10.1093/pnasnexus/pgae400

    work page doi:10.1093/pnasnexus/pgae400 2024

Show all 48 references
  1. [11]

    Edelman B, Ostrovsky M, Schwarz M (2007) Internet advertising and the generalized second-price auction: Selling billions of dollars worth of keywords. Amer. Econom. Rev. 97(1):242--259, ://dx.doi.org/10.1257/aer.97.1.242

    2007 doi

  2. [12]

    Working paper, Knight-Georgetown Institute, Washington, DC, ://dx.doi.org/10.2139/ssrn.6238578

    Gholami S, Firullo C, Cheyre C, Acquisti A (2026) Beyond search: LLM adoption and web traffic concentration. Working paper, Knight-Georgetown Institute, Washington, DC, ://dx.doi.org/10.2139/ssrn.6238578

    2026 doi

  3. [13]

    Management Sci

    Ghose A, Yang S (2009) An empirical analysis of search engine advertising: Sponsored search in electronic markets. Management Sci. 55(10):1605--1622, ://dx.doi.org/10.1287/mnsc.1090.1054

    2009 doi

  4. [14]

    Goldfarb A, Tucker C (2019) Digital economics. J. Econom. Literature 57(1):3--43, ://dx.doi.org/10.1257/jel.20171452

    2019 doi

  5. [15]

    Jeon DS, Nasr N (2016) News aggregators and competition among newspapers on the internet. Amer. Econom. J.: Microeconom. 8(4):91--114, ://dx.doi.org/10.1257/mic.20140151

    2016 doi

  6. [16]

    ://dx.doi.org/10.1287/mksc.2025.0489, ePub ahead of print April 21

    Kaiser M, Schulze C (2026) Frontiers: ChatGPT referrals to e-commerce websites: How do LLMs compare against traditional channels? Marketing Sci. ://dx.doi.org/10.1287/mksc.2025.0489, ePub ahead of print April 21

    2026 doi

  7. [17]

    Biometrics 33(1):159--174, ://dx.doi.org/10.2307/2529310

    Landis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33(1):159--174, ://dx.doi.org/10.2307/2529310

    1977 doi

  8. [18]

    Lundberg I, Johnson R, Stewart BM (2021) What is your estimand? defining the target quantity connects statistical evidence to theory. Amer. Sociol. Rev. 86(3):532--565, ://dx.doi.org/10.1177/00031224211004187

    2021 doi

  9. [19]

    Science 381(6654):187--192, ://dx.doi.org/10.1126/science.adh2586

    Noy S, Zhang W (2023) Experimental evidence on the productivity effects of generative artificial intelligence. Science 381(6654):187--192, ://dx.doi.org/10.1126/science.adh2586

    2023 doi

  10. [20]

    ://openai.com/index/introducing-chatgpt-search/, accessed June 21, 2026

    OpenAI (2024) Introducing ChatGPT search. ://openai.com/index/introducing-chatgpt-search/, accessed June 21, 2026

    2024

  11. [21]

    Working paper, London Business School, London, ://ssrn.com/abstract=5393256

    Padilla N, Lam HT, Lambrecht A, Hollenbeck B (2025) The impact of LLM adoption on online user behavior. Working paper, London Business School, London, ://ssrn.com/abstract=5393256

    2025

  12. [22]

    Greenberger M, ed., Computers, Communications, and the Public Interest, 37--72 (Baltimore, MD: The Johns Hopkins Press)

    Simon HA (1971) Designing organizations for an information-rich world. Greenberger M, ed., Computers, Communications, and the Public Interest, 37--72 (Baltimore, MD: The Johns Hopkins Press)

    1971

  13. [23]

    Internat

    Varian HR (2007) Position auctions. Internat. J. Indust. Organ. 25(6):1163--1178, ://dx.doi.org/10.1016/j.ijindorg.2006.10.002

    2007 doi

  14. [24]

    Working paper, University of Pennsylvania, Philadelphia, ://ssrn.com/abstract=5992774

    Zhao H, Berman R (2026) Strategic response of news publishers to generative AI . Working paper, University of Pennsylvania, Philadelphia, ://ssrn.com/abstract=5992774

    2026

  15. [25]

    Working paper, Bocconi University, Milan, ://ssrn.com/abstract=6438640

    Zhu K (2026) Adverse selection in the AI data commons. Working paper, Bocconi University, Milan, ://ssrn.com/abstract=6438640

    2026

  16. [26]

    , title =

    Simon, Herbert A. , title =. Computers, Communications, and the Public Interest , publisher =. 1971 , pages =

    1971

  17. [27]

    Goldfarb, Avi and Tucker, Catherine , title =. J. Econom. Literature , year =

  18. [28]

    Yannis , title =

    Bakos, J. Yannis , title =. Management Sci. , year =

  19. [29]

    Athey, Susan and Ellison, Glenn , title =. Quart. J. Econom. , year =

  20. [30]

    Edelman, Benjamin and Ostrovsky, Michael and Schwarz, Michael , title =. Amer. Econom. Rev. , year =

  21. [31]

    , title =

    Varian, Hal R. , title =. Internat. J. Indust. Organ. , year =

  22. [32]

    Management Sci

    Ghose, Anindya and Yang, Sha , title =. Management Sci. , year =

  23. [33]

    Chiou, Lesley and Tucker, Catherine , title =. J. Econom. Management Strategy , year =

  24. [34]

    Jeon, Doh-Shin and Nasr, Nikrooz , title =. Amer. Econom. J.: Microeconom. , year =

  25. [35]

    Science , year =

    Noy, Shakked and Zhang, Whitney , title =. Science , year =

  26. [36]

    , title =

    Brynjolfsson, Erik and Li, Danielle and Raymond, Lindsey R. , title =. Quart. J. Econom. , year =

  27. [37]

    , title =

    Bick, Alexander and Blandin, Adam and Deming, David J. , title =. 2024 , month =. doi:10.3386/w32966 , url =

    2024 doi

  28. [38]

    and Hitzig, Zoe and Ong, Christopher and Shan, Carl Yan and Wadman, Kevin , title =

    Chatterji, Aaron and Cunningham, Thomas and Deming, David J. and Hitzig, Zoe and Ong, Christopher and Shan, Carl Yan and Wadman, Kevin , title =. 2025 , month =. doi:10.3386/w34255 , url =

    2025 doi

  29. [39]

    Burtch, Gordon and Lee, Dokyun and Chen, Zhichen , title =. Sci. Rep. , year =

  30. [40]

    Maria and Laurentsyeva, Nadzeya and Wachs, Johannes , title =

    del Rio-Chanona, R. Maria and Laurentsyeva, Nadzeya and Wachs, Johannes , title =. PNAS Nexus , year =

  31. [41]

    Tai and Lambrecht, Anja and Hollenbeck, Brett , title =

    Padilla, Nicolas and Lam, H. Tai and Lambrecht, Anja and Hollenbeck, Brett , title =. 2025 , howpublished =

    2025

  32. [42]

    Marketing Sci

    Kaiser, Maximilian and Schulze, Christian , title =. Marketing Sci. , year =

  33. [43]

    2026 , doi =

    Gholami, Samira and Firullo, Cristiana and Cheyre, Cristobal and Acquisti, Alessandro , title =. 2026 , doi =

    2026

  34. [44]

    2026 , howpublished =

    Zhao, Hangcheng and Berman, Ron , title =. 2026 , howpublished =

    2026

  35. [45]

    2024 , note =

    Introducing. 2024 , note =

    2024

  36. [46]

    Cengiz, Doruk and Dube, Arindrajit and Lindner, Attila and Zipperer, Ben , title =. Quart. J. Econom. , year =

  37. [47]

    Callaway, Brantly and Sant'Anna, Pedro H. C. , title =. J. Econometrics , year =

  38. [48]

    2026 , url =

    Zhu, Kai , title =. 2026 , url =

    2026

  39. [49]

    , title =

    Lundberg, Ian and Johnson, Rebecca and Stewart, Brandon M. , title =. Amer. Sociol. Rev. , year =

  40. [50]

    Richard and Koch, Gary G

    Landis, J. Richard and Koch, Gary G. , title =. Biometrics , year =

Pith tools

Reviewed July 9, 2026 · model on record in the stance chip above.