THE DIRECT ANSWER

To write a LinkedIn recommendation, begin with how you know the person and the role or period you shared. Name one specific project, decision or recurring responsibility you directly observed. Describe what the person did, how they approached the work and the value or consequence you can verify. Close with the professional situations in which that capability is relevant. Use your own words, separate their contribution from team results, exclude confidential details and send only claims you are prepared to stand behind publicly.

PBL DECISION FRAMEWORK

The PBL Witness-to-Value Arc

Turn first-hand observation into useful professional proof through five bounded decisions.

01RelationshipState how you worked together and which position or period the recommendation covers.
02SituationName the project, decision or recurring responsibility you directly observed.
03ContributionDescribe what the person actually did—not only the trait you admired.
04EvidenceShow an observable behaviour, judgement or result you can verify and attribute fairly.
05RelevanceExplain where that contribution is useful without making a universal promise.

Write as a witness, not as a promoter

A LinkedIn recommendation is public professional testimony attached to your identity. Its credibility comes from a bounded relationship to the work: what you observed, in which context and why your perspective is relevant. Generic praise can feel supportive, but it gives a reader little basis for judging the person's contribution.

Approach the task as a witness. You do not need to summarise an entire career or predict future performance. Choose one useful truth you can support. The recommendation can be warm and generous while remaining precise about your vantage point, the shared work and the limit of what you know.

Choose one evidence job before drafting

Decide what the recommendation should help a relevant reader understand: how the person makes difficult decisions, leads a cross-functional process, translates expertise, builds trust, improves execution or contributes under pressure. One recommendation rarely needs to prove every capability.

Select the evidence job from your own observation, not from the recipient's desired brand language. If they want to be known for commercial strategy but you only observed their project management, write about the project management you know. A narrower account is more useful than an impressive claim borrowed from someone else's profile.

Open with the relationship and shared context

The first sentence should answer why your account is informed. Name the working relationship, the person's position at the time and a useful time or project boundary. LinkedIn's current recommendation flow asks the writer to select Relationship and Position at the time before writing the message; make the prose consistent with those fields.

Avoid turning the opening into a status display. ‘I managed Priya during the regional service redesign’ is stronger evidence than a long introduction about your own seniority. If the relationship changed over time, choose the context most relevant to the observation and do not imply authority you did not hold.

Move from adjective to observable contribution

Adjectives such as strategic, collaborative and exceptional become meaningful only when the reader can see the behaviour behind them. Replace ‘Kai is a strategic leader’ with the decision you saw Kai frame, the competing constraints considered and the action that followed.

Use a compact pattern: situation, contribution, evidence. Explain the condition the team faced, what the person specifically did and what changed or became clearer. If the work was collaborative, keep shared credit visible. Do not convert an organisation or team outcome into a personal result merely to make the recommendation sound stronger.

Use results only when you can verify and attribute them

A numerical outcome can add precision, but only when you know it is accurate, public or authorised and fairly attributable. You may have observed someone's decision quality without having access to final commercial results. In that case, write about the process and behaviour you witnessed rather than inventing a stronger ending.

Distinguish direct observation from interpretation. ‘I watched her redesign the review so country risks surfaced before approval’ is an observation. ‘She guarantees successful transformations’ is a prediction. Recommendations should not create confidential disclosure, regulated endorsement or an implication that an employer, client or board formally supports the person when you are speaking only for yourself.

Close with bounded relevance, not a universal claim

The final sentence can help a reader understand where the observed capability matters. Connect the contribution to a type of problem, team or working environment: ‘I would value that judgement on programmes where regional constraints need to shape a global plan.’ This is more informative than ‘I recommend her for anything.’

If you genuinely want to recommend the person for a defined kind of work, say so. Avoid unsupported superlatives, guaranteed outcomes and sweeping statements about contexts you have never seen. A credible boundary does not weaken the recommendation; it tells the reader how to use it.

Worked example: rebuild vague praise from real observation

Fictional teaching example — Mei, a product lead, wants to recommend Daniel, a former cross-functional partner. Her first draft says: ‘Daniel is an amazing communicator and one of the best leaders I have worked with. He always delivers incredible results.’ The support is sincere, but a reader cannot tell what Mei observed, what Daniel contributed or which result belongs to him.

Using the PBL Witness-to-Value Arc, Mei writes: ‘I worked with Daniel for nine months while our product and operations teams redesigned the customer-onboarding handover. When country requirements began to conflict, he turned the discussion into a decision log that separated fixed constraints from local choices. I saw him invite dissent before each review, clarify owners and leave teams with decisions they could execute. I would value that facilitation on any cross-functional programme where progress depends on making trade-offs explicit.’ The names and situation are hypothetical; this is not a client story or performance claim.

Respond to a request without surrendering your judgement

A recipient may send useful memory prompts about a project or capability. Treat them as context to verify, not instructions you must repeat. You can decline the request, write a narrower recommendation or ask a factual question before drafting. LinkedIn says a recommendation request can be ignored, and the recipient decides whether to display what you send.

Do not negotiate reciprocal praise. If you separately recommend each other's work, each account should remain independent and defensible. Avoid copying a draft the recipient supplied, coordinating adjectives or implying that one recommendation is payment for the other. If your relationship creates pressure—such as a current reporting, purchasing or supplier dependency—consider whether a public recommendation is appropriate at all.

Review the recommendation as a durable public claim

Before sending, check names, relationship, position, dates, attribution, confidentiality and every factual claim. Read for what an unfamiliar person might infer, not only what you intended. Remove information you lack permission to disclose and qualify conclusions that travel beyond your observation.

LinkedIn currently lets a writer revise or delete a recommendation later. A revised recommendation must be reviewed by the recipient before it appears again. That control is useful when facts, permissions or context change, but it does not replace a careful first review. Write something you can still explain after the project and the relationship have moved on.

Keep the writing recognisably yours

A recommendation should sound like a real professional account, not a template assembled from prestige words. Use the vocabulary and sentence length natural to your relationship. Specific nouns and verbs usually create more credibility than extra adjectives.

AI can help organise notes or test clarity, but it does not know what you witnessed. Supply the real relationship, situation, action, evidence, ownership and boundaries. Delete any invented result, certainty or emotional detail. The final judgement and every public claim remain yours.

DECISION TABLE

Compare the choices clearly

SituationStrong writing choiceTrade-off or boundary
You managed the person directlyDescribe a recurring behaviour and one representative momentDo not imply knowledge of work outside your team
You were cross-functional peersShow the decision or handoff you both experiencedCredit the wider team and avoid managerial language
You were a client or supplierExplain the authorised scope and observable service contributionCheck policy and avoid implying an organisation-wide endorsement
You know the outcome but not the processState the limited outcome perspective—or declinePrestigious distance is weaker than first-hand evidence
The request includes a finished draftUse it only as a factual prompt and write independentlyDo not sign language or results you cannot verify
You recommended each otherKeep the two decisions, examples and wording independentReciprocity must not become an obligation

ACTION CHECKLIST

Use this before making the next move

  • Confirm that you directly observed the work you plan to describe.
  • Choose one capability or contribution for the recommendation to evidence.
  • Select the correct relationship and position at the time.
  • State the project, period or recurring responsibility clearly.
  • Describe what the person did using specific nouns and verbs.
  • Include an observable behaviour, judgement or authorised result.
  • Separate the person's contribution from team or organisation outcomes.
  • Verify every name, role, date, figure and causal implication.
  • Remove confidential, personal or restricted information.
  • Close with a bounded professional context where the capability is relevant.
  • Make the wording your own and reject invented AI-generated details.
  • Read the final text as a durable public claim before sending.

COMMON MISTAKES

What weakens the result

  • Opening with a list of superlatives instead of the relationship.
  • Trying to summarise the person's whole career in one recommendation.
  • Claiming expertise or results you did not directly observe.
  • Copying a finished draft supplied by the recipient.
  • Using a senior title as a substitute for specific evidence.
  • Attributing a team or organisation outcome entirely to one person.
  • Including confidential client names, figures or internal decisions.
  • Writing for a relationship or position different from the selected fields.
  • Making reciprocal recommendations feel conditional or coordinated.
  • Ending with a universal promise that exceeds the evidence.

REUSABLE CHATGPT PROMPT

Apply the framework to your situation

Use the Personal Brand Lab The PBL Witness-to-Value Arc to assess my situation. Ask me one question for each stage before giving advice: Relationship, Situation, Contribution, Evidence, Relevance.

For each recommendation, show the evidence from my answer, the trade-off, the next action and what I should measure. Do not invent facts, proof or results. Flag missing information.

Framework source: https://resources.personalbrandlab.co/resource/blog/how-to-write-linkedin-recommendation
PERSONAL BRAND LAB

Check whether the profile makes the proof useful

Use the LinkedIn Profile Audit Checklist to review whether positioning, Experience, recommendations, Featured evidence and next steps reinforce one credible profile story.

Open the profile audit checklist →

COMMON QUESTIONS

Frequently asked questions

How do I write a good LinkedIn recommendation?

State how you know the person, identify one shared context, describe a contribution you directly observed, add verifiable evidence and close with the professional situations where that contribution is relevant.

How long should a LinkedIn recommendation be?

Use the shortest length that establishes relationship, context, contribution and evidence. A precise paragraph is usually more useful than several paragraphs of repeated praise; LinkedIn does not prescribe an ideal length in the current help guidance reviewed for this article.

What should I say in the first sentence?

Explain your relationship to the person and the work: the role, project or period that gives you a credible vantage point. Keep the opening about the evidence, not your own status.

Can I use a LinkedIn recommendation template?

Use a framework to organise your observation, but replace every detail with facts you personally know. Do not copy claims, adjectives, results or relationships from a template.

Should I mention results?

Mention results only when they are accurate, authorised and fairly attributable. When you cannot verify a result, describe the person's observable judgement, behaviour or contribution instead.

Can I decline a recommendation request?

Yes. LinkedIn says you can ignore a request. Decline or write a narrower account when you lack first-hand knowledge, permission or confidence in the requested claim.

Can I edit a recommendation after sending it?

Yes. LinkedIn lets the giver revise or delete a recommendation. If revised, the recipient must review it before it is displayed again.

AUTHOR & METHOD

Resources by Personal Brand Lab

This article was developed from the PBL Resources methodology for turning personal-brand questions into practical decisions. It distinguishes operating recommendations from claims about guaranteed performance; no client result or market statistic has been invented.

Sources checked

  • LinkedIn Help — Recommend Someone on LinkedIn — Checked on 12 September 2026 to verify first-degree connection eligibility, current desktop and mobile paths, Relationship and Position at the time fields, and the recipient's choice to display the recommendation.
  • LinkedIn Help — Recommendations on LinkedIn — Checked on 12 September 2026 to verify giving, declining, hiding, revising and deleting controls; the one-recommendation-per-position-per-member limit; and withdrawal after a connection is removed.
  • LinkedIn Help — Revise a Recommendation You've Given — Checked on 12 September 2026 to verify the writer's revision workflow and that the recipient must review a revised recommendation before it is displayed again.

Published and reviewed 12 Sep 2026.