B2B CRM: How to Structure an Email Funnel That Holds Up in 2026

Most B2B email funnels I inherit in audits share the same flaw: they were designed as a sequence of messages, not as a data system. I show up to seven emails already written, scoring cobbled together from opens, and a list imported from a CSV export nobody can trace back to its source. It runs for three months, then placement rates degrade, sales complains about lead quality, and someone kicks off a "nurturing overhaul" project.

A B2B email funnel that holds up rests on four stacked layers, in this order: the legal basis for collection, the data model (person and account), the sending infrastructure, and only then the message sequence. Reverse that order and you are building the floor before the foundation.

Here is the method I apply in agency work to lay down these four layers.

Layer 1: the legal basis, to settle before writing the first email

In B2C, the question is simple: consent. In French B2B, it is different, and a lot of teams get it wrong.

The CNIL states that email prospecting toward professionals can rely on legitimate interest, provided the purpose of the solicitation relates to the profession of the person being contacted. That is a change of legal basis, not a blanket exemption from the rules. The nuance has direct consequences on how you structure your segments.

Three operational points to build into your CRM from the design stage:

• The link to the profession decides everything. If you sell invoicing software and your sequence targets HR profiles, your legal basis no longer holds. That means your role-based segmentation is not a marketing refinement, it is a compliance condition. A reliable, normalized "job function" field becomes a legal asset, not just a personalization criterion.
• Identification and easy refusal, in every message. The CNIL specifies that every solicitation must let the recipient identify the sending organization and express refusal through a simple means. No exception for automated "one-to-one" emails disguised as manual messages.
• Generic addresses do not follow the same regime. Addresses like info@, contact@ or commande@ belong to the legal entity and are not subject to the same principles as named professional addresses. Concretely for your data model: these addresses deserve their own segment, handled separately from named addresses.

One last point, the one that costs the most in audits: if your data was collected indirectly or purchased, the person must have been informed that their address could be used for prospecting. A purchased list with no trace of that information does not get sent to. You set it aside, you do not "test on a small volume to see."

A practice I enforce on every B2B setup: a source_collecte property and a date_collecte property on every profile, filled in at entry, never after the fact. Without those two fields, you can neither document your legal basis nor clean your list intelligently. The same logic applies to your on-site forms, as detailed in our article on zero-party data.

Layer 2: model the account, not just the contact

This is the main break between a B2C funnel and a B2B funnel. In DTC ecommerce, the person is the decision unit. In B2B, the decision is made at the account level, with three to six people involved, different roles, and signals that belong to the company rather than to the individual.

Most ecommerce-oriented tools model the world as profiles and lists. You can store a company name in a custom property, but you then duplicate that information across every member of the account, with all the desync risk that implies.

Customer.io models this case with objects: an object is "an individual thing (a single company, online class, or account) that you want to relate to one or more people." The documentation is direct about the operational benefit: without objects, you would need to set attributes or send an event for every individual member of the account. You trigger a message at the account level, and the attached people follow.

What that changes when you design your funnel:

The open documentation does not specify the technical limits (number of object types per workspace, number of attributes, object-to-object relationships). If your model is dense, check these caps with the vendor before committing to an architecture.

If your stack is already on Klaviyo for other reasons, you can approach the account with well-structured custom properties, accepting duplication and managing sync from your source of truth. That is not the same thing as a native object model, but it holds up at limited account volumes. The detailed comparison of the two logics is in Klaviyo vs Customer.io.

Three entities, minimum, in your schema

Before writing a single email, put on paper:

1. The person: email, job function, collection source and date, marketing subscription status, engagement.
2. The account: sector, size, commercial lifecycle stage, sales owner, contract status.
3. The events: what the person does, what the account produces as an aggregated signal. One clean event beats ten static attributes that age badly.

The link between the three is what lets you score later without hacking something together. Our email engagement score method is built exactly on this structure.

Layer 3: the sending infrastructure, before volume

A B2B funnel mixes, by construction, two natures of messages: marketing nurturing and product or transactional notifications. This is where most setups sabotage themselves.

Yahoo explicitly asks senders not to send marketing email from the same IPs as user mail, transactional mail, or alerts. For a B2B funnel, the consequence is structural: your nurturing sequence and your product notifications do not share the same sending path. Separate IPs, and ideally separate subdomains, each with their own reputation.

The thresholds to know before scaling volume

Gmail sets the bar at more than 5,000 messages a day to Gmail accounts to move a sender into the bulk sender regime, applicable since February 1, 2024. One detail traps B2B teams trying to compartmentalize by subdomain: Google specifies that messages sent from the same primary domain count toward the same 5,000 limit, subdomains included. Segmenting your B2B sends onto news.yourdomain.com does not reset the counter.

What Gmail requires above the threshold:

• SPF and DKIM on the domain, plus a DMARC policy on the sending domain. Below the threshold, SPF or DKIM alone is enough.
• Spam rate under 0.30% in Postmaster Tools. Google recommends aiming under 0.10% and never reaching 0.30% or above.
• One-click unsubscribe, via the List-Unsubscribe-Post: List-Unsubscribe=One-Click and List-Unsubscribe headers, without waiving a clearly visible unsubscribe link in the message body.

On Yahoo's side, the complaint rate must stay under 0.3%, with SPF, DKIM and a valid DMARC policy at minimum p=none, which must pass.

One thing to flag about these thresholds: they are complaint thresholds across all segments, not B2B benchmarks. None of the open pages publish a B2B-specific complaint, open or reply rate. Treat 0.30% as a ceiling never to approach, not as a target.

DMARC: the three-stage rollout

DMARC rolls out in three modes: none (reporting without enforcement), quarantine, then reject. The DNS record declares the recipients for aggregate reports (rua=) and failure reports (ruf=).

The classic B2B mistake: publish p=none, check the compliance box, and never read the aggregate reports. Those reports are precisely what tells you which services send on your behalf (the sales tool, the webinar platform, the CRM, the ATS) and which ones are not aligned. Skip that work and moving to quarantine breaks flows you did not know existed. Detailed configuration is in our SPF, DKIM, DMARC guide, and the full Gmail and Yahoo rules in Google and Yahoo sender requirements for 2026.

List hygiene: B2B's weak point

Yahoo asks senders to monitor hard bounces, soft bounces and inactive recipients, and to quickly remove invalid recipients. In B2B, this rule bites harder than in B2C: staff turnover kills professional addresses continuously, with no signal to warn you. A contact who changes jobs leaves behind an inbox that bounces, or worse, one that becomes a spam trap.

Two habits to build:

• Automatic, immediate handling of hard bounces, with no exception for "big accounts."
• A fixed-interval review of inactives, with real removal rather than just exclusion from campaigns. Our article on soft bounces and hard bounces covers the decision thresholds.

Layer 4: the sequence, finally

An honest note before the detail: no vendor or normative source defines an official B2B email funnel structure. Number of steps, cadence, delays between emails, MQL or SQL scoring thresholds: these are practice choices, not rules. Be wary of any article that sells you "the 7-email sequence" as a market standard.

What I consider structural, from agency experience:

The trigger outranks the calendar. A sequence that fires at D+1, D+3, D+7 regardless of what the prospect does generates complaints. A sequence wired to events (a download, a pricing page visit, a colleague invited, a usage threshold) stays relevant longer.

Separate entries by intent, not by persona. A contact who downloads a whitepaper and a contact who requests a demo do not belong on the same journey, even with the same job title.

Scoring is built on events, not on opens. With privacy protection on mail clients, the open has become a noisy signal. We cover the impact in Apple Mail Privacy Protection. A click on a product page, a pricing page visit, a reply: those are usable signals.

The handoff to sales must be an event, not an opinion. Define the nurturing exit condition before you launch, write it into the CRM, and accept that sales will push back on it for two months. That is normal. Our complete method is developed in B2B Lead Nurturing.

The double opt-in question

A question that comes up on every B2B setup: should you turn on double opt-in?

At Klaviyo, a contact who signs up via double opt-in without confirming does not join the list: Klaviyo creates a blank profile, with no additional information. Confirmation emails are valid for 72 hours. Two implications for a B2B funnel:

• Your unconfirmed profiles exist in the account but are unusable. If you count your leads by the volume of profiles created, you are counting empty ones.
• Past 72 hours, the confirmation link expires. A B2B prospect who signs up Friday evening and comes back Tuesday will not confirm. Plan a reminder inside the window, or accept the loss.

One thing not to assume: the open documentation states that new lists inherit the account's default opt-in setting, without specifying what that default is. Check the setting in your own account rather than assuming it is active.

The rollout order, in practice

If you are starting a B2B funnel from zero, the order that avoids rework:

1. Week 1: map your collection sources, document the legal basis for each, eliminate anything untraceable.
2. Weeks 1 to 2: put the person / account / events schema on paper before any configuration. Decide where the source of truth lives.
3. Week 2: authentication (SPF, DKIM, DMARC at p=none with rua configured and monitored), separate marketing and transactional sending paths.
4. Week 3: instrument events on the product or site side. Without events, everything else is batch-and-blast in disguise.
5. Weeks 3 to 4: build two or three journeys, not ten. One entry journey per major intent, measured over a full cycle before adding more.
6. Ongoing: monitor complaint rates in Postmaster Tools and the Yahoo Sender Hub, before they drift. The tracking method is in Postmaster Tools and Yahoo Sender Hub.

What fails most often is step 4. Teams build the emails before they have the events, then spend six months personalizing messages that have nothing concrete to say.

If you want an existing architecture audited before rebuilding it, our Klaviyo agency works on this kind of diagnosis for ecommerce and hybrid stacks.

FAQ

Can you prospect by email in B2B without prior consent in France?

The CNIL states that prospecting toward professionals can rely on legitimate interest, provided the purpose of the solicitation relates to the profession of the person contacted. This is not an exemption from the rules: identifying the sender and offering a simple means of refusal remain mandatory on every message.

Do contact@ and info@ addresses follow the same rules?

No. The CNIL specifies that these generic addresses belong to the legal entity and are not subject to the same principles as named professional addresses. That justifies isolating them in a dedicated segment, handled separately.

Does a dedicated subdomain keep you under Gmail's 5,000-message daily threshold?

No. Google specifies that messages sent from the same primary domain count toward the 5,000 limit, subdomains included. Compartmentalizing by subdomain serves reputation and flow separation, not threshold avoidance.

Should nurturing emails and product notifications be separated?

Yes. Yahoo asks senders not to send marketing email from the same IPs as user mail, transactional mail, or alerts. In B2B, where both message types necessarily coexist, that is an architecture decision to make at the start.

How many emails should a B2B nurturing sequence have?

No normative or vendor source sets a standard number of steps, cadence, or delay for a B2B funnel. These choices come down to practice and depend on your sales cycle. Build on event-based triggers rather than a fixed calendar, and measure over a full cycle before adding steps.

Want your B2B funnel architecture reviewed before you rebuild it? Talk to the Deliver team.