The LinkedIn Recommendations Run, From Role History to Published Blurbs

This guide is for job seekers who want a set of specific, role-covering LinkedIn recommendations written, approved, and published before they start applying. It walks the full job in order: from a coverage map of your last several roles, through asks scripted to pre-empt boilerplate, to the approval, display, and curation steps that the quick-answer posts skip. Follow it start to finish and you end with a curated set where each visible blurb covers a different role you are selling.

Most profiles end up with two overlapping, generic blurbs because every top result is the same five-click path plus a fill-in-the-blank template. That produces volume without coverage. This run inverts the order: decide what each blurb must prove before you ask anyone, then sequence the asks so no two recommenders repeat each other.

Why this run starts from coverage, not a template

The real constraint on your profile is not how many recommendations you can collect. It is that your public profile displays only a maximum of two received recommendations, and those appear in chronological order and currently cannot be reordered. So the last two you accept decide your first impression. Volume does not fix a weak first impression; sequencing and curation do.

That changes the whole job. If only two show by default, and you cannot force your best one to the top, then the recommendations you let through, and in what order you accept them, is the decision that matters. A fill-in-the-blank template cannot make that decision for you. A coverage map can.

Coverage beats accumulation. Published guidance converges on three to five recommendations that each cover a different role, with at least one per job you held in the last five years. A fourth blurb praising a role you have already covered adds nothing that a first blurb on an uncovered role would add. Recruiters skim; a second testimonial on the same project is wasted surface.

How many recommendations, and covering what

Aim for three to five strong recommendations, each mapped to a distinct role, with at least one for every job you have listed in the last five years. Three is the floor at which LinkedIn considers the section complete; beyond five you are usually adding duplicates, not coverage.

The published sources disagree on the exact number but agree on the shape. Below is what the named guidance actually says, so you can pick the count that fits how many roles you are selling.

Read these as a range, not a contradiction. If you are selling two recent roles, three strong blurbs that cover both, from a manager and a peer or client, beats ten that cluster on one job. If you are selling four roles, you need at least four recommenders, one per role, before you think about a second blurb anywhere.

The coverage rule is the one to hold firm on. A recruiter looking at a four-role history and seeing three blurbs all on the oldest role will read the silence on your recent work as a gap. Pointing every role at a named recommender closes that gap before it opens.

Who to ask, and who will actually read it

Ask the person who directly witnessed the work for each role, ideally the direct manager, with a close peer or client as the backup. Only first-degree connections can recommend you, so your pool is whoever you are already connected to who saw the work firsthand.

Diversity of relationship matters as much as seniority. A set of three from three managers reads as narrow; a manager, a peer, and a client reads as a rounded picture. Match the recommender to the angle: a manager speaks to judgment and results, a peer to collaboration, a client to reliability and outcomes, a direct report to how you lead.

It helps to know who sits on the other side of the profile. The people most likely to read these blurbs are recruiters and talent-acquisition staff, and that audience is lopsided by geography. In Refolk's index of recruiter and talent-acquisition profiles, the split is stark.

That is roughly a 15x gap. The practical read: a US candidate competes for far more recruiter attention, which raises the premium on blurbs that survive a fast skim. The more crowded the field of profiles a recruiter scans, the more a specific, role-covering recommendation earns its place against a generic one.

The mix within that audience also shifts by market, which is worth knowing if you are targeting a specific country. This is a small directional sample, not a population share, but it hints at who you are writing for.

Treat this as a nudge, not a rule: the sample is 25 profiles, so read it as directional. If you want to see the actual recommenders and readers in your target market before you draft, a search over profiles is faster than guessing.

What makes a blurb specific instead of generic

A specific recommendation names one project and one number; a generic one offers fluent praise with neither. The difference is the whole game, because vague praise reads as filler and a recruiter discounts it on sight.

The dominant published structure is five parts, and it works for every relationship type: context, one specific project, measurable impact, a standout trait, and a confident close. Specificity is the mechanism, not the decoration. "Increased trial signups 22% in one quarter" beats "hardworking and dedicated" because the first is a claim a recruiter can picture and the second is a phrase that fits anyone.

Length is a quality signal too. The professional standard is 100 to 200 words: long enough to carry credible context and one achievement story, short enough to read in under 60 seconds. Shorter reads as an afterthought; longer stops getting read.

A blurb with no project and no number is not a recommendation, it is a greeting with your name in it.

You cannot write the blurb for your recommender, but you can hand them the raw material that makes a specific blurb the path of least resistance. That is what the brief in the next section does. Leave the writing to them; it is usually best to let the other person write it their way unless they ask for help, because a draft in your voice reads as ghostwritten.

The run, step by step

This is the full procedure, in order, with how long each stage takes and what done looks like. The two preparation steps are the ones the quick-answer posts skip, and they are what separate a covered profile from two overlapping blurbs.

Where the request mechanics bite

Two routes exist and LinkedIn documents the same fields for both. From the other person's profile, open their profile, click More, select Request a recommendation, set the Relationship and Position at the time fields, click Next, edit the message, and Send. From your own profile, scroll to Recommendations, click the plus icon, search the name, set the two fields, and send. Note that the Recommendations section only appears on your profile once you have given or received one that is not hidden.

You can only request one recommendation from a connection at a time, though there is no limit on the total you can request or receive. Set the two fields carefully: the relationship cannot be changed once sent, so a wrong tag means re-requesting.

The warm-up order is a genuine disagreement in the sources, and it is worth stating. Several guides put the system request first. Practitioners at IvyExec and The Muse insist on an email or direct-message warm-up before the LinkedIn request, on the logic that reaching out through LinkedIn cold gets a worse response. I side with the warm-up: a one-line "would you be open to this?" costs you five minutes and turns the formal request into a confirmation rather than an ambush.

The ask message itself

The most effective ask combines three things: personal context about why you are asking this person specifically, a pointer to what you would like highlighted, and a draft they can edit freely. Here is a skeleton that carries all three without reading as a form letter.

Hi [name],

I'm lining up my profile before a job search and I'd value a recommendation from you specifically, because you saw [the project] up close.

If you're open to it, the thing I'd most want captured is [the one project] and the result we got: [one number, e.g. cut onboarding time 30%]. It would speak to my [target skill] better than anything else I could point to.

I know the blank page is the annoying part, so I've written a rough draft below - please edit it however reads true to you, or ignore it entirely:

"[2 to 3 sentence draft: context, the project, the measurable result, one standout trait, a confident close]"

No rush, and happy to return the favor whenever useful.

Thanks,
[you]

The draft is the conversion lever. The blank page, not unwillingness, is what kills asks, and handing over a draft removes the most common reason good intentions never become posted recommendations. One vendor claims a draft plus named qualities triples response rates; treat that as a claim, not evidence, because it has no published methodology behind it.

Following up without pressuring

Follow up exactly once, roughly seven to fourteen days after the first ask, and make the second message even easier than the first by attaching the ready-to-paste draft again. Then stop. A second reminder past that starts to feel like pressure, which is the fastest way to turn a soft yes into an awkward no. Many silent non-responders are soft yeses; the single nudge recovers a surprising number of them.

How this goes wrong

The failure modes below are where a run that looked finished at "sent" falls apart at "published." Each has a false positive that reads fine until you check it against your coverage map.

• Generic blurb. It reads fluent but names no project and no number, so a recruiter discounts it. Check: does it contain one metric and one named project? If not, use Ask for a revision rather than accepting it.
• Overlap. Two recommenders praise the same trait on the same role, wasting a slot. Check: each blurb maps to a distinct role or skill in your map; hide the weaker duplicate.
• Wrong role tag. The relationship or position was set incorrectly, and the relationship cannot be changed once sent. Check the two fields before Send; if wrong, re-request with the correct tag.
• Display-order surprise. You assume your best blurb shows first, but order is chronological and the public view caps at two. Check which two surface publicly and hide others to curate, since you cannot reorder.
• Silent non-responder treated as a no. Many are soft yeses who hit the blank page. Check: send the one day-seven nudge with a draft before writing anyone off.
• Draft that sounds like you, not them. A pasted draft in your voice reads as ghostwritten. Check: tell them to edit freely, and let the blurb land in their words.
• Endorsement and recommendation confusion. Treating a pile of skill endorsements as reference-weight. Check: endorsements are one-click skill validations for search visibility, not testimonials.
• Vendor-stat contamination. Leaning on figures like "14x more profile views" or "87% of recruiters use LinkedIn" as proof. Both are unverified secondary claims; treat them as claims, not evidence, and do not build a decision on them.

Refolk tailors a resume to each posting and scores how well you actually fit, and the same discipline applies here: a recommendation is only working for you if it maps to a role you are selling and names evidence a recruiter can picture. If it does neither, revise it or hide it.

Endorsements, briefly, and why they are not a substitute

Seed skill endorsements separately from recommendations, because they do a different job. A recommendation is a written testimonial that someone must take time to compose; an endorsement is a one-click validation of a skill. Endorsements drive search visibility and candidate-list ranking, recommendations drive credibility with hiring managers who read them.

You need both, seeded separately, not one standing in for the other. You can list up to 50 skills for endorsement, so pick your priority skills, list them, and endorse peers to prompt reciprocity. Do not let a wall of endorsements lull you into thinking the recommendations job is done; a recruiter reads them as different signals.

The pre-publish checklist and keeping it current

Before you call the run finished, verify each item below against your coverage map. The point is not that blurbs exist, but that the visible ones cover the roles you are selling.

Keeping this current is a light touch, not a rebuild. Each time you add a role to your profile or shift which roles you are selling, revisit the coverage map and check whether the new role has a recommendation. When you change jobs, ask the manager you are leaving before the memory of the work cools; a recommendation is easiest to get while the project is fresh and the relationship is warm.

Because the public view caps at two and orders them chronologically, every new recommendation you accept pushes the display. If a fresh blurb covers a role you already had covered, accept it only if it is stronger than what is showing, or hide the weaker one so your priority roles stay on top. The curation step is never truly done; it is the part of the run you revisit each time the set changes.

One last discipline carried from the index: a combined Director-seniority plus recruiter-title query returned zero matches, a reminder that over-filtering hides real people. The same caution applies to your recommender hunt. If a tight filter returns no one, loosen it rather than concluding the right recommender does not exist. The person who witnessed your best work is usually in your connections already; the coverage map just makes you go find them.