How to Generate B2B Leads in 2026

Step 1 — Write down your ICP before you touch a tool

We've audited campaigns from ~40 teams over the last 18 months. The single biggest predictor of a bad reply rate wasn't the copy or the sending domain — it was that nobody could tell us, in one sentence, who they were reaching. If your ICP lives in someone's head, your list is guessing.

Keep it to five filters. Any more and you're describing a fantasy customer, not a market.

• Sector — narrow enough that one email angle works across the list.
• Company size — headcount band. Under 20 and over 500 behave differently; don't mix.
• Geography — matters for language, buying cycles, and compliance (GDPR bites hard in DACH/BeNeLux).
• Role of the buyer — Head of X, not "decision-maker." One title with one budget.
• Trigger — the observable event that makes now a good time to reach out (recent hire, funding, new product, tech change).

Step 2 — Pick one primary channel, one amplifier

Multichannel is the right answer eventually. It is the wrong answer on day one, because you'll end up doing four channels poorly instead of one channel well. Pair one primary — where the first-touch reply happens — with one amplifier — where you build recognition so the primary lands warmer.

Cold email

Still the highest-volume, lowest-cost primary in 2026. Deliverability is now the hard part, not writing. Budget for a warm-up tool, secondary domains, and a monthly deliverability audit. Expect 3-6% reply on a good list, sub-2% if the ICP is loose.

LinkedIn outbound

Slower and lower volume, but much higher acceptance in senior roles (VP+, C-level) and in markets where cold email is oversaturated (US SaaS, UK fintech). Works best as the amplifier to cold email, not the primary — connection request + comment + view profile before the email lands.

Intent-driven outbound

Instead of blasting the ICP, wait for a trigger (job posting, funding round, tech signal, category research) and reach out that week. Lower volume, 3-4x reply rate, and the meetings that book actually close. This is where cycle time collapses.

Step 3 — Model the numbers before you send anything

Before you write copy, work backwards from the pipeline you need. If the math doesn't clear, the copy can't save you.

Step 4 — Write messages a human would actually reply to

The templates you'll find online in 2026 are the same ones that were online in 2019. Buyers pattern-match them in a second. Two things move reply rates: relevance (why me, why now) and brevity(under 90 words, one ask).

• First line names the trigger, not the flattery. "Saw you just opened a Berlin office" beats "loved your recent post."
• Middle line is the specific outcome you deliver for a company like theirs — a number if you have one.
• Ask is a question, not a meeting request. "Worth a 15-min look?" outperforms "Do you have time next Tuesday at 3?"
• No links, no attachments in touch 1. Both tank deliverability and reply rate.

Step 5 — Sequence to 4-6 touches, then stop and restart with a signal

The single-touch reply rate is a vanity number. Roughly half of the meetings we book come from touches 3 through 6. Under 4 touches and you're leaving the majority of the pipeline on the table. Beyond 6, reply rate flattens and the annoyance signal goes up.

For no-response prospects, don't churn them. Move them to a slow-nurture list and re-enter them the week a new trigger fires (new hire, funding, product launch). This is where "second attempt" campaigns book meetings that first attempts couldn't.

Step 6 — Measure the two numbers that actually matter

• Positive reply rate (not total replies) — the truth about your targeting and message.
• Meeting-to-opportunity rate — the truth about whether the meetings are with actual buyers or just polite people.

Everything else is a leading indicator. If both of those are in range and revenue still isn't moving, the problem is downstream (offer, price, sales process), not the top of funnel.