How a B2B SaaS company grew qualified leads by 3x in 90 days using AI-assisted nurturing

The problem: a full pipeline that wasn't converting

Fieldnote, a B2B SaaS company providing project intelligence software to mid-market construction firms, had a lead generation problem that looked, on paper, like a lead generation success. Their MQL volume had grown 40% year-over-year. Their content program was producing consistent traffic. Their SDR team was booking demos. And yet, pipeline coverage was shrinking, deal cycles were lengthening, and revenue per marketing-sourced lead had dropped for three consecutive quarters.

The diagnosis took a full pipeline audit. What the data revealed was a classic B2B growth trap: Fieldnote's marketing had gotten very good at generating top-of-funnel volume without improving the quality or timing of what happened next. Leads were entering the CRM, being passed to SDRs on a fixed schedule, and either converting immediately (the rare high-intent case) or going cold — often permanently.

The team was operating what their VP of Marketing, Dara Osei, called "batch-and-blast nurturing" — a single email sequence triggered on MQL status, regardless of how the lead had behaved, what content they'd consumed, or what their buying stage actually was. A VP of Operations who had downloaded a technical integration guide was receiving the same drip as a junior project manager who had read a blog post about industry trends. The sequences didn't differentiate. The timing didn't adapt. And the SDR handoff happened on day seven regardless of whether the lead had shown any additional engagement signals.

40%MQL volume growth YoY — but pipeline shrinking7 daysfixed handoff timer — regardless of intent62%of MQLs went cold before any SDR engagement–23%revenue per marketing-sourced lead, Q3→Q4

The root cause wasn't a content problem or a sales problem. It was a signal problem. Fieldnote had the data to identify where each lead was in their buying journey — they just weren't using it to differentiate the experience. Every lead was treated as equally ready and equally interested, which meant the ones who were genuinely ready were buried in the noise, and the ones who needed more time were prematurely pushed into sales conversations they weren't ready to have.

02

Why traditional nurturing fails in B2B SaaS

Before rebuilding the system, it's worth understanding precisely why standard marketing automation nurturing underperforms in B2B SaaS — because the failure mode is structural, not executional. It's not that the emails are bad. It's that the logic is wrong.

The three fundamental flaws

Time-based, not behavior-basedTraditional drip sequences fire on a schedule — day 1, day 3, day 7. But B2B buyers don't move on a schedule; they move on triggers. A lead who visits your pricing page on day 4 is fundamentally different from one who doesn't. Static sequences can't distinguish between them.Persona-blind content routingMost nurture programs assign content based on the lead's first-touch action, then don't adapt. If your first content asset was a top-of-funnel blog post, you get top-of-funnel nurturing — even if you've since visited the product documentation, watched a demo video, and checked your competitor's pricing page.Individual-level, not account-levelIn B2B SaaS, deals involve multiple stakeholders. A single-contact nurture program misses the most important signal of all: when multiple people from the same company are engaging simultaneously. That's a buying committee forming — and it's the highest-intent signal available.Premature handoff logicMQL scoring models that trigger SDR handoff on lead source and job title alone ignore behavioral indicators that far better predict conversion readiness. A lead who has engaged with six pieces of content, attended a webinar, and returned to the site three times in a week is more ready than one who downloaded a whitepaper once and never returned.

"We were measuring marketing performance by MQL volume, which meant we'd optimized for generating the appearance of interest — not for identifying genuine purchase intent. The funnel was wide at the top and leaking everywhere below it."

The AI-assisted approach that Fieldnote implemented with Reacusto addressed all four of these failure modes simultaneously. It replaced the time-based trigger logic with behavioral signal triggers, introduced content routing based on revealed intent rather than initial action, incorporated account-level engagement aggregation, and rebuilt the handoff model around a composite readiness score rather than a binary MQL flag.

03

Building the AI-assisted nurturing architecture

The architecture Reacusto built for Fieldnote has five layers, each feeding the next. Understanding the structure as a whole is essential before implementing any individual component — because the value of each layer compounds when connected to the others.

Layer 1Signal ingestionUnified behavioral data collectionAll behavioral signals — website visits, content downloads, email opens and clicks, webinar attendance, product trial activity, and G2/review site visits — are unified into a single contact-level timeline. Every touchpoint is tagged with intent category: educational, evaluative, competitive, or purchase-signal.Layer 2Intent classificationAI-powered intent stage assignmentReacusto's model classifies each contact into one of four intent stages based on their signal pattern: Awareness, Consideration, Evaluation, or Decision. Stage assignment is dynamic — it updates in real time as new behavioral signals arrive, not on a weekly batch cycle.Layer 3Account aggregationBuying committee detectionIndividual contact signals are aggregated at the account level. When three or more contacts from the same company are classified as Consideration or Evaluation simultaneously, an account-level "buying committee signal" is triggered — one of the highest-value events in the entire nurturing model.Layer 4Content routingIntent-matched content deliveryBased on intent stage, persona (determined from job title and firmographic data), and content consumption history, Reacusto's engine selects the next content asset from a mapped content library. No contact receives content they've already engaged with, and no contact receives content more than one stage ahead of their current classification.Layer 5Handoff logicComposite readiness scoring for SDR routingContacts are routed to SDRs only when their composite readiness score — a weighted combination of intent stage, engagement recency, account-level signals, and persona fit — crosses a calibrated threshold. SDRs receive a pre-populated context card showing the contact's engagement history, intent stage, and recommended opening message.

The critical design decision in this architecture is that layers four and five are dependent on layers one through three. You cannot implement intelligent content routing without accurate intent classification. You cannot calibrate the handoff threshold without account-level aggregation informing the readiness score. Teams that attempt to deploy just the "personalized email" layer without building the signal and classification infrastructure beneath it will get marginally better open rates and nothing more.

04

The lead scoring model that actually works

Fieldnote's original lead scoring model was a standard fit-plus-engagement framework: demographic fit (job title, company size, industry) plus a basic engagement score derived from email opens and website visits. The composite score triggered an MQL at 100 points. It had a 19% SQL conversion rate, which sounds reasonable until you realize that means 81% of SDR time was spent on contacts who wouldn't convert.

Rebuilding the scoring model with behavioral intent signals

The new model adds three scoring dimensions that the original model lacked entirely: intent signal type, engagement recency decay, and account-level amplification. Together, these dramatically improve the signal-to-noise ratio of the handoff threshold.

Signal

Score

Intent weight

Priority

Pricing page visit (2+ sessions)

+35

Purchase signal

High

ROI calculator completion

+30

Evaluation

High

Demo video watched >80%

+28

Evaluation

High

Integration docs page visit

+22

Evaluation

High

Webinar attendance (live)

+20

Consideration

Medium

Case study download (vertical match)

+18

Consideration

Medium

Email click (bottom-of-funnel CTA)

+15

Consideration

Medium

Blog post read (3+ in session)

+8

Awareness

Low

Newsletter open (no click)

+3

Passive

Low

Recency decay and account amplification

Every behavioral score decays at a rate of 10% per week of inactivity. A contact who engaged six weeks ago and hasn't returned has a fundamentally different readiness profile than one who visited yesterday — even if their raw cumulative score is the same. The decay function ensures that score reflects current intent, not historical curiosity.

Account amplification multiplies the individual readiness score by a factor when account-level signals are present. A contact at a company where two or more colleagues are also in active nurture receives a 1.4x multiplier. Three or more colleagues triggers a 1.8x multiplier and generates an account alert in Reacusto's dashboard for immediate SDR review — regardless of whether the individual contact has crossed the handoff threshold independently.

The most common scoring model mistake is over-weighting job title fit and under-weighting behavioral signals. A perfect-fit persona who hasn't engaged is less ready than an imperfect-fit contact who has visited pricing, watched the demo, and returned three times this week. Behavioral recency should outweigh demographic fit in the final composite score.

05

Content sequencing by intent signal

The content library restructuring was one of the most labour-intensive parts of the Fieldnote implementation — and one of the highest-impact. Most B2B content programs are organized by topic or format. Reacusto restructured Fieldnote's content library by intent stage and persona, creating a routing map that the AI engine could traverse based on each contact's classification.

The four-stage content map

1Awareness stage — problem recognition content. Contacts classified as Awareness have shown interest signals but no evaluation intent. Content served here is educational and industry-framing: benchmark reports, trend analyses, "state of the industry" guides, and problem-definition content that helps them articulate the challenge they're experiencing. No product content at this stage. Goal: advance to Consideration by helping the contact recognize and name their problem clearly.2Consideration stage — solution category content. Contacts who have engaged with Awareness content and returned for more are moved to Consideration. Here, content shifts to solution category education: "what to look for in a project intelligence platform," buyer's guides, capability comparisons, and use-case-specific content matched to the contact's industry vertical. The goal is to help them build an evaluation framework — one that, ideally, maps to Fieldnote's strengths.3Evaluation stage — proof and differentiation content. Evaluation contacts are actively comparing vendors. Content here is proof-heavy: customer case studies matched by industry and company size, ROI calculators, integration documentation, security and compliance materials, and analyst reports where Fieldnote is positioned favorably. This is also where the live demo invite enters the sequence — but only after at least two Evaluation-stage content engagements. Sending the demo invite too early (before the contact has consumed proof content) reduces show rates significantly.4Decision stage — conversion-enabling content. Decision-stage contacts have high composite scores, recent engagement with pricing or integration content, and in many cases have had initial SDR conversations. Content here accelerates the final buying decision: proposal templates, contract overview materials, implementation timelines, customer reference offers, and executive-to-executive outreach templates for the SDR to use with economic buyers. At this stage, the nurturing is largely sales-assisted rather than automated — but Reacusto continues to feed engagement data to the SDR in real time.

The demo invite timing finding

One of the most operationally significant findings from Fieldnote's implementation was around demo invite timing. In the original nurture program, the demo CTA was included in the first email sequence — present from day one. In the restructured model, the demo invite only fires after a contact has consumed at least two Evaluation-stage content assets.

The impact was stark: demo show rates increased from 31% to 67%. No-shows dropped from 69% to 33%. And the quality of demo conversations improved measurably — SDRs reported that contacts who arrived via the intent-gated demo invite had already self-educated on the product category, asked more specific questions, and required less foundational explanation during the call.

"The demo invite used to be the loudest thing in our nurture sequence. We put it everywhere because we wanted meetings. What we actually needed was qualified meetings — and those only come when the contact has done their own research first."

06

The 90-day results and what drove them

At the 90-day mark after full implementation, Fieldnote's marketing team ran a cohort comparison between MQLs generated in the 90 days prior to implementation and those generated in the 90 days post. The results exceeded the targets set at the project kickoff.

3.1xincrease in qualified leads (SQL output)67%demo show rate, up from 31%41%reduction in average sales cycle length2.8xSDR productivity (SQLs per rep per month)

Attribution breakdown: what drove the 3x result

The 3.1x SQL increase wasn't driven by a single change. Analyzing the contributing factors, Reacusto's attribution model identified four primary drivers:

Reactivation of dormant leads (accounts for ~38% of the SQL increase). The behavioral scoring model surfaced 212 contacts in the existing database who had gone cold under the old model but were actively re-engaging with website and content assets. These contacts had high fit scores but had never crossed the old MQL threshold because their engagement was spread over time. The recency-decay model identified their current re-engagement as a high-priority signal, routing them back into active nurture immediately.Buying committee detection triggering earlier account-level action (accounts for ~29%). Seventeen accounts generated buying committee alerts in the 90-day period — meaning three or more contacts were simultaneously in active nurture. In twelve of these accounts, no single contact had crossed the individual handoff threshold, so under the old model, no SDR action would have been taken. Under the new model, account-level alerts generated proactive SDR outreach that resulted in eight new opportunities.Intent-matched content improving progression rates (accounts for ~21%). The Awareness-to-Consideration progression rate increased from 18% to 44%. This was the direct result of routing contacts to stage-appropriate content rather than a fixed sequence. Contacts whose first-touch was a blog post now received a second content asset aligned to their revealed interest topic rather than a generic product overview email, which dramatically improved second-touch engagement.Demo invite timing improvement (accounts for ~12%). The shift to intent-gated demo invites, combined with the SDR context card providing a personalized opening message, improved both show rates and post-demo conversion rates. SDRs closed at a 28% higher rate from intent-gated demo attendees compared to the previous batch-scheduled demo invites.07

How to implement this for your team

The Fieldnote implementation is a reference architecture, not a template. Your specific configuration will depend on your existing tech stack, content library depth, SDR team structure, and the complexity of your buyer journey. What follows is a prioritized implementation roadmap that applies regardless of company size.

Phase 1 — Signal infrastructure (weeks 1–2)

1Audit your existing behavioral data. Before implementing any scoring or routing logic, map every touchpoint where you collect behavioral signals today: website analytics, email platform, CRM activity log, product analytics (if applicable), and ad platform engagement. Identify gaps — most companies are missing intent signals from third-party review sites (G2, Capterra) and webinar platforms.2Implement unified contact timeline tracking. All behavioral signals should flow into a single record per contact. If you're using HubSpot or Salesforce, this means configuring custom activity objects or using a CDP layer (Segment, RudderStack) to route events into a unified timeline. This step is prerequisite to everything else — you cannot build intent classification on fragmented data.3Tag every content asset by intent stage and persona. Go through your content library and assign each asset to an intent stage (Awareness, Consideration, Evaluation, Decision) and a primary persona. This exercise will also surface content gaps — most B2B SaaS companies are heavily over-indexed on Awareness content and severely under-resourced at Evaluation and Decision stages.

Phase 2 — Scoring model (weeks 3–4)

4Rebuild your scoring model with behavioral intent signals. Use the signal scoring table in Section 04 as a starting point. The specific point values matter less than the relative weighting — purchase-signal behaviors (pricing page, ROI calculator) should be weighted 3–4x higher than passive awareness behaviors (blog read, email open). Start conservative on the MQL threshold and tighten it based on SDR feedback over the first 30 days.5Implement recency decay. Most marketing automation platforms support score decay natively (HubSpot calls it "score degradation"). Configure it to decay behavioral scores at 10% per week of inactivity. Demographic fit scores should not decay — a CFO at a 200-person SaaS company doesn't become less of a fit over time.6Add account-level aggregation. Configure your CRM to generate account-level alerts when two or more contacts from the same company cross the Consideration threshold simultaneously. This requires connecting individual contact scores to their associated account record — straightforward in Salesforce, achievable in HubSpot with custom properties and workflows.

Phase 3 — Content routing and SDR handoff (weeks 5–8)

7Build intent-gated content sequences. Replace your existing email sequences with stage-triggered workflows. Contacts entering Consideration receive Consideration-stage content; contacts advancing to Evaluation receive Evaluation-stage content. Build a minimum of three content assets per stage before going live — you need enough depth to avoid repetition within a single stage for contacts who engage quickly.8Gate the demo invite behind two Evaluation-stage engagements. This is a change that will feel counterintuitive to SDRs accustomed to high-volume outreach. The data strongly supports it. Configure your demo invite trigger to fire only after a contact has engaged with two or more Evaluation-stage assets and has a composite score above your calibrated threshold.9Build the SDR context card. When a contact is routed to an SDR, provide a pre-populated context card showing: the contact's intent stage classification, their engagement history (last three touchpoints), account-level signals (if present), and a recommended first message angle based on their highest-intent signal. This is the difference between an SDR cold-reading a CRM record and an SDR walking into a conversation with genuine context.Before you start: prerequisites checklist

• Unified behavioral tracking across web, email, and product (or a plan to implement it)
• Content library with a minimum of 3 assets mapped to each intent stage
• CRM configured with account-level contact association
• SDR team briefed on the new handoff model and context card format
• A 90-day cohort comparison plan agreed before go-live (so you can measure impact cleanly)
• Marketing and sales alignment on the new MQL definition — this is a political as much as a technical requirement

The most important mindset shift

Every technical component in this guide is implementable with existing marketing automation tools, a well-structured CRM, and a content library of reasonable depth. The harder change is organizational. AI-assisted nurturing requires marketing teams to stop measuring success by MQL volume and start measuring it by SQL conversion rate and pipeline quality. It requires SDRs to trust that a lower volume of better-qualified leads will outperform a high volume of poorly-qualified ones. And it requires leadership to accept a short-term dip in MQL numbers while the model calibrates — typically two to four weeks — before the SQL improvement becomes visible.

Fieldnote's Dara Osei put it plainly in a retrospective at the 90-day mark: the 3x SQL result was less about the technology and more about the willingness to define "qualified" rigorously, measure against it honestly, and rebuild the entire pipeline around that definition. Reacusto's AI layer accelerated the signal detection and removed manual work from the routing logic. But the strategic clarity had to come first.

"If you're not willing to redefine what a qualified lead means and hold the whole system accountable to that definition, no amount of AI-assisted tooling will change your pipeline outcomes. The technology executes the strategy. It doesn't replace the need for one."