Agentic Campaigns are the fastest way to get intent working for you. You pick a strategy, goal and add your experiences. The agent figures out who sees what and when - learning and improving with every session.
This guide covers all six strategies: what they target, how to set them up, and what to expect.
If any of your experiences include a discount, we recommend setting your goal to Revenue Per User. This ensures the agent accounts for the cost of the discount and avoids giving it away to visitors who would have bought anyway. For all other campaigns, your goal depends on what you’re optimising for.
Welcome Visits
Who it targets: First-time visitors at the very start of their journey.
Most brands treat every new visitor the same. But a first visit is a high-stakes moment - visitors are forming an impression, deciding whether to stay, and in most cases not yet ready to buy. A Welcome Visits campaign reads intent from the first interaction and responds with the experience most likely to build confidence and keep them on site.
What to set up
Example experiences: Email capture, USP or trust messaging (service guarantees, delivery promises, social proof), discovery tools (product finder, quiz)
We recommend adding a “show nothing” option as one of your five experiences — it gives the agent somewhere to route visitors who don’t need an intervention
What the agent does
Focuses on the visitor’s first session, looking for the right moment within the first few page views to engage. Tests experiences such as welcome messages, newsletter sign-up offers, new customer discounts, and brand storytelling to find what builds the best first impression.
Targets: first session only, not yet converted.
Recommended settings:
Multi Experience: Off — one consistent welcome message per visitor
Frequency: Once per journey — fires once across the entire customer journey
Ideal for: Welcome pop-ups, newsletter sign-ups, first-visit offers
Returning Visits
Who it targets: Visitors who’ve been to your site before and are coming back.
A returning visitor isn’t starting from scratch. They have history, preferences, and a reason to be back. A Returning Visits campaign uses that context to meet them where they are — surfacing recently viewed products, relevant interests, or a re-engagement message at the right moment.
What to set up
Example experiences: Recently viewed products, re-engagement banners, category or product recommendations based on past browsing
Think about what’s most likely to re-ignite interest: a reminder of what they were looking at, or a reason to take the next step
What the agent does
Fires when a visitor lands on their second session or beyond. Tests experiences such as welcome-back messages, previously viewed products, browse history nudges, and loyalty or repeat-customer offers to find what brings them back into the journey.
Targets: second session or later, landing page visit, not yet converted.
Recommended settings:
Multi Experience: On — content adapts to find the best returning visitor experience
Rebucketing Frequency: Every session — tests new variations each time they return
Frequency: Once per journey — fires once per return to avoid repetition
Ideal for: Welcome-back messages, browse history, abandoned basket reminders
Browse Abandonment
Who it targets: Visitors who are browsing but not committing — showing intent to explore but at risk of leaving without engaging further.
Not every visitor who’s about to leave needs the same response. Or any response at all. A Browse Abandonment campaign identifies which visitors are genuinely persuadable, finds the moment they’re most open to an intervention, and decides which experience — a discount, email capture, or brand message — will land best for each one.
What to set up
Goal: Revenue Per User — accounts for the cost of discounting so the agent doesn’t give offers to visitors who’d have engaged anyway
Example experiences: Email capture, a discount offer, a brand or product message, a “show nothing” baseline
What the agent does
Targets browsers with nothing in their basket who are showing signs of leaving. Tests the timing to catch the critical exit moment, and tries different responses such as; discount codes, free shipping thresholds, delivery messages, and exit overlays, to find what works for each visitor.
Targets: no items added to basket, not yet converted.
Recommended settings:
Multi Experience: Off — one consistent message per session
Frequency: Once per session — captures abandonment intent each session
Ideal for: Exit-intent pop-ups, discount overlays, urgency messaging
Basket Abandonment
Who it targets: Visitors who’ve added something to their basket but are showing signs of leaving without buying.
Giving a discount to everyone who abandons their basket means giving a discount to people who’d have bought anyway. A Basket Abandonment campaign distributes your options — tiered discounts, trust messages, email capture, or nothing at all — across hundreds of visitor contexts, learning who actually needs the nudge and who doesn’t.
What to set up
Goal: Revenue Per User — essential for protecting margin when discounts are in the mix
Example experiences: A trust or reassurance message, tiered discount levels (e.g. 0%, 5%, 10%, 15%), email capture, a “show nothing” baseline
Don’t just add your biggest discount — add a range. The agent learns which visitors need a price nudge and which just need reassurance
What the agent does
Targets visitors with items in their basket who are at risk of leaving. Tests the timing to catch the abandonment moment before exit, and tries different responses such as discounts, free shipping offers, delivery guarantees, and trust signals like secure checkout messaging.
Targets: at least one item added to basket, not yet converted.
Recommended settings:
Multi Experience: Off — one consistent recovery message
Frequency: Once per session — one message per session to avoid intrusiveness
Ideal for: Cart recovery overlays, exit offers, urgency messaging
Product Visits
Who it targets: Visitors on product pages — the moment consideration is highest and the right message can tip the decision.
What works for a visitor who’s nearly ready to buy is very different from what works for someone still comparing. A Product Visits campaign surfaces the right message — social proof, urgency, a USP, or a discovery tool — for each visitor based on where they are in their decision.
What to set up
Example experiences: Social proof (real-time signals — add-to-carts, purchases, views), trust or USP banners, urgency or delivery messaging, tooltips, discovery or consideration tools
Think about the different mindsets a product page visitor might have — still comparing vs. nearly ready to buy. Give the agent experiences that speak to both
What the agent does
Targets visitors actively viewing product pages. Tests on-page content such as social proof, recommendations, urgency around stock or delivery, and trust signals, to find what moves each visitor closer to a decision.
Targets: visitor is on a product page, not yet converted.
Recommended settings:
Multi Experience: On — content adapts as the agent learns optimal variations
Rebucketing Frequency: Every page — tests new variations on each product viewed
Frequency: Once per page — fires on every product page to optimise each interaction
Ideal for: On-page product content, social proof, trust badges
Experience Optimisation
Who it targets: Every eligible visitor, across the entire journey — no specific moment, no single trigger.
Sometimes you don’t want to target one touchpoint or moment. You want the right experience to reach each visitor wherever they are. Experience Optimisation campaigns surface prompts, recommendations, content, and more based on real-time intent across the full session, adapting as visitors move through the site.
What the agent does
Targets all eligible visitors across the whole journey with no pre-set behavioural filters. Segments by behaviour and intent as it goes, testing different experiences at different moments to find what works for each visitor context.
Targets: not yet converted (no other restrictions).
Recommended settings:
Multi Experience: On — content adapts continuously across all interactions
Rebucketing Frequency: Every page — tests new variations on each page view
Frequency: Once per page — fires on every page for maximum flexibility
Ideal for: Functionality that exists on site which you’d like to work harder for you