London research

14 reading notes on strategy, consulting, operations and AI, with original sources and explicit reading limits.

Notes by Brief & Field. Source access checked 8 October 2026; this layout updated 9 October 2026. Linked versions, reading scope and evidence limits are retained below.

Strategy

LBS · Free book chapter · 2017 · LBS01

The New Business Road Test, fifth edition: Chapter 1 — Our opportunity: why will or won’t this work?

John Mullins

London Business School (author and official host); Pearson Education (publisher)
2017 (copyright of this free chapter; exact publication day not stated)

Open original source Free chapter 1; the full book is paid.

B&F reading note

A useful starting point for assessing a new service before investing heavily in it. Mullins separates the customer opportunity from competitive economics and the team’s ability to deliver. The framework encourages investigation of weaknesses as well as attractive features, without treating the result as a simple score. This is a free sample chapter, not an experimental validation. Apply it to a specific buyer and problem, then test the assumptions with evidence.

Question for your work: Which assumption about the buyer, economics or team could make this service fail?

Application: What evidence would make you reject this opportunity before investing in it?

Evidence and limitations

Synthesises interviews with investors and entrepreneurs plus cases. This sample does not give the underlying sample size or validation design.

A practical framework, not a causal trial, demand estimate or validated scoring instrument. Full book is paid; this chapter is free.

Reading scope and source locations

Passages examined: PDF pp. 2 and 5–26: permission, complete chapter 1. The rest of the book and research appendix were not available/read.
Where to look: PDF pp. 8–10 (seven domains); 13–14 (first segment); 17–23 (economics, team, evidence and non-additive assessment). Printed pp. 6–8, 11–12, 15–21.

LSE · Evidence review · 2020 revision · LSE04

Unlocking SME productivity: Review of recent evidence and implications for the UK's Industrial Strategy

Isabelle Roland

Centre for Economic Performance, London School of Economics and Political Science; author also affiliated with University of Oxford
2020-01 revised PDF, replaces September 2018 version; CEPIS05 catalogue dated 2018-10-02

Open original source Free institutional research review.

B&F reading note

Small businesses do not all face the same growth problem. This LSE policy review separates financing constraints from management capabilities and explains why both deserve attention. Technology adoption can depend on the ability to monitor work, set useful targets and change routines. The review also cautions against treating size or age as reliable indicators of growth potential. It is a historical synthesis of varied evidence, useful for framing a diagnosis rather than predicting a client's financial return.

Question for your work: Is the next constraint better technology, clearer management routines or something else?

Application: What happened in the most recent real case, and what prevented the team from acting?

Evidence and limitations

Narrative review of UK SME finance and management evidence. Draws on WMS/MES surveys, cited experiments and observational manufacturing research. No single sample or pooled denominator; small firms in cash-saving analysis means below-median assets.

Historical UK policy context and heterogeneous cited evidence; no new consulting trial. Some evidence overlaps the MES family used by LSE02.

Reading scope and source locations

Passages examined: Cover/version, abstract, executive summary, full sections 3-4; Table 2 visually checked. The finance section was also reviewed.
Where to look: Sections 3.1-3.2, printed pp.12-16 (PDF 14-18); Table 2, printed p.15 (PDF 17); section 4, printed pp.16-17 (PDF 18-19).

LBS · Case-based book sample · 2014 · LON04

The Customer-Funded Business: Start, Finance, or Grow Your Company with Your Customers’ Cash — public sample and About the Research

John Mullins

London Business School / Wiley
2014

Open original source Free official sample and research appendix; the full book is paid.

B&F reading note

Payment timing is part of a business model, not just an administrative detail. Mullins distinguishes advance payments, subscriptions and other ways customers can finance delivery, including a gradual move from services to products. The sample also explains why cash arriving early does not establish profitability. Its framework comes from selected company cases and interviews, not a pricing experiment. Use it to examine a small offer’s cash needs, delivery obligations and evidence of paid demand.

Question for your work: When does cash arrive, what must we deliver for it, and is the work still profitable?

Application: Map the payment milestone, accepted output, full delivery effort and remaining obligation.

Evidence and limitations

Inductive case selection: more than 40 initial candidate companies narrowed to nearly 20, with entrepreneur and investor interviews; research phase began in 2012. Non-random, not a pricing experiment.

Selected historical cases, no counterfactual or representative success rate. Does not establish feasible prices, deposit levels, retention or willingness to pay for B&F.

Reading scope and source locations

Passages examined: PDF pp. 13–14, 17, 19–20, 27–34, 48–50 and 56–58; model definitions, selected examples, pitfalls and complete About the Research appendix. The remaining sample and paid full book were not read. LBS publication announcement also read.
Where to look: Chapter 1, printed pp. 10–17 (PDF 27–34), including Table 1.1; printed pp. 31–33 (PDF 48–50); About the Research, printed pp. 267–269 (PDF 56–58).

LSE · Observational research · 2009 · LON05

Specialization, Firms, and Markets: The Division of Labor within and between Law Firms

Luis Garicano, Thomas N. Hubbard

LSE / Northwestern University
2009 issue; online 7 May 2008

Open original source Free full published PDF on the coauthor’s Northwestern faculty page.

B&F reading note

A specialist person does not necessarily belong in a narrowly specialised firm. This study of US law offices distinguishes individual expertise from the range of services organised inside one business. The observed patterns differ across fields and partly support the importance of referrals and coordination. They do not prove that a boutique should diversify or specialise. Use the distinction to ask which capabilities need close collaboration and which can be supplied through a clearly managed external relationship.

Question for your work: Which expertise belongs inside the team, and which can be coordinated reliably outside it?

Application: Which specialist is confirmed, what must they receive, and who checks the handoff?

Evidence and limitations

1992 US Census: 26,151 law offices, 219,033 lawyers; main small-market regressions use 6,032 offices. Weighted grouped-data regressions compare specialization with county market size, controlling for demand composition; observational, not a randomized test.

Historical US legal markets, geographic market assumptions and ambiguous expertise classification. No boutique pricing or current SME-demand test; authors do not emphasize a causal interpretation of the main coefficient ratio.

Reading scope and source locations

Passages examined: PDF pp. 1–3, 12–23, 25 and 28–30: framing, data and identification, main results, sensitivity summary and conclusions. Full theoretical section, remaining tables and appendices not independently reanalysed. LSE repository record read. Confidential microdata unavailable; no replication.
Where to look: Data and sample: printed pp. 350–356 (PDF 12–18), Table 3. Specifications and main results: pp. 357–363 (PDF 19–25), especially Table 6 and footnote 26 (PDF 21). Mechanism caveats and conclusion: pp. 366–368 (PDF 28–30).

LSE publication record

LBS · Theory & simulation · 2018 · LBS02

A basic theory of inheritance: How bad practice prevails

Freek Vermeulen

London Business School; Strategic Management Journal / Wiley
2018

Open original source Free accepted manuscript.

B&F reading note

Widely used practices can survive even when they weaken the organisations adopting them. Vermeulen develops a theory and simulation explaining how apparent success, uncertain causation and rapid imitation can sustain such practices. This does not establish that any particular consulting method is harmful. It provides a reason to ask what evidence supports a routine, how its effects will be measured and whether removing it might serve the client better than adding another tool.

Question for your work: Are we adopting this practice because it works, or because successful firms appear to use it?

Application: Which routine could you remove, and how would you check whether the work improved?

Evidence and limitations

Conceptual theory, earlier IVF research as illustration and a stylised simulation of 1,000 firms. These are simulated firms, not a sampled business population.

A possibility argument under assumptions; no measured prevalence of harmful practices or experiment in boutiques. Accepted manuscript may differ from published version.

Reading scope and source locations

Passages examined: PDF pp. 1–6, 16–28 and 41–42: metadata, theory framing, illustration, simulation, discussion and notes. Not all background/references or figure details read; simulation not reproduced.
Where to look: PDF pp. 2–3 (claim); 17–19 (simulation); 24–28 (implications/limits); p. 1 (version and terms). Published journal 39(6):1603–1629, DOI 10.1002/smj.2713.

UCL · Faculty essay · 2026 · LON03

The Hidden Architecture of Effective Groups

Colin M. Fisher

UCL School of Management
2026-09-29 (institutional index)

Open original source Free faculty article.

B&F reading note

Before adding meetings or coaching, examine how the work is arranged. Fisher highlights the goal, the task, the people involved and the norms governing their collaboration. This is a practical starting point for a small project team: make the output, dependencies and review responsibility clear. The article synthesizes earlier research; it does not establish a universal staffing formula or guarantee that a particular team structure will work.

Question for your work: Does the team have the structure the task needs?

Application: Name the output, its owner, its dependencies and its reviewer before assigning the work.

Evidence and limitations

Author synthesis drawing on prior team research and his book. The linked underlying studies were not independently reanalysed in this research pass.

A synthesis rather than a new trial of a small consultancy; no universal staffing ratio or guaranteed outcome.

Reading scope and source locations

Passages examined: Full substantive public article and its reference links. Underlying studies were not independently reanalysed.
Where to look: Why is structure so important? and Structure first, then process sections; publication date from UCL Research and insights index.

Client development

LSE · Randomized experiment · 2024 · LSE01

Does subsidising business advice improve firm performance? Evidence from a large RCT

Gonzalo Nunez-Chaim, Henry G. Overman, Capucine Riom

Centre for Economic Performance and What Works Centre for Local Economic Growth, London School of Economics and Political Science
2024-01-29; CEP Discussion Paper 1977, January 2024 version

Open original source Free working paper.

B&F reading note

This UK randomized voucher study suggests a short-term turnover benefit, with statistical uncertainty after adjustment for multiple tests. It found no statistically significant turnover benefit after two years or employment effect; that does not prove effects disappeared. Only a third of firms offered vouchers used them. For practitioners, the useful question is what changes after advice and whether that change persists. The study does not establish returns from any particular consultant or automation project.

Question for your work: How can we tell whether business advice creates a lasting operational change?

Application: Which changed action will you check after delivery, and when will you check it again?

Evidence and limitations

Randomized offers of subsidised advice to UK small businesses, with administrative and survey follow-up. Voucher offers and actual use are different treatment definitions.

Short-term turnover evidence weakens after multiple-testing adjustment. No statistically significant 24-month turnover or employment effect; this does not prove a zero effect. Subsidised advice does not establish a consultancy’s returns.

Reading scope and source locations

Passages examined: Abstract, introduction, programme and identification sections, data, section 5.1 and Table 2, conclusions, appendix A.1; Table 2 visually checked.
Where to look: Sections 2-5.1, printed pp.5-14 (PDF 6-15); Table 2, printed p.14 (PDF 15); conclusions, printed pp.25-26 (PDF 26-27); A.1, printed p.32 (PDF 33).

LSE · Quasi-experimental research · 2022 · LON06

Managerial input and firm performance. Evidence from a policy experiment

Francesco Manaresi, Alessandro Palma, Luca Salvatici, Vincenzo Scrutinio

Centre for Economic Performance, LSE
5 October 2022

Open original source Free CEP working paper; archived October 2022 version.

B&F reading note

An Italian policy study suggests specialized export advice can improve established firms' performance, with some gains emerging over several years. It compares applicants close to a funding cutoff rather than randomly assigning consultants. The practical lesson is to specify the capability being added, the client's readiness and the time needed for change; it does not establish the return from a particular consultancy.

Question for your work: Which missing capability would outside advice add, and when could its effect realistically become visible?

Application: Schedule separate checks for adoption of the process and the later business outcome.

Evidence and limitations

Italy: EUR10,000 export-consulting vouchers; 6-12-month services. Administrative data 2013-2019; intention-to-treat comparison within 30 seconds of funding cutoff. Table 1: 1,765/553 firms; sales regression: 15,315 firm-years.

Requires parallel trends; self-selected applicants. Provider matching and continuation evidence are noncausal. Long-term results may include subsequent advice. Not evidence of Czech demand or B&F returns. The pooled extra-EU export estimate was not statistically significant; later effects should not be read as immediate results.

Reading scope and source locations

Passages examined: Sections 2-3, 4.1-4.2, opening external-validity discussion and conclusions; Tables 1, 3-4 visually checked.
Where to look: Printed pp.5-17, 22-26 (PDF 6-18, 23-27); especially Table 4, PDF 27.

LBS · Practitioner essay · 2017 · LBS04

Unleash the power of your networks

John Dore

London Business School — Think
2017-01-30

Open original source Free practitioner article.

B&F reading note

A practical way to distinguish collecting contacts from developing useful relationships. Dore describes four approaches: deepen existing ties, expand through people already known, reach beyond the current circle and explore new networks. The article draws on work with relationship managers but provides no controlled evidence of sales impact. Use the distinctions to choose where limited business development time goes, and judge progress by useful conversations and follow-through rather than the size of a contact list.

Question for your work: Which relationship deserves attention because there is a relevant problem we can help someone understand?

Application: What useful question or piece of evidence could you bring to your next conversation?

Evidence and limitations

Author reports working with 100 relationship managers. No control group, outcome dataset or measured conversion effect is provided.

Experience-based advice, not causal evidence. The proposed 100-hour campaign and 10,000-hour analogy are not validated thresholds or sales forecasts.

Reading scope and source locations

Passages examined: Complete substantive article, Evaluate your current network through Applying the Four Routes to Growth; excludes navigation and promotional footer.
Where to look: Sections Evaluate your current network; Route one to Route Four; Applying the Four Routes to Growth.

Operations

LBS · Field research · 2008 · LBS03

Expertise utilization in project teams: A status-based account of process and performance

Heidi K. Gardner

University of London: London Business School
2008-05

Open original source Free doctoral dissertation.

B&F reading note

Having knowledgeable people on a project does not ensure their knowledge shapes the result. Gardner’s field research distinguishes recognising expertise from actually using it, including knowledge of the particular client. Under pressure, a team may give greater weight to general professional status. The study offers a useful prompt for reviewing whose evidence influences a decision. It is observational research in large professional firms, not proof that a prescribed team structure will improve every boutique engagement.

Question for your work: Whose knowledge of this client is missing from the decision, even though we know they have it?

Application: Ask the person closest to the client process to challenge the proposed explanation.

Evidence and limitations

Qualitative teams in two firms and a team survey in one Big Four firm. Observational relationships, including client-rated outcomes.

Historical professional-services setting, expertise proxies and intended rather than verified repeat purchases. The dissertation has a sample-count discrepancy; see PDF pp. 133 and 147–151.

Reading scope and source locations

Passages examined: PDF pp. 1–3, 5–6, 72–83, 133 and 147–161; methods and pressure/client-outcomes chapter. Remaining literature, chapter 4 and references not fully read.
Where to look: PDF pp. 74–82 (design/sample), 133 versus 147–151 (sample discrepancy), 152–160 (results/limitations). Printed pp. 64–72, 123, 137–150; tables 5.2–5.5.

LSE · Observational research · 2025 revision · LSE02

Do well managed firms make better forecasts?

Nicholas Bloom, Takafumi Kawakubo, Charlotte Meng, Paul Mizen, Rebecca Riley, Tatsuro Senga, John Van Reenen

Programme on Innovation and Diffusion, London School of Economics and Political Science; ONS/ESCoE survey collaboration
2025-10; POID Working Paper 023, replaces March 2025 version; catalogue retains original 2022-01-05 date

Open original source Free revised working paper.

B&F reading note

Better forecasts are associated with more structured management, not simply with more confident managers. This study links UK management surveys with business data and finds that firms with stronger monitoring, targets and incentives tend to make smaller forecast errors. Its observational design cannot show that installing a dashboard causes better decisions. A useful application is to preserve assumptions, compare expectations with outcomes and investigate errors before expanding a reporting process.

Question for your work: Are our forecasts improving, or are we only becoming more confident?

Application: Write down an expectation before the next data update, then investigate the difference.

Evidence and limitations

UK management surveys linked to business records; forecast errors compared with management practices using regression controls.

Brexit/Covid periods; response and linkage selection; unexplained forecast variation. Table 2's footnote reverses column descriptions; headers and sections 3.2-3.3 establish interpretation.

Reading scope and source locations

Passages examined: Cover and rights page; sections 1-4, focusing on survey design, identification limits and sections 3.2-3.5; Tables 2-4. Table 2 visually checked.
Where to look: Section 2.1, printed pp.5-6 (PDF 6-7); causal caveat footnote 15, printed p.10 (PDF 11); sections 3.2-3.5, printed pp.12-17 (PDF 13-18); Table 2 (PDF 24), Table 4 (PDF 26).

Imperial · Randomized experiment · 2026 · LON02

How forced intervention facilitates AI adoption

Xinyu Cao, Chenshan Hu, Jiankun Sun, Dennis J. Zhang

Imperial College London
2026-02-24 (online); 2026-07-01 (issue)

Open original source Free accepted manuscript, CC BY 4.0.

B&F reading note

A tool can look ineffective when people give it only the cases they already expect to fail. This randomized workplace study examines that selection problem in algorithm adoption. Temporary mandatory use changed later usage, but the observed operating outcomes did not significantly improve. For a pilot, the useful lesson is to compare representative cases, record how cases were selected and measure business results separately from use of the tool.

Question for your work: Are we testing the tool on a fair set of cases?

Application: Are the comparison groups receiving cases of similar difficulty and relevance?

Evidence and limitations

171 sales workers in one online education company; three randomized groups in 2019 and about one month of follow-up. A matching algorithm, not a generative chatbot.

One firm and task, 2019 data, brief follow-up, no significant operational-performance gain. Adoption is not return on investment. The intervention design is not a general recommendation to force unvalidated tools into high-stakes decisions.

Reading scope and source locations

Passages examined: Accepted manuscript PDF pp. 1, 3–8, 12–15 and 18; repository metadata and substantive Imperial commentary. Supplementary proofs and all robustness appendices not independently verified.
Where to look: Section 4.1 pp. 12–13; Figure 1 p. 14; Table 1 p. 15; section 5.3/Table 4 p. 18.

AI and work

LSE · AI field experiment · 2025; updated 2026 · LSE03

The Cybernetic Teammate: a field experiment on generative AI reshaping teamwork and expertise

Fabrizio Dell'Acqua, Charles Ayoubi, Hila Lifshitz, Raffaella Sadun, Ethan Mollick, Lilach Mollick, Yi Han, Jeff Goldman, Hari Nair, Stewart Taub, Karim Lakhani

Programme on Innovation and Diffusion, London School of Economics and Political Science; multi-institution authors with Harvard, ESSEC, Wharton and Procter & Gamble affiliations
2025-06-02; POID Working Paper 118, June 2025 version

Open original source Free 2025 working paper; see the linked 2026 journal version.

B&F reading note

This working-paper version examines AI and collaboration in product development. Experienced professionals worked alone or in pairs, with or without AI, and independent experts assessed their proposals. AI-assisted individuals scored higher than unaided individuals and had similar average scores to unaided pairs; this does not establish equivalence. The one-day workshop at one company supports testing how commercial and technical perspectives can combine, while preserving human review and avoiding claims about replacing expertise or guaranteeing business returns.

Question for your work: When can AI help one person work across commercial and technical perspectives?

Application: Compare manual and AI-assisted work on the same case, including review and correction time.

Evidence and limitations

One-day product-development workshops at P&G: individual versus pair, crossed with AI versus no AI; blinded expert ratings. The 2025 working-paper sample and 2026 participant accounting differ.

One company, ideation task, temporary pairs, trained professionals. Expert scores are not revenue, operational accuracy or long-term expertise. Disclosed company support/author roles.

Reading scope and source locations

Passages examined: Abstract, sections 3-4, 5.1 and opening 5.2, discussion 7, Table 2; table visually checked. Newer journal methods and Table 2 cross-checked online separately.
Where to look: Archived PDF pp.8-14: design and main results; pp.20-23: limitations; p.42, Table 2: quality estimates and observation count.

2026 journal version — same experiment

Imperial · Theoretical model · 2026 · LON01

The need for absorptive capacity alleviates the free-rider problem in knowledge production

Jerker Denrell, Jerry Luukkonen, Nick Chater, Chengwei Liu

Imperial / Warwick
2026-03-13

Open original source Free accepted manuscript, CC BY 4.0; hosted by Warwick.

B&F reading note

Sharing answers does not automatically preserve the ability to evaluate them. This model examines why teams may underinvest in learning when they can simply reuse others’ work, and how prior knowledge changes that incentive. It gives a useful lens for designing an AI-supported research practice: preserve independent thinking and the ability to explain a conclusion. The result is theoretical; it does not establish that adding checks always improves performance.

Question for your work: How do we keep expertise while sharing answers?

Application: Before reading an AI answer, write down your own hypothesis and one thing you do not know.

Evidence and limitations

Game-theoretic model of individual versus social learning. Results depend on opportunity costs, knowledge spillovers and absorptive-capacity assumptions; not an empirical GenAI trial.

No measured boutique productivity, customer demand or tested optimum for an AI review process. Imperial faculty commentary translates the model into AI workflow proposals; this is an application, not a field result of the paper.

Reading scope and source locations

Passages examined: Accepted manuscript: PDF pp. 1–4, 9–10, 19–22, 25–26 and 28. Repository metadata and full substantive Imperial author commentary also read. Mathematical proofs and all appendices not independently verified.
Where to look: Sections 3.1–3.3 model assumptions; section 5 model sensitivity; section 6.1 implications. PDF pp. 9–10, 19–22, 25–26.

About the notes

Download the bibliography (.csv). The full annotations and evidence limits are on this page.

We select sources for the business question they illuminate, their methods and their availability. Older foundational work sits alongside recent research. A faculty essay, a theoretical model and a randomized experiment answer different kinds of questions.

Versions and multiple university affiliations do not make one study into independent evidence. A free manuscript, an interview transcript and an institutional worksheet are different materials; reading one does not mean completing a course or reading a paid book. Reading scope is recorded for every note.

These are B&F annotations and applications. They do not validate demand for a service or promise a client outcome. Universities and authors do not endorse Brief & Field. Full publications remain with their original providers and retain their own terms.

Reuse our original notes with attribution