What each side wants from its expert

One line for each thing the tribunal, counsel, the client and the expert want from expert evidence, and for each thing they do not want, with the source. The survey figures come from CIArb's 2026 Report on party-appointed experts. They record what 419 respondents perceive, not measurements, and the in-house counsel sample was small.

WantsDoes not want

The tribunal

  • Help, not advocacy: the expert assists the tribunal and does not argue the appointing party's case (CIArb Protocol, preamble 4(ii); RHC Appendix D para 4).
  • Reasons it can test, with the method, evidence and information set out (Griffiths v TUI [37], [38]; Nina Kung [20]; IBA Rules art 5.2(e)).
  • Reports it can compare, built on an early meeting of the experts on common data and method (CIArb 2026 para 24.1.1; Chevalier [5]).
  • Reports it can absorb: even several hundred pages may be more than a tribunal can fully take in (CIArb 2026 para 26.1.3).
  • A clear statement of what the expert cannot give an opinion on, and of any opinion that is only provisional (CIArb Protocol art 4.4(h); RHC Appendix D para 10; Ikarian Reefer, duty 5).
  • Questions answered early: 78% of arbitrators want to put questions to the experts after the first reports (CIArb 2026 para 2.1.3).
  • Over-instructed experts: 66% of arbitrators say it happens often or very often (CIArb 2026 para 5.1.3).
  • Opinions held back and sprung in a reply report (IBA Rules art 5.3; CIArb 2026 para 23.1.2).
  • An expert who chooses between the parties' versions of the facts, or decides how the contract should be read (ICI v MMT [184], [237]).
  • To be asked to defer: the tribunal cannot hand its decision to the expert, and should give reasons for preferring one (Kennedy v Cordia [49]; ICC Commission 2021; CIArb Guideline 2015).
  • Private meetings with the experts after the hearing: only 16% of arbitrators think them more helpful than harmful (CIArb 2026 para 39.1.1).

Counsel and solicitors

  • The expert appointed early enough to sense-check the claim before it is filed (CIArb 2026 para 13.1.5).
  • Instructions written for the tribunal to read: the report describes them, and under the CIArb Protocol they are not privileged (IBA Rules art 5.2(b); CIArb Protocol art 5.1).
  • All the material, good and bad, to both sides' experts alike (Kennedy v Cordia [57]; ICI v MMT [237]).
  • Assumptions made visible, with the expert free to decline or reformulate an instruction (CIArb 2026 para 5.1.11; Presbar v GW Atkins [68]).
  • A joint statement the experts own, reached without counsel (TCC Guide 2026 para 14.6.3; Chinachem [490]).
  • A position that is defensible from the start, not negotiated towards the centre at the hearing (Cabo v MGA [49]).
  • The other side's expert left unchallenged on a point it needs the tribunal to reject (Griffiths v TUI [70]).
  • A change of view saved up for the hearing: the sooner the expert says so, the better (Guntrip v Cheney Coaches [21]; PD 35 para 2.5).
  • Its own fingerprints on the joint statement: permission to rely on the expert was revoked in Andrews v Kronospan (2022) and Glover v Fluid (2024).
  • An expert who strays outside the questions asked or outside his own field (Vainker v Marbank [58]; Beattie Passive v Canham [79]).
  • Experts approaching the tribunal directly, even with counsel copied: 55% of external counsel oppose it (CIArb 2026 para 10.1.3).
  • A tribunal-appointed expert who becomes a fourth arbitrator in all but name (QMUL 2021 survey).

The client

  • The expert appointed once the issues are known, not at the last minute: 80% of in-house counsel favour this (CIArb 2026 para 12.1.2).
  • Cost under control: expert evidence is a large share of the total, and expert fees push up counsel's fees and hearing time too (CIArb 2026 para 12.1.6; ICC Commission 2021 s II(D)(iv)).
  • A real expert, not a fact witness doubling as one: in-house counsel were unanimous (CIArb 2026 para 33.1.1).
  • Evidence that survives the hearing: a biased expert who sticks to the party line harms the client's own case (Secretariat v A Company [110]).
  • Conflicts checked at appointment and disclosed; the tribunal, not the parties, decides whether one matters (Toth v Jarman [102], [112]).
  • Short summaries and expert presentations at the hearing: 83% of in-house counsel want each (CIArb 2026 paras 26.1.1, 38.1.1).
  • Its own lawyers steering the expert: in-house counsel reported the highest rate of over-instruction, on a small sample (CIArb 2026 para 5.1.4).
  • Late instruction: it compresses the expert's time and raises the cost, a false economy (CIArb 2026 para 14.1.5).
  • To give up its own expert to save money: only 13% of users would (QMUL 2021 survey).
  • A tribunal-appointed expert sold as the cheaper route: it is not necessarily so (Bell, GAR Guide to Construction Arbitration, 2025).
  • To lose its expert for a breach of the rules, as in Dana v Freudenberg (on day 7 of the trial) and Marples (the client had rewritten the joint statement).

The expert

  • Enough time: only 44% of experts say they get it, against 77% of arbitrators (CIArb 2026 para 16.1.1).
  • Clear instructions and a list of the specific questions to answer (IBA Rules art 5.2(b); PD 6.1 para 15(4)).
  • All the documents, including the ones that hurt (Kennedy v Cordia [57]; ICI v MMT [237]).
  • Room to agree with the other expert without counsel involved in the joint report, and no instruction to withhold agreement (CIArb 2026 para 6.1.1; RHC Appendix D para 13; SCL(S) Protocol 2026).
  • To be brought in early and treated as more than a report writer (CIArb 2026 para 14.1.3).
  • Clear rules on AI: 89% of experts in the 2025 Bond Solon survey want guidance, and RICS members must identify AI-generated work from 30 November 2026 (RICS 5th edn s 4.1.1(b)).
  • Opinions withheld by the other expert and sprung in reply: 85% of experts have met it (CIArb 2026 para 23.1.2; IBA Rules art 5.3).
  • To be over-instructed, though 73% of experts say it happens never or not often (CIArb 2026 para 5.1.5).
  • Pressure to change an opinion: 23% of UK experts reported it from instructing solicitors in 2024 (Bond Solon 2024); edits that distort the opinion are to be disregarded (RICS 5th edn).
  • An AI-drafted report handed over to sign: 14% of experts would accept one (Bond Solon 2025).
  • An instruction not to agree a point, or not to sign the joint statement (RHC Appendix D para 13; Chinachem [490]; Vainker v Marbank [58]).
  • A hired gun on the other side: more than a third of UK experts met one in the past year (Bond Solon 2024).