Your CRM is full of alerts nobody acted on. Three people from one company commented on your last post on Tuesday; a contact you lost a deal to last year changed jobs into exactly the seat you sell to. Both sat there until Friday, by which point those buying signals had become trivia. Plenty of signal, almost no response — that is the honest state of signal-based selling at most B2B companies.
The instinct is to buy more signal: another intent feed, another tracking pixel, another alert nobody owns. Supply was never the constraint. Most third-party B2B intent data is a probability, not an event — it says something may be happening at an account, not who decided what, and it arrives with no owner and no play attached.
What follows fixes the demand side. Rank signals by how much they should change a rep's behavior, attach a decay window and one play to each, then route the best ones to a human the same day.
Why Most B2B Intent Data Is a Probability You Rented
Third-party intent watches content consumption across publisher networks and flags when an account's research runs above its own baseline. Real information — but account-level and probabilistic by construction. Someone inside a 400-person company read more about your category this month than usual. Not who. Not why.
Take one in five flagged accounts as noise — a working assumption, not a measurement — and a feed of 400 a week hands your reps eighty dead ends to discover personally. Within a quarter the surge list is a tab nobody opens. Four things go wrong:
- It is account-level, not person-level. You get a company and a topic. Turning that into a named human who owns the problem is still the expensive part.
- It is late. A surge only registers once a pattern establishes itself, which usually means the shortlist already exists.
- Competitors rent the same feed. Anything sold as a data product is not proprietary. Everyone in your category gets the same alert on the same Tuesday.
- It arrives without an owner. The feed does not assign, does not expire, and does not say what to do. So nobody does anything.
That is a prioritization input, not a trigger.
A Buying Signal Is Only Worth What Your Response Time Makes It Worth
Every signal has a half-life, and it is shorter than you think. A reply is live for hours — the person is at their desk, thinking about what you asked. A comment on your post is warm for a day or two, a reaction for less, after which referencing it reads as surveillance. A job change holds for the first quarter in seat, when a new leader is still allowed to change things.
Which gives you the rule the rest of this hangs on: a signal without a defined play and a response deadline is just a notification. The practice worth killing is the shared alerts channel, where every trigger lands in one undifferentiated stream. It feels like visibility. It guarantees nobody is responsible for any specific signal — which is why so many teams can prove they detected a deal they never worked.
The Four Sources of Buying Signals, Ranked by Trust
Sort signals by how directly you can see them. The closer one sits to something you generated yourself, the more it should move a rep.
- First-party. Anything on surfaces you own — replies to your outreach, comments and reactions on your LinkedIn posts, and whatever your own product analytics show about trial accounts. Lowest volume, highest trust, and the only category where you know exactly who.
- Second-party. Someone else's first-party data, shared deliberately: review-site comparison activity, partner referrals, event attendee lists. Person-level and explicitly commercial, with one catch — several vendors watch the same person.
- Third-party intent. The rented, aggregated layer described above: it reports companies, not people. Broad reach, weak resolution. Prioritize with it; do not trigger on it.
- Public and situational. Job changes, funding rounds, hiring patterns, leadership appointments, tech-stack changes. Free, verifiable, person-level, and independent of whether anyone has heard of you. These are the classic sales triggers, and they hold up longest.
The order has an uncomfortable implication: the best signal layer is one you manufacture, not one you buy. Publishing to the audience you sell to produces first-party intent every week, free, at person-level resolution and invisible to your competitors. That is the underrated case for content — not awareness, but a private signal feed, and the reason running inbound and outbound as one All-Bound engine beats running them as rivals.
Rank Signals by Decay Window and the Play They Trigger
Write your own version before you automate anything. Every signal needs an implication, a window and one named play — if you cannot fill all three columns, it is not ready to reach a human.
| Signal | What it implies | Decay window | The play |
|---|---|---|---|
| Comment on your post | Their own wording for the problem, attached to a real employer | 24-48 hours | Route it like an inbound lead, not a social notification |
| Reaction on your post | A name and an employer, without their wording | 12-24 hours | Score for fit, then stack it — too thin to trigger on alone |
| Reply to a sequence, including a no | A live human with an opinion about your category | Same day | Automation stops; a person takes over |
| Job change into an ICP role | New mandate, new budget, permission to replace things | First 90 days | Re-introduce, referencing what they owned before |
| Open role naming your problem | They are fixing it with headcount | 2-4 weeks | Quote the responsibility from the posting, offer the alternative |
| Third-party topic surge | Someone there is researching the category | Weeks, fuzzy | Never alone; move the account up an existing cohort |
Notice what is missing. Open tracking has been unreliable since mail clients started prefetching images on the recipient's behalf, so a play triggered by an open is closer to a coin flip. Treat opens and clicks as tiebreakers, never as reasons.
Stack Weak Signals Instead of Firing on One
Single-signal automation on a thin signal is how outreach gets creepy and pipeline gets fake. Fire on one thin signal and the message either overstates what you know or has nothing to say beyond the trigger. Fire on a combination and it writes itself, because a combination is a story.
- Job change plus role relevance. A new VP of ops, thirty days in, whose team owns the process you improve. Either alone is thin; together they describe a mandate and a problem.
- Third-party surge plus first-party engagement. The account was already researching; now a named person from it has reacted to two of your posts this week and commented on a third. The rented signal supplied the company; your own feed supplied the human.
- Engagement plus fit. Someone commented on your post about a specific failure mode, holds a title in your buyer set, and sits in your size band. The comment does not make them a buyer — it dates the fit you already had to this week.
Formalize that and you have arrived at scoring, which is a solved problem. The two-axis fit and intent model is the right container — signals feed the intent axis, decay out of it, and only high-fit leads earn a same-day play, so a hyperactive bad fit never outranks a quiet perfect one.
Reference the signal without reciting it
Name public signals explicitly; let private ones shape only the timing. A prospect expects you to know they changed jobs. They do not expect you to keep a tally of every post they touched.
Weak, and slightly alarming: "Hi Dana, I saw you liked three of my posts this week — got 30 minutes?"
Better: "Hi Dana — you're two months into the ops seat and hiring two analysts, which usually means the routing you inherited is dropping leads somewhere. It normally breaks at the handoff. Worth ten minutes?"
The second uses the private signal (the engagement) for timing and the public ones for the opening line, then makes a claim instead of asking for a slot. Our trigger-led cold email templates are organized the same way — by the signal that justifies sending — and the rule carries over unchanged to a LinkedIn DM.
Build the Routing: Capture, Score, Assign, Escalate
Every step here exists to stop a signal dying in a channel.
- Capture in one place. Every source writes to the same lead record. A signal living only in an email alert or a notification does not exist operationally.
- Enrich and score on arrival. Fill in title, seniority, headcount and location, then score fit from the whole profile before anything routes. Unenriched signals cannot be thresholded, and unthresholded signals route at full volume.
- Threshold deliberately. Set the bar so the daily queue is something a rep can finish. Ten worked properly beat a hundred skimmed, and a queue that is never empty is never trusted.
- Assign a person and a clock. Not a team, not a channel — a name, with an SLA matching the decay window from your table.
- Escalate, then expire. If the clock runs out, reassign once. If it runs out again, the signal expires into nurture and gets counted. Signals that rot quietly are how a team convinces itself the data was bad.
That last step is the one everybody skips, and the only one that answers "is this working?" honestly — unworked signals belong in the weekly pipeline review, beside meetings booked.
Step one is where most teams stall. LinkedIn engagement is the richest first-party signal a B2B team has and the one that never leaves the feed — nobody exports a notification. That is the gap Sendpilot's inbound campaigns close: point one at a post and everyone who engages with it is captured as a lead the moment they engage, then scored for fit by ICP Scoring. The automation answers comments carrying your Action Words — the strongest post-engagement signal in the table — and the DM follows for the commenters you are already connected to on LinkedIn; anyone else is asked to connect first. The comment-to-DM workflow is worth building whether or not you buy a tool.
The Three Numbers That Prove Signal-Based Selling Works
Most signal dashboards measure supply — the one thing you already know you have. Measure the response instead.
- Signal-to-touch time. Median and 90th percentile, per signal type. The median flatters you; the tail is where the system fails. A four-hour median beside a nine-day p90 is a routing problem, not coverage.
- Signal-to-reply rate by signal type. It tells you which signals deserve automation and which deserve deletion. Judge each type against your own cold baseline, not an industry chart. In our experience first-party triggers run several times higher than cold — teams we work with typically see them in the 15-25% band — while a signal replying at cold-list rates is a list you paid extra for.
- False-positive rate. The share of routed signals a rep marks not relevant. Above roughly 30%, in our experience, your thresholds are too loose, reps stop working the queue, and you find out a quarter late.
Leave the vanity numbers out: signals captured, accounts surging, alerts sent. They move every time you buy another data source and say nothing about whether a human spoke to anyone. If you are weighing an AI SDR layer to close the gap between detection and response, those three are the test.
The Bottom Line
Almost every team that thinks it has a buying signal problem has a response-time problem in a data-purchasing costume. The ones that win are not detecting the most — they decided in advance what each signal means, who owns it, and how long it stays worth acting on. Rank yours by how much they should change behavior, stack the weak ones rather than firing on them, and let anything you cannot staff expire honestly. Sendpilot exists to make the LinkedIn half of that loop automatic — post engagement captured and scored for fit, and the comments carrying your Action Words answered inside the window — because the signals you generate yourself are the only ones your competitors cannot also buy. Pick the one signal you detect and never work, give it an owner and a same-day deadline, and see what a week of responding does to your reply rate.
Where Sendpilot Fits
Sendpilot captures one class of buying signal, and captures it completely: engagement on your own LinkedIn posts. Point an inbound campaign at a post and everyone who engages with it becomes a lead as they engage. The automation runs on comments: when a comment carries one of your Action Words, Sendpilot replies to it from a randomizable template, and the DM — first-name and full-name variables included — goes to the commenters you are already connected to. Anyone who is not a connection gets the connect-first reply instead, asking them to send a request; once you accept, the DM goes out. That is the useful part of the gate, because a commenter who connects is worth more than a commenter who does not. Either way the lead can be transferred into an outreach campaign for follow-up messages or voice notes.
ICP Scoring gives captured leads a 0–100 fit score read from the whole profile — title, seniority, industry, company size, skills, experience and activity — so the threshold in step three is a number you set, not a judgement a rep makes at five o'clock. Scoring draws on your plan's monthly credits, the same balance enrichment and the lead extractor spend. Third-party intent feeds and your own product analytics stay separate products; Sendpilot runs the LinkedIn side, and from Growth up it can hand leads on to the rest of your stack through webhooks and the API.
Inbound campaigns and ICP Scoring are on every plan, priced per connected LinkedIn account — never per user. See /pricing for the plans, or start a free trial and point your first inbound campaign at the post that drew the most comments last month.
Frequently Asked Questions
What are buying signals in B2B sales?
Buying signals are observable actions or changes suggesting a prospect is closer to purchasing than they were before. They come from four places: first-party activity on surfaces you own, second-party data shared by partners or review sites, rented third-party intent, and public events like job changes, funding and hiring. Useful ones always come with a response window.
What is the difference between intent data and buying signals?
Intent data is one type of buying signal, usually the rented third-party kind that reports elevated research activity at account level. Buying signals is the wider category and includes replies, comments and reactions on your posts, job changes, hiring patterns and whatever your own product analytics show. Third-party intent tells you which accounts to prioritize; first-party signals tell you which person to message today.
How fast should you follow up on a buying signal?
Match the follow-up to the signal decay window. Replies and comments on your posts need a human response the same day, ideally within a few hours. Hiring signals — an open role naming your problem — stay useful for two to four weeks. A job change holds far longer: a new leader is still allowed to change things through their first quarter in seat, roughly 90 days.
Is third-party intent data worth paying for?
Only if you already act on your first-party signals reliably. Third-party intent is account-level and probabilistic, so expect a meaningful share of flagged accounts to be noise. It works as a prioritization layer over cohorts you already run, not as a trigger for outreach. Teams without a routing SLA get very little from it.
How do you use a buying signal in outreach without sounding creepy?
Reference public signals explicitly and let private ones shape only the timing. Mentioning a job change, a funding round or an open role is normal. Reciting a tally of every post they touched is not. Use the signal to justify why you are writing today, then spend the rest of the message on a specific claim about their problem.
Which buying signals does Sendpilot capture?
LinkedIn engagement only. An inbound campaign watches a post you choose, captures everyone who engages with it, and scores each person against your ICP from their whole profile. The automated response runs on comments: a comment carrying one of your Action Words gets an automatic public reply. The DM is connection-gated — commenters you are already connected to receive it, and everyone else is asked to connect first, so the DM follows only once you accept. Sendpilot does not track website traffic or trial usage — keep those in your analytics or a dedicated intent tool, and route them beside the LinkedIn signals in the same lead record.
