Why Intent Data Made B2B Outreach Worse (And How to Fix It)

Intent data was supposed to fix the outreach problem. Most teams have more data on their accounts than they've ever had, more signals layered into more dashboards, more "intent flags" lighting up in their CRMs every week. Reply rates haven't moved. Meetings booked per SDR haven't moved. Cost per pipeline dollar has gone up.

The argument here isn't that intent data is broken. It's that scoring outreach on inference - rather than on evidence of a buying motion - did to B2B sales what algorithmic feeds did to social media: it scaled the volume of low-quality output without improving the underlying quality of what was sent. We covered the structural argument in the Purchase Intent vs Intent Data pillar. The practical effect on outreach is what this piece is about.

The promise versus the outcome

The original argument for intent data was specific. The B2B funnel was built on firmographic targeting plus the seller's own engagement data, and that combination missed accounts that were actively in-market because they hadn't engaged with the seller yet. Intent data closed that gap: third-party publisher activity, anonymised IP-level behaviour, content syndication signals - all aggregated into a "who's researching your category right now" view.

The category sold a single promise: with more signal upstream of the seller's own engagement, sales teams could prioritise outreach against accounts that were genuinely in-market. The metric that mattered was reply rate, because reply rate measures whether the rep is reaching the right buyer at the right time.

That's not what happened in most companies. Forrester's evaluation of B2B intent data providers found 50% of teams report too many false positives. Reply rates on cold outbound have stayed flat or declined for most teams over the same period intent data spend has climbed. The category didn't fix the outreach problem. It scaled it.

Why scoring on inference makes outreach faster, not better

The mechanism is simple. Intent data scores accounts on inferred interest - someone at the account loaded a category page, an anonymised IP appeared in a bidstream, a topic surged in a third-party publisher network. The signal is real in the sense that something happened. It's just not specific enough to predict a buying motion at that account in the next 60 days.

A rep who gets a list of "high intent" accounts every Monday morning treats the list the way they used to treat the ICP list - as a queue to work through. The cadence stays the same: day 1 email, day 3 LinkedIn, day 5 call. The messaging stays the same: a value-prop opener wrapped around a content piece. The only thing that changes is that the rep feels more confident the account is in-market, because the score told them so.

That confidence is the problem. When the signal is real evidence of a buying motion - a competitor trial just started, a renewal opens in 47 days, a vendor was recently churned - the rep can open with the named event and the message lands harder. When the signal is a score, the rep opens with a generic value prop and the message reads the same as every other outreach the buyer received that week. Same touch, more confidence behind it, no improvement in reply rate. The data layer changed; the cadence didn't.

The compounding effect: scoring the same accounts at higher confidence means the team works those accounts harder. More touches per account, more sequence escalation, more inbound from marketing. The signal didn't tell the team to find different accounts. It told the team to push the existing ones harder.

The three ways intent data quietly broke outreach

Three specific failure modes recur across teams that scaled intent data without rethinking the cadence underneath.

Generic touches feel more confident, so reps send more of them

A rep who's been told an account is "high intent" sends touches with more conviction than a rep working from an unscored list. That conviction doesn't translate into a different message - the rep doesn't suddenly know what specifically the account is doing - but it does translate into volume. More touches, longer sequences, more aggressive follow-up.

The buyer experiences this as more pressure with no new information. The intent score told the rep the account was a priority. The buyer's actual buying motion (or lack of one) hasn't changed. The rep is now pushing harder on an account that wasn't necessarily buying. Reply rate drops. The score the rep relied on starts being correlated with worse outcomes, not better ones.

The bidstream-correlated cohort problem

Most third-party intent data is sourced from the same underlying signal layer - a small number of bidstream networks, content syndication co-ops, and aggregator publishers. Two intent vendors selling to two competing sellers in the same category will often flag the same accounts as "high intent" in the same week, because they're seeing the same underlying behaviour.

The buyer gets contacted by both sellers (and often a third and fourth) inside the same window with similar messaging. Each seller thinks they're early. They're all arriving simultaneously. Reply rates on whichever seller arrives first are mediocre; reply rates on every subsequent seller are worse. The shared signal layer scales coordinated noise at the buyer without anyone intending to.

The intent-confidence loop

Most intent data products score recency and intensity together. An account that's surged in the last 30 days gets a higher score than one that surged 90 days ago. A rep who works the high-recency, high-intensity scores will, on average, get marginally better engagement - because the score correlates loosely with actual buying motion.

The trouble is that the correlation is just strong enough to keep the team using the layer, but not strong enough to compound. Reply rate on top-decile intent accounts is higher than reply rate on the rest of the list. But it's not high enough to fundamentally change the team's pipeline math. The team gets a small lift, takes the win, and assumes the layer is working. Over time, the team scales harder on the layer (more spend, more reps, more sequences) and the lift doesn't compound - it dilutes. The signal got noisier as more sellers piled into the same vendors.

This is the most insidious failure mode because it doesn't look like a failure mode. Reply rate is up - just not by enough to matter at the bottom of the funnel.

What the data actually looks like

The pattern in practice, across teams that scaled intent data hard:

• Cold email reply rates of 1-3% on "high intent" accounts - inside the range of cold email rates on any reasonable B2B list. The intent score isn't producing a fundamentally different funnel.
• 3 in 100 of "high intent" accounts producing a meeting within 30 days - the Tanvi framing from candidate sales calls. The score correlates with interest, not with a near-term buying motion. Most "high intent" accounts are out-of-market on a quarterly cycle and stay that way.
• Marketing-sourced pipeline contribution flat year-over-year despite higher MQL volumes from intent-driven audience segmentation. The volume goes up; the conversion to opportunity stays the same; net pipeline contribution doesn't grow.
• Rep adoption of the intent layer dropping in months 4-6 - a leading indicator that reps are quietly stopping using a layer they were told would change their week. The CRM keeps the signal field populated. Nobody sorts on it.

None of these data points are damning on their own. Together they describe a category that delivered a small lift sold as a large one, and then over-scaled the small lift into noise.

The fix: intent as a filter, evidence as the trigger

Intent data has a legitimate use case. It works as a coarse top-of-funnel filter - useful for audience segmentation in paid media, useful for narrowing a 500,000-account universe down to a 50,000-account research priority. The mistake most teams made was using it as the trigger for outreach, not as the filter above outreach.

The corrective is to put a different kind of signal underneath the intent layer - one that produces evidence of a real buying motion at a specific account. Subscription Intelligence is one category of that signal layer - it detects observable buying events (competitor trials starting, renewal windows opening, vendors churned, tech stacks changing) at the account level, sourced from observable data, not from inferred behaviour. We've covered the broader category of evidence-based platforms in the Best Purchase Intent Data Platforms for 2026 directory.

Put intent on top and evidence underneath, and the workflow inverts. Outreach gets triggered by the evidence - the trial that started, the renewal that opened - not by the intent score. The cadence anchors on the named event. The rep opens with something the buyer recognises, not with a value prop. The score gets used for audience targeting and territory planning, not for sequencing decisions.

What fixing it looks like in practice

Four moves separate teams that fixed this from teams that scaled the noise.

1. Stop using the intent score as the queue sort. The rep's working list should sort on a time-bound event (renewal opening, trial started, vendor churned), not on a static or rolling intent score. The score becomes a filter, not the sort.
2. Anchor every cold touch on the named event. "I saw your team started evaluating Vendor X four days ago" lands. "I noticed your company has been showing intent in our category" doesn't. The opening line tells the buyer the rep knows something specific - or it tells them they don't.
3. Audit what the layer flagged 90 days ago. Pull the cohort of accounts your intent layer flagged as high-intent a quarter back. How many converted to an opportunity? How many are still flagged at the same intensity? The delta is the layer's actual signal quality.
4. Pair the intent layer with an evidence layer. Intent for audience and segmentation, evidence-based purchase intent for sequencing and prioritisation. Two layers, two different jobs - not one layer asked to do both badly.

Teams that make these four moves see the lift the original intent-data category promised but never delivered. Not because they replaced the intent layer - because they stopped asking it to do a job it was never structured to do.

Frequently Asked Questions

Is intent data dead?
No. Intent data still works as a top-of-funnel filter and as an audience-targeting layer in paid media. It's the wrong signal to base sales prioritisation on, especially in B2B SaaS where the buying window is narrow and the inferred-intent signal sits too far upstream of the actual buying motion.

Why have my intent data reply rates dropped?
Most likely because more sellers in your category are now sourcing from the same underlying intent data and contacting the same accounts in the same week. The signal didn't degrade - the surrounding competitive density did. The shared bidstream layer scales coordinated noise even when no individual seller intends it.

What should I use instead of intent data for sales prioritisation?
Evidence-based purchase intent - the observable buying events at an account (trials, renewals, churn, tech-stack changes) - is the layer that produces a trigger a rep can actually anchor outreach on. The pillar piece on purchase intent vs intent data covers the difference, and the platform directory covers who produces it.

Can I keep my intent data tool and add an evidence-based layer on top?
Yes - in fact that's the recommended stack for most teams. Intent for audience filtering and territory planning; evidence-based purchase intent for sequencing and sales prioritisation. Two layers, two different jobs. They don't compete.

How do I tell if my intent layer is producing signal or producing noise?
Run a 90-day cohort audit. Pull the accounts your intent layer flagged as high-intent 90 days ago. What proportion converted to a real opportunity? What proportion are still flagged at the same intensity without converting? A high "still flagged, not converted" rate means the layer is producing recurring noise on the same population of accounts, not real signal.

How long does it take to see results from a signal-led, evidence-anchored motion?
Reply-rate lift is usually visible within the first sequencing cycle (30-45 days), because event-anchored opening lines convert better than generic value-prop openers. Pipeline impact tracks to the underlying sales cycle - a 90-day cycle produces a 90-day measurement window. By month four most teams have enough data to confirm the new motion.

Is this a Niall O'Gorman / Episode 1 argument?
The "intent data accelerated the noise" framing comes from Episode 1 of Intent Decoded, where the conversation centred on intent data as a starting gun rather than a confirmation signal. The full episode is on the podcast hub.

Recommended reading

• Purchase Intent vs Intent Data - the pillar this spoke ladders into.
• Best Purchase Intent Data Platforms for 2026 - the directory of evidence-based platforms that produce the layer this piece argues for.
• Most GTM Teams Use Intent Data at the Wrong Time - the Episode 1 recap where the "starting gun, not confirmation signal" framing first ran.
• What Is Subscription Intelligence? - the data category that produces the evidence-based signal layer.
• Your Sales Outreach Isn't Broken. Your Timing Is. - the timing-side of the same argument.
• Why Qualified Leads Don't Close (And How Timing Kills Conversion) - the downstream effect on the funnel of the failure mode this piece describes.