B2B SaaS demand-gen playbook from 4 years running an agency

Between 2017 and 2021 I ran Marketers — a B2B digital marketing agency based in Tel Aviv. We started as me, alone, after the army. We finished as a team of around 20, with more than 100 clients spanning fintech, B2B SaaS, e-commerce, hospitality, and a long tail of weirder things I won't bore you with.

I sold or wound down most of those engagements at the end of 2021 to focus on operating roles. But the playbook from those four years — what works in B2B SaaS demand-gen, what's a complete waste of money, what I'd do differently if I started over today — that's still the foundation of how I look at growth now. This piece is that playbook, written the way I'd hand it to someone starting an agency or a B2B SaaS growth team in 2026.

The single most important framing

Most B2B SaaS founders ask the wrong question first. They ask, "what marketing channel should we use?" The right question is, "what's the shortest path from a stranger noticing us to a stranger paying us?"

That path has stages — awareness, interest, consideration, decision, purchase — and almost every demand-gen mistake comes from optimising one stage in isolation. You add awareness traffic that doesn't convert. You bury yourself in MQLs that don't reach SQL. You spend on ads that drive trial signups your sales team can't close.

The playbook below is structured by stage, but the mental model is: fix the bottleneck, then the next bottleneck, then the next. Don't try to optimise five stages at once. You'll dilute every effort.

Stage 1 — Awareness: get noticed by the right strangers

For B2B SaaS specifically, three awareness channels matter in 2026:

1. Content + organic search. Still the highest LTV channel. Slow to start, compounds for years. Mandatory for any B2B product where the buyer is researching.
2. LinkedIn. The single best targeted-paid platform for B2B in the world. The cost-per-click is high; the cost-per-customer is often the lowest of any channel if your ICP is professionals.
3. Founder thought leadership. A founder writing in public — on LinkedIn, on a blog, on Twitter/X — is worth more than a $20K/month content team for the first two years. Buyers in 2026 buy from people they recognise.

What we used to recommend that I no longer do:

• Cold email. Conversion rates have collapsed. Inbox protection has gotten too good. The few campaigns that still work need extreme personalisation that doesn't scale.
• Display advertising for awareness. Banner blindness is total. Save your budget.
• "Brand video" production. Useful at $50M+ revenue, money-burner before that.

Stage 2 — Interest: turn passive readers into engaged readers

This is where most teams under-invest. They build awareness traffic, then send everyone to a homepage with a single "Book a demo" button. The interested but not-yet-ready visitor has nowhere to go.

The fix is having middle-of-funnel content people can engage with without committing to a sales call:

• Comparison pages (X vs Y, X vs Z) — these have the highest commercial intent of any organic content type.
• Templates, calculators, tools — anything that does one job well and gets bookmarked.
• Substantial guides on the buyer's actual problem (not "what is SaaS?", real specific problems).
• A clear newsletter people actually want to read.

The metric that matters at this stage is not signups. It's return visits. If you have a million sessions a month and 90% of them are first-time, your awareness is great and your interest layer is broken.

Stage 3 — Consideration: make it cheap to engage with sales

Most B2B sales motions in 2026 are still over-engineered. Three calls, qualification, discovery, demo, follow-up, proposal. The buyer is comparing you to four other vendors who collectively want twelve calls.

The agencies and SaaS teams that won in our portfolio were the ones that compressed this:

• Self-serve product trial wherever possible. Even for "enterprise" products, give buyers a way to look without booking.
• Async demo videos before the sales call. 8-minute Loom of the product solving a specific problem. Send before the call. The call becomes shorter, sharper, more closeable.
• Public pricing. The "request a quote" wall costs you 30% of qualified buyers. Yes, even at six- and seven-figure deals.
• One sales call to close, not three. Do the qualifying async; do the closing in person.

Founders push back on these because their existing sales motion has revenue attached. That's fair. But run a controlled test on 10% of incoming leads with a compressed motion and compare conversion. We did this at half a dozen client SaaS companies. The compressed motion won every time, except in regulated industries.

Stage 4 — Decision: answer the buyer's last objection before they ask

The deals that stall in late-stage rarely stall on price. They stall on a specific worry — security, integration, support quality, whether the product still exists in two years. The conversion lift comes from identifying the worry and addressing it on the product page, not in the sales call.

Things that demonstrably moved late-stage conversion in our agency work:

• A security page with the actual SOC 2 report linked. Not just "we're SOC 2." The real document.
• A status page with multi-month uptime history visible.
• Logos of similar customers, with linkable case studies, not just a wall of branded squares.
• Testimonials from people whose names you can verify on LinkedIn, not "Sarah J., Director."
• An honest "who this isn't for" section. Removes 30% of bad-fit buyers and increases conversion of good-fit buyers.

Stage 5 — Purchase and beyond: the cheapest customer is the one you already have

Demand-gen people obsess over acquisition because that's their job. But the most growable line in any B2B SaaS P&L is expansion revenue, and the playbook for it is different.

• Make onboarding actually solve the buyer's problem in week one. Most B2B SaaS onboarding is product tours when it should be guided "first-success" sessions.
• Track activation, not just signup. Activation = the user got real value at least once. Most products have this metric undefined.
• Have a customer-success motion that's not a glorified support queue. CS should be aware of expansion triggers and routing them to AEs.
• Build a referral mechanism that doesn't feel sleazy. The simplest one: ask for a quote when a customer says something nice, with permission to publish it.

The four meta-lessons from 100+ engagements

1. Channel matters less than message

Two clients in similar verticals, same channel mix, same budget, sometimes within the same quarter. One would 3x revenue, one would barely move. The variable was almost always positioning. The ones with sharp, specific messaging that named the buyer and the pain point won. The ones with vague "platform for [verb]ing" lost.

Spend the first month of any new engagement on positioning. Don't run a single ad until the message is sharp.

2. Attribution is mostly a lie, but track it anyway

Anyone who tells you they have full-funnel B2B attribution working is either selling you software or self-deceiving. The buying journey for B2B is dark, multi-touch, and partially happens in conversations between humans you'll never see.

That said: track what you can. Self-reported attribution on intake forms ("how did you hear about us?") is more honest than UTM-based attribution. Combine both.

3. Don't outsource your story

The best content we ever produced for clients came from interviews with their founders. The worst came from briefs handed to us by their marketing manager. If a founder won't sit for a 60-minute interview every six weeks, the content engine will be mediocre.

This is why "founder-led" content beats agency-managed content for the first three years. Agency-managed content doesn't fail because agencies are bad. It fails because the founder's actual voice is the most differentiated thing the company has, and outsourcing it is throwing it away.

4. Ship in public

The clients that built marketing audiences fastest were the ones who shipped product changes in public. New feature → blog post + LinkedIn + email + customer call-out, every week. Not "case studies" published quarterly. Constant, public, real shipping.

This requires the product team to ship, which is usually the gating constraint.

The honest bit: what I'd do differently if I started a B2B agency in 2026

If I were standing up Marketers again today, I would:

1. Niche harder, faster. We were too generalist for too long. The agencies that survive the next three years are the ones with one ICP, one playbook, one repeatable motion.
2. Charge for outcomes, not hours. Hourly billing is dying everywhere. The agency model that works in 2026 is fixed-price-for-outcome or revenue-share.
3. Build a content engine for ourselves before any client. The agencies with no public presence have no leverage. We were the cobbler's barefoot children for years.
4. Lean fully into AEO over SEO. The next four years of demand-gen are about being cited in AI answers, and most agencies are still selling the 2018 SEO playbook.
5. Refuse founders who can't write. If the founder can't or won't be the voice, the engagement will fail. Politely decline.

That's the playbook. None of it is novel — it's a compression of four years of real engagements and the patterns I saw repeat. Most B2B SaaS demand-gen problems are solvable by removing things, not adding things.