Skip to main content
liftstack
Sign In Start Free Trial

How liftstack works

A longer-form walkthrough of the product workflow. Visual, non-technical language throughout. Five steps from content variants to revenue-denominated verdicts.

1

Create Your Snippets

Reusable content with multiple variants

A snippet is a reusable piece of content with multiple variants. Snippets work across all channels: email hero blocks, subject lines, CTAs, push notification titles and body, SMS copy, in-app message content, and content cards. Each variant is a different creative approach to the same content position.

One variant is marked as the control, your baseline. This is typically the copy you would have sent anyway. All performance is measured against this baseline so you know the true incremental value of testing.

Snippets live in a library that persists across campaigns and channels. You can reuse the same snippet in multiple campaigns, and liftstack accumulates performance data over time.

Snippet Preview

Spring Sale Hero

Control Exclusivity, “Members-only early access to our spring collection”
Variant B Urgency, “Spring sale ends tonight. Don't miss 40% off.”
Variant C Social Proof, “Join 12,000 customers who already shopped the spring sale”
2

Build Your Campaign

Connect snippets to slots in your message

Choose your channel (email, push, SMS, in-app message, or content card) and connect snippets to slots. A slot is a named content position where you test different variants. For email, that might be a subject line, hero block, or CTA. For push, it could be the title or body. For SMS, the message copy.

You choose:

  • Which snippets go in which slots
  • Which variants to include
  • Your channel and primary metric (open rate, click rate, conversion rate, tap rate, or revenue per exposure)
  • Custom conversion goals (optional): choose which platform events count as conversions
  • Campaign group (optional): coordinate tests across multiple channels with shared variant assignment
  • Holdout percentage (optional, advanced): reserve up to 20% of the audience to measure the value of having content vs not having it
  • How traffic should be allocated (equal split, manual percentages, or Thompson Sampling)

After your audience is synced, liftstack shows a sample size guidance card that tells you whether your audience is large enough for the number of variants you're testing. If it's not, you'll see specific guidance on how to adjust.

Campaign Builder

Channel selector: Email

Subject Line

3 variants

Hero Block

3 variants

CTA

2 variants

Primary metric: Revenue per exposure
3

Assignment: The Key Differentiator

A proper randomised experiment, every time

Before your message is sent, liftstack:

  1. Pulls every recipient’s profile from your CRM platform
  2. Assigns each recipient a variant per slot, recorded in a permanent assignment ledger before any message is sent
  3. Writes the assignment to each recipient’s profile as custom properties
  4. Generates a conditional template in your platform’s native syntax (Liquid, Handlebars, Velocity, Braze Liquid)

Because assignments are recorded before outcomes are observed, this is a proper randomised experiment. You know exactly who saw what, and any difference in conversion rates is attributable to the content they saw, not to selection bias.

Audience liftstack Assignment Profile Properties Conditional Template Platform Sends Message
4

Track Results

Events flow back and join the assignment ledger

For email and SMS, each link is tagged with tracking parameters that identify both the slot and the variant. For push and in-app, attribution works via profile-based assignment lookup. When a recipient clicks, opens, converts, taps, or triggers any tracked event, the events flow back to liftstack via webhooks or polling.

liftstack joins each event back to the assignment ledger: “Profile 001 was assigned Hero Variant B and subsequently converted with a £67 order.”

5

Read Your Verdict

Clear outcomes, no statistical expertise required

Every slot in every campaign gets a clear verdict:

Winner Found

One variant outperformed the others with high confidence. Includes conversion rate difference, probability, and estimated additional revenue.

= No Meaningful Difference

All variants performed within a negligible range. Pick whichever fits your brand voice best.

More Data Needed

The current leader is shown with its probability. Includes an estimate of how many more exposures are needed.

! Guardrail Violation

A variant would win on the primary metric but triggered a safety guardrail (e.g., unsubscribe spike). Review the content before proceeding.

Thompson Sampling: Smart Allocation

After 2–3 campaigns with the same snippet, you can switch to Thompson Sampling. This sends more traffic to better-performing variants while still testing all options.

Control floor (10%)

The control always gets at least 10% of traffic

Variant floor (5%)

No variant drops below 5%

New variant warm start (20%)

New variants get at least 20% on their first campaign

Recency weighting

Recent results count more than old ones (60-day half-life)

You can accept the system recommendation, adjust manually, or override to equal split.

Start compounding revenue from the campaigns you already send

14-day free trial on the Growth tier. No credit card required.