The Person-Level Buying Signal Reference: Proof, Mimic, Shelf Life

This reference is for founders, account executives, SDR leads, and partnerships teams. Use it when you've just watched a named person at a target account do something in public and you need to decide what it means. Each row says what the act proves about that person, how it fakes intent, how many days it stays live, and which second signal it needs before the person goes into a sequence. Other references in the library cover company trigger events and repository signals. This one covers one person and one act.

What is a person-level buying signal, and what does each row record?

A person-level buying signal is a public act by one named person at a target account that makes it more likely that they, specifically, are weighing a purchase now. It's weaker evidence than most teams assume. Gartner puts a complex B2B purchase at 6 to 10 decision-makers, so one person's act speaks for one seat on the buying committee at most.

Every row in this reference records the same four fields:

• Proves. The narrowest claim the act supports about the person's intent. Usually smaller than it looks.
• Lies when. The mimic: a reason unrelated to buying that produces the same visible act. The classic mimic is an engagement pod, a group of accounts that agree to like and comment on each other's posts to boost reach.
• Shelf life. How many days the act stays usable as a reason to reach out. Where a published figure exists, it's a vendor claim, and I label it that way.
• Corroborate with. The second, independent signal you need before enrolling the person. Independent means a company-level event at the account, or a different kind of act by the same person. A second like doesn't count.

One limit up front: nobody has published a dated decay curve for person-level signals along with the method behind it. The shelf lives here are the best public numbers available, and they disagree with each other. Use them as starting cutoffs and tune them against your own reply data. They aren't measurements.

How long does a buying signal stay fresh?

Vendors' shelf lives for the same signal differ by as much as 14x. The only signal where they roughly agree is a substantive comment, at about 7 days. Cheap acts like a like or a profile view have no agreed shelf life at all.

Four sales-tool vendors publish shelf lives for person-level signals, and none of them publishes a method. The only vendor that mentions a sample at all cites outbound testing across 6,000+ accounts, and even that vendor doesn't give a separate method for each half-life.

The disagreement itself is informative. On a like, the vendors are 10x apart: 3 days against 30. On a profile view they're 14x apart: 24 hours against 14 days. On a comment, they're within roughly 1x to 2x of each other. My read is that a comment takes effort and leaves a public record, so every vendor is looking at similar behaviour. A like costs nothing, so each vendor's number mostly reflects its own pairing rules, meaning which other signals it requires alongside the like. It says little about the buyer.

How to use the table:

• Shortest figure in the row = your enrollment deadline. If the act passes every check, act before that figure runs out. One signal vendor uses a 48-hour time-to-enroll SLA, which is a sensible operating target.
• Longest figure in the row = your nurture horizon. After that point, park the contact with a re-check date. Don't enroll them.
• Job changes: pay more attention to the 30-day reply drop-off one outreach-tool vendor reports than to the 60-day figure.

Which signals lift reply rates over a cold baseline?

Job-change plays have the largest documented lift: 4.2x to 5.3x over the same sender's cold baseline, against 2.3x for engagement on a founder post. None of these cases publishes a sample size, so they show which direction the effect runs, not how big it is.

What the table supports:

• A job change outperforms engagement. A new role comes with a new budget and a reason to review tools. Engagement only shows interest. The job-changer case also reports meetings: 1.89% for recent changers against 0.2% for people 7+ months into their role.
• The vendor claim on former champions isn't proven. A job-change data vendor says previous champions convert 114% more often than cold prospects. It's the vendor's own figure with no published method, so log it and don't plan around it.
• The only large-sample row is about recruiting, not buying. It covers candidates replying about jobs. Recently updated profiles replied 23% more often in the same study. It's the best evidence that a change at the person level moves replies, but it doesn't measure buyers.
• Peer-community tool requests have no published lift. I haven't found one anywhere. Measure it yourself before you staff a play around it.

For a sense of normal rates, a LinkedIn outreach benchmark across 15M+ contacts puts the overall reply rate at 7.2%. A 5% team baseline is ordinary, not unusually weak.

The signal reference: 25 rows by family

Each row stands on its own. Where I mark a shelf life "working default", no source covers that signal, so I've borrowed the nearest published window and said which one. Swap in your own number once you have reply data.

Engagement acts

1. Like or reaction

• Proves: they saw the post and didn't scroll past. Says nothing about timing or budget.
• Lies when: they were scrolling passively, a pod is returning the favour, or a colleague is supporting a coworker's post.
• Shelf life: 3 to 30 days, depending on the vendor.
• Corroborate with: never enroll on this alone. It needs a different act by the same person or a company-level event.

2. Like on a high-relevance post (pricing, comparison, your product)

• Proves: attention to a topic close to a purchase.
• Lies when: the same mimics as row 1 apply. The topic is specific, but the cost of the act is still zero.
• Shelf life: 3 days, from the one vendor that lists it separately.
• Corroborate with: a comment, reshare, or profile view from the same person.

3. Company page follow

• Proves: low-cost interest in your company.
• Lies when: the follower is a competitor's employee, an investor, a candidate, or a partner.
• Shelf life: no published figure. Working default: treat it like a single like.
• Corroborate with: employer and function checks first, then any second act.

4. Single profile view

• Proves: they looked you up, often as a response to your own earlier touch.
• Lies when: a recruiter is sourcing, a peer seller is researching, or they're returning a view.
• Shelf life: 24 hours to "days".
• Corroborate with: a second view within the week, or any public act.

5. Repeated profile views (2+ in a week)

• Proves: sustained attention to you or your company.
• Lies when: the viewer is a competitor or a candidate.
• Shelf life: 14 days, from the one vendor that defines it.
• Corroborate with: a company-level event at the account.

6. Generic comment ("Thanks for sharing", "Great post")

• Proves: nothing.
• Lies when: it is the mimic. Pods and auto-scripts produce exactly this.
• Shelf life: none. Discard it.
• Corroborate with: not applicable.

7. Substantive comment on category content

• Proves: effort and a public opinion on a topic you serve.
• Lies when: a pod, an auto-script, or an AI-written comment produced it (see the failure modes section).
• Shelf life: 7 days. Three of the four vendors agree within a week.
• Corroborate with: a different act by the same person, or a company-level event.

8. Comment on your founder's post

• Proves: interest in your company's point of view. This is the play behind the 11.6% case above.
• Lies when: the commenter is a peer, recruiter, investor, or pod member.
• Shelf life: 7 days.
• Corroborate with: a confirmed function on the buying committee, plus a company-level event.

9. Reshare without commentary

• Proves: mild endorsement of the post.
• Lies when: they're reposting habitually or pushing employer-brand content.
• Shelf life: 14 days, from the one vendor that lists content shares.
• Corroborate with: any act that has their own words in it.

10. Reshare with commentary

• Proves: they've attached their name to a view, and their words tell you the angle.
• Lies when: they're building a personal brand, not looking for a vendor.
• Shelf life: 5 to 14 days.
• Corroborate with: a company-level event or a second act.

11. Engagement with a competitor's content

• Proves: interest in the category, and possibly an active evaluation.
• Lies when: the person is the competitor's employee, partner, or investor.
• Shelf life: working default of 14 to 21 days, borrowed from Vendor D's window for engagement with category content.
• Corroborate with: an employer check, plus a second act or account event.

12. Sustained engagement across several weeks

• Proves: an ongoing interest in the topic, not a one-off click.
• Lies when: the same names keep showing up across posts, which is a sign of a pod.
• Shelf life: 14 to 21 days from the most recent act.
• Corroborate with: a company-level event.

Profile changes

13. Job change into a target account

• Proves: a new budget and a reason to review tools. This is the strongest documented lift in this reference.
• Lies when: a tracker surfaces an old move late.
• Shelf life: replies reportedly fall off after 30 days; one vendor allows 60.
• Corroborate with: the start date on the profile.

14. Former champion at a new company

• Proves: they know your product.
• Lies when: they weren't actually a user, or the move is old.
• Shelf life: same as row 13.
• Corroborate with: the start date, plus evidence they used the product at the previous employer.

15. Promotion or internal title change

• Proves: possibly more scope.
• Lies when: the promotion only renamed the title.
• Shelf life: same as row 13 if the scope really changed.
• Corroborate with: a new reporting line or a newly visible team.

16. Senior title at a company under 50 people

• Proves: very little about budget.
• Lies when: the "VP" has no team (see the titles section below).
• Shelf life: whatever the underlying act's row says.
• Corroborate with: headcount, and evidence that someone reports to them.

17. Recently updated profile or headline

• Proves: activity, and in recruiting data, a higher chance they'll reply.
• Lies when: the platform generated the update automatically, or the person is job-hunting.
• Shelf life: no published figure for buyers.
• Corroborate with: an act that is specific to buying.

18. Between roles

• Proves: they aren't a buyer right now. The 74% lift above is a recruiting result.
• Lies when: a buying list treats it as intent.
• Shelf life: not applicable. Track it until they start somewhere new.
• Corroborate with: their next employer, then row 13.

Asking in public

19. Tool request in a peer community (Reddit or a private group)

• Proves: an active evaluation. This is the most specific intent in this reference.
• Lies when: the account belongs to a consultant, a student, or a vendor planting the question. Identity is also the hardest thing to confirm here.
• Shelf life: no published figure. Treat it as live while the thread is active.
• Corroborate with: the person's identity and employer, which you must confirm before anything else.

20. LinkedIn post asking for recommendations

• Proves: an evaluation under the person's real name.
• Lies when: the post is fishing for engagement.
• Shelf life: working default of 7 days, borrowed from the comment row.
• Corroborate with: their function on the buying committee, plus an account event.

21. Post about switching or leaving a tool

• Proves: dissatisfaction, or a switch that's already finished.
• Lies when: the decision has already been made.
• Shelf life: no published figure. If the switch is done, you've missed the window.
• Corroborate with: a company-level signal that the decision is still open.

22. Written software review

• Proves: they use or used the tool and have an opinion about it.
• Lies when: the review is about a previous employer's stack.
• Shelf life: no published figure.
• Corroborate with: the current employer, plus a second act.

Visibility acts

23. Conference talk or panel

• Proves: they own the topic and have taken a public stance on it.
• Lies when: the talk is employer marketing or personal brand-building.
• Shelf life: no published figure. A talk gives you a durable topic hook but no timing.
• Corroborate with: a company-level event or a recent act, and take your timing from that.

24. Podcast guest appearance

• Proves: the same as row 23, plus a long record of how they talk about the problem.
• Lies when: the same as row 23.
• Shelf life: no published figure.
• Corroborate with: the same as row 23.

25. Their own long-form post on a problem you solve

• Proves: they own the problem in public.
• Lies when: the post is ghostwritten or AI-written thought leadership.
• Shelf life: working default of 14 to 21 days, from the window for category content.
• Corroborate with: a company-level event.

Rows 23 to 25 are the hard case. They're effortful and public, but they point at a topic, not a purchase. The useful move is to find people who own a topic at accounts large enough for their titles to mean something, and then look for the timing signal separately.

Does the title carry budget? Company size and country

A senior title is good evidence of budget authority only when the company has more than about 50 people and you read the title by its own country's conventions. Below 50, inflated titles are common. In the UK, the top sales leader is far more often a Director than a VP.

Index Ventures estimated that 30% of executives hired in a startup's "first 10" stage have inflated titles, with less than six years of prior experience, and that this lasted until 50 headcount. 1752.vc's check is the practical one: does anyone report to this person? Nobody has published data linking titles to budget authority by headcount band, so this is a check you run, not a lookup.

Country matters as much as size. Here are counts of title matches in Refolk's index:

The keyword counts are also noisy. In a UK Director sample of 25 from Refolk's index, only 13 headlines literally said "Sales Director". The rest included CEOs, a Managing Director, and a Non Executive Director. A title-keyword match is where you start reading, not a verdict. This is the friction that asking in plain English removes. In Refolk I ask for "sales leaders at UK fintechs" instead of filtering on "VP", and Directors stay in the results instead of being filtered out.

How signals lie: failure modes and false positives

Most false positives come from five places: pods and auto-scripts, AI-written comments, people who engage for a living, inflated or misread titles, and stale profile updates. Each has a check that takes about five minutes, and skipping it is how engaged-looking lists end up full of people who will never buy.

1. Pod or auto-script comment

LinkedIn's VP of Product Management, Gyanda Sachdeva, defined the target as comments "posted to LinkedIn through a third party auto-script or a browser plugin without any human oversight or review". When LinkedIn detects these, it removes them from the "Most Relevant" comment sort. Its help documentation also says it may limit the visibility of comments where it detects "excessive comment creation or use of an automation tool".

Check: under "Most Recent", look for a burst of comments right after the post went up, generic wording, and the same commenters showing up on the person's other posts. These timing and wording tells come from third-party automation-tool vendors. LinkedIn hasn't confirmed them, so treat them as rules of thumb.

2. AI-written comment

False positive: a fluent comment that restates the post and asks nothing specific. Check: compare it with the person's last 5 comments. If unrelated posts get comments with an identical structure, the comments are automated, however well written they are.

3. Recruiter, peer, or investor engagement

False positive: a recruiter, a competitor's rep, or an investor engaging with category content. Check: the current employer must be the target account, and the function must sit on the buying committee. This failure is common: a LinkedIn outreach benchmark across 15M+ contacts found HR replies at 10.9%, the highest of any group, which inflates "engaged" lists with people who aren't buying.

4. Senior title at a company under 50 people

False positive: a "VP" with no team. Check: headcount band, plus whether anyone reports to them.

5. UK title read by US conventions

False positive: skipping a UK "Sales Director" who is actually the top sales leader, because the filter expects "VP". Check: the table above, where UK Directors outnumber UK VPs 45 to 1.

6. Stale or cosmetic job-change update

False positive: an old move that a tracker surfaced late, or a promotion that only renamed the title. Check: the start date on the profile. Replies reportedly fall off after 30 days, so a move from four months ago that just appeared in a feed isn't a fresh signal.

7. Like or follow treated as intent

False positive: passive scrolling. Check: never enroll on a like or follow alone. Even a vendor that publishes a 3-day like window says to pair it with another signal.

The procedure: from a captured act to a sequence

Run every captured act through eight checks in order. Most acts drop out at step 2 or 3, and that's the point. The full run takes about 25 to 30 minutes of SDR and AE time for one person.

Step 2 is where most of the time goes, because the employer on a profile is often out of date. Checking the person's current role across LinkedIn, the open web, and the public GitHub graph in one pass is the kind of lookup Refolk handles well. Steps 3 and 4 still need a human eye.

Sources disagree on ordering. One school recommends 2 to 3 engagement touches over a couple of weeks before any ask. Two signal vendors push for enrollment within 24 to 48 hours. My rule: for rows with a 7-day or shorter shelf life, enroll fast. For topic-only rows (11, 12, 23 to 25), engage first and let the corroborating signal set the clock.

Person:
Current employer (confirmed on):
Function / buying-committee seat:
Act (row number):
Act URL:
Act date:
Mimic screen: pass / discard (reason)
Headcount band:
Title flag: none / under-50 / non-US convention
Age at review (days):
Row cutoff (days):
Corroborating signal:
Corroborating URL and date:
Decision: enroll / nurture until (date)

Writing the first message: cite the topic, not the act

The first message should name the topic the person engaged with and never mention the act you watched. Mentioning the like, the view, or the comment adds nothing persuasive and makes the message more likely to feel like surveillance.

Gartner found that content is 3.6 times more likely to be perceived as creepy when marketers use ten data dimensions instead of one. Each observed act you mention is another dimension. The topic is already implied by the conversation, so citing it adds nothing to the count. One outbound agency's guidance draws the same line: "Mentioning a conference talk signals homework, bringing up vacation photos signals you went looking where you weren't invited."

The limit: no published study measures how B2B buyers react when outreach mentions a specific like or profile view. The rule rests on the general finding about creepiness, not on a direct test. A public talk (rows 23 and 24) is the exception you can safely cite, because the speaker chose to be on stage.

Subject: {topic in their words} at {company}

Hi {first name},

Teams working on {topic in their words} at {company}'s size usually run into {specific problem} around {stage or trigger from the corroborating signal}.

{One sentence on how a comparable team handled it, with a concrete outcome you can stand behind.}

Worth a 20-minute look at how that would apply to {company}? If the timing is off, tell me when to check back.

{Your name}

Keeping the reference current and the collection clean

The shelf lives in this reference are borrowed claims, so replace them with your own as soon as you have the data. Keep collection inside each platform's rules while you do it.

Tuning the cutoffs locally:

• Log the signal's age at enrollment on every record (the template above has the field).
• Once a row has enough volume, plot reply rate by age band and find where it drops off.
• Replace the vendor cutoff with your observed drop-off point, and note the date you changed it.
• Re-run the comparison each quarter, because platform ranking changes can move the curve.

Collection rules to re-check before automating anything:

• LinkedIn. Reading public posts and comments by hand is fine. LinkedIn does not allow "crawlers, bots, browser plug-ins, or browser extensions that scrape, modify the appearance of, or automate activity". It also bans fake engagement, "including any tools or services that try to manipulate LinkedIn's content algorithms". Pod participation on your own side breaks the same rules that make prospects' pod comments worthless.
• Reddit. The free API tier allows 100 queries per minute per OAuth client ID and is for non-commercial use only. Commercial use needs approval. One secondary source reports that Reddit said it will gradually start restricting all new requests, so read the current terms before building a pipeline on it.
• Review sites, speaker listings, and podcast guest lists. I haven't established the collection terms for these. Read each site's terms before automating, and collect by hand until you have.