Conditional sequences change what a lead receives next based on what they did or said: opened, clicked, replied "not now", accepted a LinkedIn invite, booked a meeting. Buyers care because linear sequences waste touches on people who have already signalled. Agencies especially want "interested but not now" replies handled without a human.
How it works
There are two architectures:
- In-sequence branches. A condition node splits the flow into yes/no paths, with a time window. lemlist's advanced conditions create exactly two branches per condition. Each can be checked "within" a window (take the yes path as soon as the condition is met) or as "wait until" (pause until it is met).
- Subsequences. The main sequence stays linear, and a trigger moves the lead into a separate follow-up sequence, usually in the same thread. Instantly and Smartlead both use this model.
| Tool | Model | Triggers | Gate (Oct 2026) |
|---|---|---|---|
| lemlist | Branching canvas | Opened, clicked, unsubscribed, booked meeting, LinkedIn invite accepted, LinkedIn message opened, score, has phone, call status, has WhatsApp, custom variables | Help says Multichannel Expert or Enterprise only (help) |
| Instantly | Subsequences | Lead status (incl. custom), opens, clicks, completed without reply, exact-phrase reply keywords | Hyper Growth and Light Speed (help) |
| Smartlead | Subsequences | Lead categories (AI-categorised replies), keywords, re-engagement, bounces | Not stated (help) |
| EmailBison | "Reply follow-up campaigns" | Conditional logic, in-thread | $599 tier (homepage) |
| Woodpecker | If-else / condition-based campaigns | Not detailed | Included (pricing) |
| Saleshandy | Subsequences | Behaviour-based | All plans (pricing) |
| Reply.io | Conditional sequences | Not detailed | All plans (pricing) |
Instantly expanded subsequences in a November 2025 update to allow several interest statuses per subsequence and triggers on clicks, opens and "completed without reply".
lemlist's plan gating is contradictory. The help article restricts advanced conditions to "Multichannel Expert or Enterprise" plans, but the current pricing page lists advanced conditions across plans, and the plan names in the help article do not match the current Email and Multichannel plans. Check before quoting.
Who does it well
lemlist has the richest condition vocabulary because it owns the other channels: it can branch on a LinkedIn acceptance or a call outcome. Smartlead's subsequences are the most useful for volume senders, because they key off AI reply categories. "Interested", "not now" and "wrong person" route automatically, which is what an agency's SDR would otherwise do by hand (AI reply agent, Unified inbox (Unibox)). Instantly matches this with status and keyword triggers, but keyword matching is exact-phrase, which is brittle against real replies.
Where implementations differ
- Retroactivity. Instantly subsequences only fire for leads that meet the trigger after launch. Smartlead says subsequences must exist before the campaign is activated. Both push users into manual lead moves.
- Open-based branching. Branching on "opened" is common (lemlist, Instantly), but opens are unreliable (see A/B testing, Custom tracking domains). Branching on a fake open sends the wrong follow-up.
- Plan gating. Instantly keeps subsequences off its $47 Growth plan. lemlist's help article gates advanced conditions to its top plans. Saleshandy and Reply.io include them broadly.
- Visual vs settings-based. lemlist offers a canvas. Instantly and Smartlead configure subsequences in separate screens, so the full logic is hard to see in one place.
Apollo, Outreach and HubSpot sequence branching could not be verified from their help centres in this pass. HubSpot's KB documents tasks and a 10-email cap per sequence but not branching.
Risks and abuse
Little direct abuse. The operational risk is runaway automation: a lead marked "interested" by a wrong AI classification gets a pushy follow-up, or "wait until" conditions park thousands of leads indefinitely. Instantly's own help centre has an article on why follow-ups still send to leads who replied, a sign that state handling is fiddly at scale.
What this means for an entrant
- Build one model, a branching graph where subsequences are just branches, rather than two bolted-together systems. This is easier to explain and to audit.
- Branch on reply intent from an LLM classifier, with confidence thresholds and a human-review lane. This is where AI reply agent and conditional logic meet, and it is the frontier.
- Make triggers retroactive (apply to leads already in the campaign) by default. Both leaders make users move leads by hand.
- Do not gate branching behind the top tier. Instantly's Growth-plan omission is an easy comparison-page win (Pricing strategy).
- Avoid open-based branches by default. Prefer reply, click and CRM events (CRM sync).