The B2B SaaS Demand Generation Playbook

Most demand generation programs fail in the same way. They are assembled channel by channel, judged on lead volume, and quietly disconnected from the revenue number the company is actually chasing. This playbook is the sequence we use to build programs that survive contact with a board meeting.

It is written for B2B SaaS companies between roughly 2 million and 50 million in ARR. Below that the answer is usually founder led sales plus a narrow capture program; above it the constraint becomes organisational rather than tactical.

What is demand generation?

The distinction that matters operationally is between creating demand and capturing it. Demand creation makes people want the thing. Demand capture collects the ones who already do. Both are necessary. Funding only capture means you compete for a fixed pool of in market buyers, and your cost per acquisition rises every quarter until the model breaks.

The 2026 numbers this playbook is calibrated to

Two benchmarks changed enough in the last year to invalidate plans built on 2025 assumptions. Read these before setting any channel target.

Together these say the same thing from two directions: fewer of the opportunities you create will close, and more of the education happens before a buyer ever reaches your site. A program that only counts form fills will read both changes as failure.

Step 1: Define the ICP narrowly enough to hurt

Most ICP definitions are too broad to make decisions with. A usable one names the firmographics, the trigger event, the buying committee, and the alternative you displace. If your ICP does not exclude a large share of the market, it is not doing any work.

• Company profile: size, stage, geography, tech stack, industry.
• Trigger: the event that turns this from a someday problem into a this quarter problem.
• Committee: who feels the pain, who evaluates, who signs, who blocks.
• Status quo: the spreadsheet, the internal tool, or the incumbent you replace.
• Disqualifiers: the segments you will deliberately not chase.

Step 2: Build the model backwards from pipeline

Start at the revenue target and divide until you reach a monthly spend. Every number in the chain is an assumption you will correct with real data, and having them written down is what makes the correction possible.

1. 1

   Start with new ARR target

   Take the number marketing is accountable for, not the whole company target.

2. 2

   Divide by average contract value

   This gives the number of closed won deals required.

3. 3

   Divide by win rate

   This gives the qualified opportunities required.

4. 4

   Divide by opportunity conversion rate

   This gives the qualified meetings or SQLs required.

5. 5

   Apply pipeline coverage

   Multiply by three to four to account for slippage and timing.

6. 6

   Divide by expected cost per opportunity

   This gives the budget. If it exceeds what you have, the constraint is in win rate or ACV, not in ad spend.

Step 3: Allocate across the channel mix

For a company under roughly 50 million in revenue, a 60/30/10 split is a defensible starting point: 60 percent to capture, 30 percent to creation, 10 percent to structured experiments. Shift toward creation as capture saturates, which you will detect as rising cost per opportunity at flat volume.

Budget splits by company stage

Channel economics, worked

Reddit and community: the channel most B2B teams get wrong

We run this for clients, so the caveats are firsthand. Community works because buyers trust peers and because forum threads are retrieved constantly by AI answer engines, which multiplies the value of a single good thread. It fails when it is treated as a paid placement.

Step 4: Fix the conversion path before scaling spend

Doubling traffic into a broken path doubles the waste. Before any budget increase, audit the sequence a buyer actually walks.

1. Does the landing page make the same promise as the ad that sent them there?
2. Is the primary call to action the one this buyer stage is ready for?
3. How many fields are on the form, and is each one earning its place?
4. What happens in the first five minutes after submission?
5. Is speed to lead measured in minutes or in days?
6. Are unqualified submissions being routed away from sales rather than into it?

Step 5: Treat creative as the main variable

In paid channels, targeting options have largely commoditized. What separates a program that scales from one that plateaus is the volume and quality of distinct creative concepts. Not variations on a headline, genuinely different angles: the problem angle, the competitor comparison, the customer proof, the contrarian point of view, the product demonstration.

• Ship several genuinely distinct concepts per month, not twenty colour variants of one.
• Kill on statistical evidence, not on internal taste.
• Reuse winners across channels before inventing new ones.
• Feed the sales team's actual objection language into the copy.

Targeting is commoditised. Creative volume is the only paid lever left that competitors cannot buy their way past.

Step 6: Report on pipeline, not leads

The lead count is the number most likely to be growing while the business is not. Report the chain.

The first 90 days

If you are inheriting or rebuilding a program, this is the order that produces the fastest defensible result. Nothing here requires a budget increase.

1. 1

   Days 1 to 14: establish truth

   Pull closed won and closed lost by segment, rebuild the pipeline model with real conversion rates, and audit the conversion path end to end. Do not change spend yet.

2. 2

   Days 15 to 30: stop the leaks

   Fix routing, form length, message match, and speed to lead. Cut the single worst performing channel and hold the budget rather than reallocating it immediately.

3. 3

   Days 31 to 60: rebuild capture

   Rework search structure and landing pages, complete review platform profiles, and restructure the ten highest intent organic pages. Add self reported attribution to the demo form.

4. 4

   Days 61 to 90: restart creation

   Launch a creative program with three distinct concepts, begin community participation, and set the reporting cadence. Present the model with three scenarios.

Step 7: Run a cadence that produces decisions

1. 1

   Weekly

   Spend pacing, creative performance, anomalies. Fifteen minutes, no slides.

2. 2

   Monthly

   Pipeline by source against plan, cost per opportunity trend, reallocation decisions.

3. 3

   Quarterly

   Channel mix review, ICP validation against closed won data, experiment budget reset.

The programs that compound are boring in the best way: a narrow ICP, a model everyone agrees with, a conversion path that works, a steady supply of new creative, and a report that shows pipeline. Everything else is variation on top of that foundation.