Demand generation framework

The AI-native B2B demand generation framework

Six phases, from the right customer to measured pipeline. For each one: the job to do, where AI fits into the process, how to measure it, and resources you can use today.

Federico Anedda · Updated October 2026 · 10-minute read

In shortA demand generation framework is the order in which you do the work: choose who to sell to, decide which accounts come first, create and capture demand, turn it into pipeline with sales, measure what worked, and feed what you learn back into the start. This one asks one more question in every phase: which part of the job can AI do, and where does a person still decide?

Demand generation is not lead generation

Lead generation collects contact details. Demand generation gets you onto the shortlist before the buyer talks to anyone, and then into the pipeline.

	Demand generation	Lead generation
Goal	Be on the shortlist of accounts that aren't buying yet, and be chosen when they are	Collect contact details from people who show interest
Who it reaches	The buying group in every account of your market	People ready to fill in a form
Main measure	Pipeline, cost per opportunity, win rate by segment	Leads, MQLs, cost per lead
Time horizon	Quarters	Weeks
Common mistake	Judged on leads in the first month	Gated content for people who aren't buying yet

In this framework, lead generation is one part of capturing demand, inside phase 3.

Why you need a framework

Without one, each team fixes its own part and the mistakes add up. These are the five most common, and the phase that prevents each one.

Phase 1The ICP is guesswork, and the message describes the product the way the company sees it.
Phase 2Marketing builds the account list alone, and sales never signs it off.
Phase 3All the budget goes to capturing demand that already exists.
Phase 4Leads reach sales with no context, and nobody records why they were rejected.
Phase 5One attribution model is presented as the truth, read by contact instead of by account.

The framework at a glance

Six phases in a row. What you learn in phase 6 goes back to phase 1. Click a phase to read it.

1ICP and positioningWho the right customer is, and what we tell themFinds what your best customers share2Segmentation and accountsWhich companies, in what orderEnriches and researches accounts (Clay)3Demand generation engineChannels, content, SEO and GEO, budgetTests whether AI answers name you (GEO)4Pipeline conversionWhere marketing meets salesGives sales the account's context5Measuring demand genWhich campaigns produce pipelineWrites the campaign review6Measuring GTMThe result of the whole systemFlags anomalies and deals at risk

Marketing

Marketing and sales

Marketing ops, RevOps, leadership

Under every phaseData and technology: HubSpot, Salesforce, Clay, n8n, ClaudePeople and roles: one owner per phase

Full page online

The six phases

Each phase has a job, a few things to do, one place where AI changes the work, and a number that tells you if it works.

ICP and positioning

Who is the right customer, and what do we tell them?

The problem

The ICP is based on guesswork and the message is generic: buyers don't describe the offer the way the company does.

Do this

Find what your best customers have in common, from your own data
Write down the buying triggers and the buying group
Write the message in the words buyers use

Where AI fits in

Finds what the best customers have in common and the words they use; checks whether the message lands.

Generative AI

Measure

Win rate and deal size of accounts that match the ICP, compared with those that don't.

Who owns it

Marketing and sales, with leadership

TopicsICPPositioning and messagingPipeline goals and market sizeFull page coming soon

Segmentation and accounts

Which companies, and in what order?

The problem

Marketing chooses the account list alone, and nobody in sales has signed it off.

Do this

Build the list of accounts that fit the ICP
Split it into tiers, with sales
Map the buying group in each tier 1 account

Where AI fits in

A Clay workflow enriches every account, an AI column researches its website and news, and a score puts it in a tier; AI also maps the buying group.

Generative AI + rules

Measure

Share of pipeline from tier 1 and 2 accounts, on a list signed off by sales.

Who owns it

Demand generation with sales; RevOps for the data

TopicsAccount list and tieringBuying groupsIntent data

Demand generation engine

Which channels, which content, and how much budget?

The problem

Teams optimise for the number of leads and ask people who aren't buying yet to convert.

Do this

Split the budget between creating and capturing demand
Be found on Google and named in AI answers (SEO and GEO)
Fund the channels that produce pipeline, not leads

Where AI fits in

Pulls themes from the market, tests whether AI assistants name you when buyers ask for suppliers (GEO), and optimises ads and bids for opportunities.

Generative AI

Measure

Pipeline and cost per opportunity by channel; share of the buyer's questions where search and AI answers name you.

Who owns it

Demand generation

TopicsSEO and GEOContent and thought leadershipLinkedIn AdsEvents, webinars and communityPaid searchLanding pages and demo requestsABM playsPlan and budget by channelPartner marketingRead the phase →

ResourcesAI shortlist test (GEO) · SEO and GEO page checklist Go to resources →

Pipeline conversion

How does marketing effort become a sales opportunity?

The problem

Marketing generates leads, sales opens opportunities, and nobody can connect the two.

Do this

Score fit and intent on separate axes
Agree an SLA: every lead gets a decision
Use a closed list of rejection reasons

Where AI fits in

Summarises the account's context for the salesperson and classifies rejection reasons.

Generative AI

Measure

Share of leads with a decision within the SLA; lead-to-opportunity rate by source.

Who owns it

Demand generation, sales and RevOps

TopicsLead and account scoringSLA and handover to salesNurture and lead recyclingFull page coming soon

Measuring demand gen

Which campaigns produce pipeline?

The problem

The pipeline is there, but nobody knows which campaign produced it.

Do this

Set the rules before you measure
Check data quality before every report
Read sourced and influenced pipeline separately, by account

Where AI fits in

Writes the campaign performance review and suggests tests.

Generative AI

Measure

Share of opportunities with a tracked source; sourced pipeline per euro spent.

Who owns it

Marketing ops and RevOps

TopicsAttribution and campaign reviewTesting and optimisationRead the phase →

ResourcesAttribution simulator · Campaign measurement · Campaign performance review · Data quality before the report Go to resources →

Measuring GTM

Is the whole system working?

The problem

Each team measures its own activity, not the result of GTM as a whole.

Do this

Read win rate, sales cycle and payback by segment
Compare segments, not teams
Feed what you learn back into the ICP

Where AI fits in

Explains the funnel numbers and flags anomalies and opportunities at risk.

Predictive model

Measure

Pipeline coverage, win rate, sales cycle and payback by segment.

Who owns it

Leadership and RevOps

TopicsGTM metrics

Where AI fits in, and where it doesn't

AI does one job in each phase. A person still makes the decision, and every job has a limit.

Phase	What AI does	Type	A person still	The limit
1 ICP and positioning	Finds what your best customers share	Generative AI	Chooses the ICP	Only as good as your customer data
2 Segmentation and accounts	Enriches and researches accounts (Clay)	Generative AI + rules	Signs off the tiers with sales	Data coverage is lower in Europe; GDPR applies to people data
3 Demand generation engine	Tests whether AI answers name you (GEO)	Generative AI	Decides what to publish	AI answers change from one run to the next
4 Pipeline conversion	Gives sales the account's context	Generative AI	Accepts or rejects the lead	Needs a closed list of reasons
5 Measuring demand gen	Writes the campaign review	Generative AI	Checks the hypotheses	Hypotheses, not causes
6 Measuring GTM	Flags anomalies and deals at risk	Predictive model	Decides what to change	Needs enough closed deals

Where to start

You don't need all six phases at once. Start where it hurts most.

If you can't say which campaigns produce pipelineStart with phase 5 → If sales rejects most of the leads you sendStart with phase 4 → If buyers don't find you on Google or in AI answersStart with SEO and GEO → If marketing and sales target different accountsStart with phase 2 →

Frequently asked questions

What is a demand generation framework?

The order in which you do demand generation work, from choosing the right customers to measuring pipeline, so that each step feeds the next.

How is demand generation different from lead generation?

Lead generation collects contact details. Demand generation works on the whole market, including accounts that aren't buying yet, and is measured on pipeline.

What are the stages of demand generation?

In this framework: ICP and positioning, segmentation and accounts, the demand generation engine, pipeline conversion, measuring demand gen and measuring GTM, which feeds back into the ICP.

How do you measure demand generation?

On pipeline, not leads: sourced and influenced pipeline, cost per opportunity and win rate by segment, read by account.

Where does AI help in demand generation, and where doesn't it?

It helps with research, enrichment, summaries and first drafts. It doesn't choose your ICP, sign off the account list or prove what caused a result.

Building AI into your demand gen?

Share what works for you, or ask about a resource.

Message me on LinkedIn ↗