A founder I worked with last year showed me his nurture program. Six emails, sent on day 1, day 4, day 8, day 14, day 21, and day 30. Same six emails to every MQL who downloaded any of his eleven gated assets. Open rate around 22%. Click rate around 1.4%. Pipeline created in the last quarter: $0.
He kept asking me what was broken with the copy. The copy was fine. The premise was broken.
Most B2B lead nurturing programs are built like a treadmill. Someone downloads a thing, you drop them into a queue, the queue ships content on a fixed cadence regardless of who they are or what they care about, and 90 days later 80% of the list is dead and the other 20% have unsubscribed. The "drip" model came out of email marketing software in 2011. Buyer behavior moved on. The drips did not.
What works now is signal-based. The nurture program reads what the contact does after the first touch, scores the behavior, and sends a message tied to that signal. No signal, no message. Lots of signal, fast follow-up. The point is to identify the 8 to 12% of MQLs who are actually moving toward a buying decision and get them to a meeting before a competitor does.
This is the playbook I rebuilt for that founder and for four other teams in the last 18 months. It is the same shape every time.
Why drip nurture stopped working
The B2B buying journey moved from linear to non-linear sometime around 2019. Gartner has been tracking this for years and their numbers are worth knowing. The average B2B buying group is now 6 to 10 people. Buyers spend only 17% of the total purchase time meeting with potential suppliers. They spend 27% researching independently online, 18% researching offline, and the rest internally.
Translation: by the time a contact downloads your ebook, they have already done most of their research. They are not sitting in their inbox waiting for your day-14 follow-up email about "why our platform is different". They are on G2 reading reviews, on LinkedIn asking peers, and on three competitor websites running parallel comparisons.
A six-email drip ignores all of this. It treats every MQL as a blank slate who needs to be educated from scratch on a 30-day timeline. The contact who downloaded your ROI calculator last week and just visited the pricing page three times today gets the same email as the contact who grabbed one whitepaper in 2024 and hasn't opened anything since.
That is the core problem. The nurture is built around your content calendar, not their buying timeline.
The 73% number is from the SiriusDecisions waterfall research, replicated by Forrester. It has not improved since 2015. We keep building bigger lead funnels and a smaller fraction of them ever turns into pipeline.
The signal-based model
The model I use has four layers. Each one does one job. None of them are time-based drips.
Layer one: behavioral scoring
Every action a contact takes gets a score. Page view of a comparison page is worth more than a page view of a blog post. A pricing page visit is worth a lot more than an ebook download. A second pricing page visit within seven days is worth even more.
The trick is to score on intensity and recency, not just count. A contact who fired 14 signals in the last week is hotter than a contact who fired 30 signals over six months. Decay matters. Most scoring models I see in HubSpot do not decay. A score of 85 from 2024 still says "send to sales" in 2026. That is how reps end up on cold calls to people who barely remember the company.
A simple decay rule: any signal older than 30 days loses 50% of its weight. Older than 60 days, drop another 50%. Older than 90, it is dead unless they fire a new signal.
Layer two: content matched to stage
I split content into three buckets, not seven. Anything more granular gets impossible to maintain.
- Awareness: blog posts, frameworks, "how to" guides. The contact does not know they have the problem yet.
- Consideration: case studies, comparison guides, ROI calculators. They know they have the problem and are evaluating approaches.
- Decision: pricing pages, customer references, demo videos, free assessments. They are picking a vendor.
Behavioral scoring tells you which bucket the contact is in. A pricing page visit is decision-stage. A "what is account-based marketing" search is awareness. You send content from the matching bucket. Nothing else.
Layer three: trigger-based sequences
These are the actual nurture messages. They fire when a signal trips a threshold, not on a calendar.
Examples I have running for clients:
- Visited pricing page twice in seven days but no demo booked, send a 2-sentence email from the AE with a Calendly link. No marketing footer, no CTA button, plain text.
- Downloaded a comparison guide between us and a competitor, send a 1-page customer story of someone who switched from that competitor, plus an offer for a 20-minute call with the customer.
- Watched 60% of a product demo video, send the deep-dive whitepaper on the specific module that segment cares about, plus a meeting link.
- Opened the last 3 sales emails but never replied, switch to LinkedIn outreach from the AE with a video message.
Each trigger has one job. Get the contact to a person. Not nurture them for another month.
Layer four: handoff and re-cycle
If a contact crosses an SQL threshold (in my models that is usually a score of 65 plus a decision-stage signal in the last 14 days), they go to a rep immediately. The SLA we set is 5 minutes during business hours.
If a contact never crosses the threshold but is still active (any signal in the last 30 days), they stay in the nurture. If they go silent for 60+ days, they get a "re-engagement" sequence: one offer, one personal note, one final touch. After that, out of the active list. Move to a quarterly newsletter only.
This last piece matters more than people think. Most nurture databases are full of corpses. Cleaning them out improves your deliverability, your reporting, and your conversion math.
How to build it in HubSpot
I run most of these in HubSpot because the buyer is usually already on it. The build takes about two weeks if your data is clean. Three to four if it is not.
Step one: clean the data
You cannot score behavior if the behavior data is broken. The most common issues I see:
- Form submissions firing two contacts because of a junk email plus a real one
- Page tracking missing on the pricing page because someone removed the HubSpot script during a redesign
- Email opens inflated 300% from Apple Mail Privacy Protection so the score is meaningless
Audit before you build. Run a data quality audit and fix what you find. If you skip this step the scoring model will be wrong and nobody will trust it.
Step two: build the scoring property
In HubSpot, use the new "engagement score" property type if you are on Marketing Hub Enterprise. If you are on Pro, build it as a calculated property using workflows. Either way, the rules look like this:
- Page view, generic: +1, decay over 30 days
- Page view, comparison/pricing/customer story: +5, decay over 30 days
- Form submit, top of funnel: +10, decay over 60 days
- Form submit, mid funnel (case study, ROI calc): +20, decay over 60 days
- Form submit, bottom of funnel (demo, contact us): instant SQL, no decay needed
- Email engagement, click: +3 (do not score opens because of Apple MPP)
- Video, 25-60% watched: +8
- Video, 60%+ watched: +15
- LinkedIn ad click (via the LinkedIn integration): +4
- Sales email reply: +25
Tune these weights to your business. The exact numbers matter less than the relative weights. A demo request should always score 10x what an ebook download scores.
Step three: build the triggers
Each trigger is a HubSpot workflow with a behavioral enrollment criteria, not a time-based one. The workflow shape:
- Enroll when [signal X] AND [score above Y] AND [no recent rep activity]
- Send the right asset
- Wait 3 days, check for reply or meeting booked
- If yes, suppress further nurture
- If no, move to next trigger or back to scoring pool
This part is the most fragile. Workflows in HubSpot can interact in ways nobody intended. I always build with re-enrollment off by default and test each trigger on a small slice before turning it on for the full database.
Step four: integrate with the rep workflow
The hand-off has to be instant. We set up a workflow that fires when a contact hits SQL threshold: it creates a task in HubSpot for the owning AE, posts a Slack alert to the rep's DM, and starts a 30-minute SLA timer. If the rep does not action it in 30 minutes, escalate to the manager.
This is where most lead nurture programs fail in execution. The marketing team builds beautiful nurtures, the SQL crosses the threshold, sales gets a notification 24 hours later when they finally look at the lead view, and the contact has already booked with a competitor. Sub-5-minute response on inbound is one of the highest-impact fixes in B2B. The classic InsideSales.com / MIT study found a 100x drop in qualification rates between 5 minutes and 30 minutes.
If your SLA is over 5 minutes, your nurture is doing the work and your sales team is killing it.
The single biggest leak in B2B inbound is the gap between SQL trigger and rep response. Fix the gap before you fix the copy.
Where most nurtures break
After auditing maybe 40 of these in the last few years, the same five failures keep showing up.
The list is dead
Half the contacts have not opened anything in 12 months. Sending more email to a dead list trains Gmail to throw your domain in spam. Before any new nurture goes live, suppress anyone with zero engagement in the last 180 days. Yes, this will cut your list. Yes, your deliverability will get better in two weeks.
Scoring has no decay
A static score is a lie. A contact who scored 80 in 2024 and went silent is not warm anymore. Without decay, your nurture treats them as warm forever and your reps stop trusting the score within a quarter.
The "nurture" is product marketing in disguise
The email talks about your features, your awards, your release notes. Nobody in nurture cares yet. The job of nurture is to teach, not to pitch. The pitch comes after the contact talks to a person. If the email reads like a one-pager, it does not belong in nurture.
No segmentation by ICP fit
A bad-fit contact does not get better with more email. They get more annoyed. Before any nurture sends, the contact has to clear an ICP filter. Company size, industry, geo, role. Anyone who fails the filter goes to a newsletter, not a nurture sequence. This alone usually doubles reply rates on the remaining list.
The rep does not know what the contact has read
This is the easiest one to fix and the most often missed. When a lead becomes an SQL, the rep gets a notification with a meeting link and... that is it. They do not know the contact has read three competitor comparison guides and the GDPR whitepaper. They run a generic discovery call and the buyer rolls their eyes.
Fix: include a "content history" summary in the SQL hand-off notification. Last 5 pages viewed, last 3 assets downloaded, last comparison guide read. The rep walks in with context. The discovery call lands 2x better.
Numbers from real rebuilds
The founder I mentioned in the opening was running a $14K MRR program with 8,000 MQLs in his database. After we rebuilt his nurture on this model:
- Active list dropped from 8,000 to 2,100 (suppressed dead contacts and out-of-ICP)
- MQL to SQL conversion went from 0.3% to 2.4% over the next 90 days
- Pipeline created in Q2 of last year: $412K, up from $0 in Q1
- SQL response time went from "the next day" to a median of 3 minutes
- One contact, sourced through a competitor comparison trigger, closed for $86K ACV nine weeks later
The rebuild took 14 working days. Most of that was scoring data cleanup and rewriting the trigger copy.
I have run versions of this for a Series A fintech, a 30-person dev tools company, and a B2B services agency. The shape is identical. The thresholds vary by ACV and sales cycle.
Average MQL to SQL conversion improvement after switching from a calendar-based drip to a signal-based nurture, across five rebuilds since 2024.
What you do not need
A few things I see teams over-invest in that do not move the number.
A CDP
If you have HubSpot or Salesforce plus a clean website tracking setup, you do not need Segment, mParticle, or RudderStack to run nurture. Save the $80K and spend it on better content. CDPs become useful when you have 6+ data sources you actually need to unify. Most Series A teams do not.
AI-written nurture emails
Tempting, expensive, and almost always worse than what the founder or AE would write themselves. Use AI to research the account and personalize a hook. Do not use it to generate the body of a 4-paragraph email. The "AI sheen" kills reply rates faster than bad copy ever will. Our AI SDR breakdown goes into where AI actually helps in outbound, and where it just makes you look lazy.
Predictive lead scoring out of the box
Vendors sell this hard. The "predictive" model is a black box trained on industry-generic data. For Series A and B teams, a hand-built scoring model based on your last 200 closed deals will outperform a vendor's predictive model 9 times out of 10. Build your own. Tune it every quarter.
A 12-month nurture map
If your nurture map has 47 nodes and a 12-month horizon, you have built a content calendar, not a nurture. Real nurture programs have 8 to 15 triggers and run in parallel, not in sequence. Map them as a hub of triggers, not a linear journey.
Tools that actually help
- HubSpot Marketing Hub Pro or Enterprise for the scoring, workflows, and reporting. Enterprise gets you the new engagement score property which is genuinely useful.
- Clay for the enrichment that powers ICP scoring. Match a contact to a company, get firmographic data, score against ICP rules.
- n8n for the integrations HubSpot does not natively support. We run the Slack-DM hand-off through n8n because the native HubSpot to Slack integration is clunky.
- LinkedIn Sales Navigator with the HubSpot integration, so a LinkedIn ad click counts as a real signal. The native integration is fine since the 2024 release.
That is the whole stack for a Series A nurture program. Under $3,500 a month all-in.
A note on B2B vs B2C nurture
If you read most lead nurturing content online, it is written for ecommerce. Abandoned cart sequences, post-purchase upsell, win-back campaigns. None of that maps cleanly to B2B. Our buying group is 6 to 10 people, our deal sizes are 5 to 6 figures, and the buying timeline is 45 to 180 days.
What B2B borrows from B2C: behavioral triggering, signal decay, content matching to stage.
What B2B has to do differently: account-level scoring (not just contact-level), buying group identification, longer trigger horizons, much heavier sales involvement in the nurture itself.
If you want the deeper version of how this maps to account-based motions, that is its own post.
Need to rebuild your nurture?
Book a free 30-minute audit. We will pull your last 90 days of MQL data and show you the three triggers that would have created the most pipeline.
Book an audit →FAQ
How long should a B2B lead nurture be?
There is no fixed length. The nurture runs as long as the contact keeps firing signals. A hot contact might convert in 4 days. A slow one might sit in nurture for 9 months before they show buying intent. Stop thinking in weeks and start thinking in signals.
What is a good MQL to SQL conversion rate?
For Series A and B B2B SaaS, 8 to 14% is the realistic target. Below 5% and your scoring model is broken or your MQL definition is too loose. Above 20% and you are calling people MQLs who are actually SQLs already. Both are common failure modes.
Should I run nurture on contacts or accounts?
Both. For SMB and mid-market, contact-level nurture is enough. For accounts with 6+ buying-group members or 6-figure ACV, run account-level signal aggregation on top. If three different contacts at the same company touched your pricing page in 14 days, that account is hot even if no single contact crossed a threshold.
Can I use AI to write nurture emails?
Not for the email body. Use AI for research and personalization hooks. Write the body yourself or have your AE write it. AI-written nurture emails read like AI-written nurture emails, and reply rates show it.
How often should I audit and tune the scoring model?
Quarterly. Pull the last 50 closed-won deals, look at what signals they fired in the 90 days before the deal, and adjust the weights. Anything that consistently shows up in winners but not in non-winners should be weighted higher. Anything that shows up everywhere is useless as a signal and should be dropped.
If you want a hand running the audit or rebuilding the program from scratch, we do this for a living. The first call is free and we will tell you in 30 minutes whether the program is worth rebuilding or whether the bigger problem is somewhere else entirely.