Signal Decay: The Outbound Metric Nobody Measures (2026 Benchmarks)

Signal decay is the outbound metric nobody measures

Every GTM team in 2026 says they run signal-based outbound. Almost none of them can tell you how old their signals are when a rep touches them. That number — the gap between the moment something actually happened at an account and the moment a human said something useful about it — is signal decay, and it is quietly the difference between a 20% reply rate and a 2% one.

The reason this stays invisible is a measurement trick. Sales intelligence vendors publish database refresh cadence. What determines whether you win the deal is event-to-alert latency. Those are different numbers and, according to freshness benchmarks published this year, they can differ by weeks.

A 2026 analysis of documented vendor refresh cadences laid the spread out plainly: some platforms genuinely touch their data daily, others weekly, others every two to four weeks, and several large contact databases show practical refresh cycles of 90 to 180 days under independent testing. None of those numbers describes how long it takes for a real-world event to reach a rep. That is the number you are actually buying, and almost nobody publishes it.

This post is about what signal decay costs, how fast different signal types actually rot, and what an outbound clock looks like when you design around decay instead of pretending it does not exist.

Why late signals cost deals: the buying window is front-loaded

The case for speed is not vibes. It is the shape of the modern B2B buying journey.

6sense's 2025 B2B Buyer Experience Report, based on responses from more than 4,000 buyers across North America, EMEA and APAC, found something uncomfortable for anyone running a monthly outbound cadence:

• Buying groups form the bulk of their vendor shortlist on day one of the journey, before any vendor is contacted.
• 94% of buying groups rank a preferred vendor before first contact, and they buy from that pre-contact favourite roughly 77% of the time.
• The winning vendor is on the day-one shortlist about 95% of the time.
• Buyers now initiate contact themselves in the large majority of cases, and they reach out first to the vendor they already intend to buy from.

6sense also reported that the point of first contact moved from about 69% of the journey to 61% — a shift worth roughly six to seven weeks of influence window that sellers used to have and no longer do.

Read those findings together and the implication is brutal. The commercially valuable moment is not when a buyer fills in a form. It is the window in which the shortlist is still being assembled — which is early, short, and almost entirely invisible unless something external tips you off. That tip-off is a signal. If the signal arrives after the shortlist is set, you are not doing outbound. You are doing archaeology.

Speed compounds inside the funnel too

The same pattern shows up further down. The 2026 Blazeo speed-to-lead benchmark, covering 573 businesses, found that 74% miss their own five-minute response window, and that slow responders were substantially more likely to report losing leads outright. Speed-to-lead research has been saying a version of this for two decades; what changed in 2026 is that the benchmark itself moved. Teams now measure the window in minutes, not hours, and AI-assisted responders hit the target far more often than manual-only teams.

The point is not that everyone needs a five-minute SLA on every signal. It is that decay is a property of the signal, not a preference of the sales team — and different signals decay at wildly different rates.

How fast do different buying signals actually decay?

Here is a practical half-life table. These windows are drawn from published research where it exists and from observed outbound performance where it does not, so treat them as planning defaults rather than laws of physics.

Two things fall out of this table immediately.

First, the highest-converting signals decay fastest. Engagement, complaints and pain-point posts are the warmest triggers available and they are worthless within a week. The signals that stay fresh for a quarter — firmographics, technographics — are the ones that produce the coldest messages.

Second, a single refresh cadence for everything is a design error. If a vendor gives you one "refresh rate" number covering news, contacts, intent and firmographics alike, they have not modelled decay at all.

The three layers of lag (and the one you can actually control)

Signal lag stacks. Understanding where it comes from tells you where to spend.

Layer 1: the source is late

This is the biggest layer and it usually sits outside any vendor's control.

• Job changes inherit self-report lag. People update their profile one to four weeks after starting a new role, and plenty never do. Any job-change alert built on profile scraping starts life weeks behind.
• Funding rounds are typically disclosed well after the paperwork closes; Crunchbase's own methodology notes that the lag is longest at seed stage.
• Registry and filing data can lag by months depending on the domain.

The fix here is not a faster pipeline. It is a different source. If you want to beat the LinkedIn self-report lag on a leadership change, you watch the press release, the company page post, the job posting for their first hire, or the earnings-call commentary — surfaces that move before the profile does.

Layer 2: the vendor's own cycle

This is the number vendors publish. The documented spread in 2026 runs from daily (6sense intent scores, G2 Buyer Intent) to weekly (Bombora Company Surge, refreshed Sundays and scored by comparing a three-week window against a twelve-week baseline) to every two to four weeks for some contact-tracking products.

Note the Bombora design detail, because it generalises: a surge score that compares recent weeks to a longer baseline trails by construction. That is not a flaw, it is what the metric is for. But it means a surge score describes roughly what happened last month, and pairing it with datestamped event signals is not optional if you need same-week timing.

Layer 3: delivery adds the last mile

This is the layer teams control completely and neglect entirely. Real-time detection plus a Monday digest email equals weekly signals. Owner-only routing on an unassigned account equals no signal. A Slack channel nobody opens equals no signal. An alert that fires into a queue a rep works down every second Thursday equals a fortnightly signal, regardless of what you paid for detection.

Most teams we talk to could halve their event-to-touch latency this month by fixing layer 3 alone, at zero incremental cost.

Build an outbound clock, not an outbound list

The mental model shift is this: stop thinking about outbound as a list you work through, and start thinking about it as a clock you respond to. Lists are a stock. Signals are a flow. Working a flow at list cadence is how you end up sending funding-round congratulations three weeks late.

Five rules for building the clock.

Rule 1: match scan cadence to decay rate, per signal type

News, funding, leadership changes and social engagement need scanning several times a week at minimum, and ideally daily. Contact and firmographic verification can run monthly. Financial fundamentals can run quarterly. Write this down as an explicit policy per signal type. If you cannot state it, you do not have one.

Rule 2: delete the weekly digest

Digests exist for the convenience of the person configuring the tool, not the rep working the account. Any signal with a half-life under seven days must route same-day, to a named owner, with the context attached. If your only delivery mechanism is a scheduled email, your effective latency is the digest interval no matter what the detection layer does.

Rule 3: front-run the lagging source

For every high-value signal type, ask: what moves before the obvious source? A few that consistently pay off:

• Job posting for a role that implies a project, rather than waiting for the tool purchase.
• Company page announcement of a leadership hire, rather than waiting for the profile update.
• Engagement on a competitor's post, rather than waiting for a review-site intent surge.
• A pain-point thread on Reddit or LinkedIn, rather than waiting for the buyer to hit your pricing page.

Rule 4: score signals by freshness, not just fit

Most lead scoring models weight firmographic fit heavily and recency barely at all. That is backwards for outbound. A perfect-ICP account with a 60-day-old signal is a worse use of a rep-hour than a decent-ICP account with a signal from this morning. Add a time-decay term to your score and watch which accounts move to the top.

Rule 5: measure event-to-touch latency as a first-class metric

This is the metric that makes the other four real. Not touches per day, not sequences launched, not connects. The median number of hours between the event firing and a rep sending something relevant about it. Instrument it, report it weekly, and set a target per signal tier.

Most teams that measure this for the first time find a median in the multiple-day range with a long tail into weeks. Getting the median under 48 hours for tier-one signals is usually worth more than any messaging rewrite.

What to ask vendors before you buy

If you are evaluating a signal or intelligence platform this quarter, the freshness conversation is the one that separates the marketing from the product. Five questions:

1. Show me timestamped examples. Pick three real events at accounts you know — a funding round, an executive hire, a competitor complaint — and ask when the platform surfaced each one. The gap between event date and alert date is the only freshness number that matters.
2. Give me cadence per signal type, in writing. "Real-time" is a marketing word. "News scanned daily, contact verification monthly, financials quarterly" is an answer.
3. What sources front-run the slow ones? A vendor that only watches the lagging source has a structural ceiling no amount of engineering fixes.
4. Walk me through the delivery chain. Detection to routing to the rep's actual working surface. Count the hours.
5. Run it against my current stack for two weeks. Count who reported each event first. This settles the argument empirically in a fortnight.

Vendors who are confident in their freshness will agree to all five. That in itself is signal.

Why decay matters more in 2026 than it did in 2024

Two structural changes have raised the price of being late.

The buying window shrank. Per 6sense, first contact moved earlier by roughly six to seven weeks of journey. There is simply less runway between "shortlist forms" and "decision made" than there was, and the shortlist forms before you get a call.

The team working the signals got smaller. SaaStr's analysis of survey data covering 560+ B2B software companies found 36% of companies reduced SDR/BDR headcount over the year — the largest decline of any sales role surveyed — against just 19% increasing it. Meanwhile Emergence Capital's Beyond Benchmarks work found that while a majority of companies use generative AI somewhere in GTM, satisfaction with AI in go-to-market motions sits well below satisfaction with AI in product.

Put those together: fewer humans, a shorter window, and AI tooling that has mostly been pointed at volume rather than timing. That combination is exactly how you end up with more messages sent and fewer meetings booked. The teams pulling ahead are not sending more. They are sending sooner, on fewer, warmer triggers.

This is the whole argument for signal-based, warm outbound over list-based cold outbound, and it is an argument about clocks rather than personalisation. Personalisation makes a timely message better. It cannot make a late message timely.

Where AI actually helps with decay

Worth being precise about this, because the market is noisy.

AI does not help much with layer 1. No model makes a person update their profile faster. What AI does compress is the gap between signal detected and rep sends something worth reading — historically the slowest human step in the chain, because it required someone to notice the alert, research the account, decide whether it was worth a touch, and write the message.

Collapsing that step from hours to minutes is where the leverage sits. That is the design principle behind Updately: capture the fast-decaying warm signals — profile views, post engagers, competitor mentions, hiring signals, pain-point posts on Reddit, LinkedIn and X — score them against ICP, research the prospect, and draft the message in the user's voice while the signal is still warm, sending safely inside LinkedIn's limits rather than blasting past them.

The important part is not the automation. It is that the clock keeps running whether or not a human is at their desk, so a Friday-evening competitor complaint does not become a Tuesday-morning archaeology project.

Key takeaways

• Signal decay is measurable and mostly unmeasured. The metric is event-to-touch latency: median hours from real-world event to a relevant human touch. Instrument it this week.
• Published refresh cadence is not event-to-alert latency. Documented 2026 cadences range from daily to quarterly, and none of them tells you how fast a real event reaches a rep.
• The highest-converting signals decay fastest. Engagement and complaint signals are worthless within a week; firmographics stay fresh for a quarter and produce cold messages.
• The buying window is front-loaded. Shortlists form on day one, 94% of buying groups rank a favourite before first contact, and that favourite wins about 77% of the time.
• Layer 3 is free money. Kill weekly digests, route same-day to a named owner, and most teams halve latency without spending anything.
• Add time decay to your lead score. A fresh signal on a decent-fit account usually beats a stale signal on a perfect-fit one.
• Test vendors on timestamps, not adjectives. Three known events, one demo call, and a two-week overlap against your current stack settles it.

If you only change one thing after reading this: pick your three highest-value signal types, measure the median event-to-touch latency on each for the last 30 days, and put the number on a slide. It is usually worse than anyone on the team expects, and it is usually the cheapest thing in the entire outbound stack to fix.