Analytics & Learning Agent

Your account gets smarter with every send.

It reads behaviour continuously, keeps your segments rebuilt from what customers actually do, and turns every campaign result into a rule the other agents follow next time.

Self-improvement loopRunning continuously
PROPOSE CAMPAIGN IDEAS02 · SEND03 · ANALYSE04 · LEARN
Inputs
Shopify orders, catalog, and every Klaviyo send and open.
Refresh
Segments rebuild as behaviour changes, not on a schedule you set.
Output
Learnings the other agents read before they write anything.
Setup
None. It starts from the history already in your store.

Segmentation

01

Segments that rebuild themselves.

Nobody writes the filter. The agent watches behaviour as it happens and keeps a working set of segments cut from real signals, so a customer moves the moment their behaviour does.

A discount-led buyer who starts paying full price is re-classified before your next campaign goes out, not at the next quarterly clean-up.

  • No static lists to maintain or refresh
  • Membership updates as orders and opens land
  • Every segment is explainable, down to the signal
Live segmentationRebuilt 6 min ago
SegmentLTVFreqDisc

High-LTV loyalists

1,840 profiles

Full-price repeat

3,110 profiles

Discount-led buyers

6,420 profiles

Lapsing after one order

4,205 profiles

Learning

02

Every campaign teaches the next one.

The agent tracks how each segment responds across every send: which angle landed, which offer was unnecessary, which product story earned the click.

Those conclusions are stored as rules, not dashboards. The Personalization and Email agents read them before writing a single line, so the account compounds instead of starting over.

  • Per-segment performance, not one blended open rate
  • Findings written as reusable rules
  • Applied automatically on the next campaign
What it has learnedApplied automatically

High-LTV loyalists

Early access beats any discount

6 sends · +34% revenue per recipient vs. a 15% off offer

Offer depth locked at 0% for this segment

Discount-led buyers

Deadline language moves them, product story does not

9 sends · 2.1× click rate on time-bound subject lines

Angle set to urgency, 48-hour window

Lapsing after one order

Category they bought first outperforms bestsellers

4 sends · +22% reactivation on same-category picks

Product picks pinned to first-order category

Catalog

03

It knows your catalog, not just your customers.

Products are categorised automatically from your Shopify data, then scored against each other: what gets bought together, what follows what, and what a first order usually leads to.

That map is what makes an up-sell or cross-sell campaign specific. The agent already knows which second product belongs next to the one they own.

  • Automatic categorisation, no manual tagging
  • Co-purchase and sequence correlation across order history
  • Feeds product picks in every generated campaign
Product correlationsTop pairings

Curl creamSilk pillowcase

Cross-sell

Starter kitFull-size refill

Up-sell

Trail Runner ProMerino socks

Cross-sell

Travel setWash bag

Bundle

Experimentation

04

It runs the tests you never get around to.

For each micro-segment the agent quietly trials one variable at a time: the angle, the language, the tone, the position of the offer. Small holdouts, real sends, honest reads.

Winners are promoted into the segment’s rules and used from then on. Losers are retired. Nothing waits for someone to open a report.

  • One variable per test, per micro-segment
  • Winners promoted into the rules automatically
  • The account keeps improving between campaigns
Running experimentsThis week
Full-price repeat+18% CTR

Variant A

“Back in your size”

Variant B · winner

“Restocked, 40 left”

High-LTV loyalists+11% revenue / recipient

Variant A · winner

Warm, personal tone

Variant B

Short, editorial tone

Discount-led buyers+27% open rate

Variant A · winner

Offer in subject line

Variant B

Offer revealed in email

The loop

Watch, segment, test, apply. Then again.

No weekly review, no analyst chasing reports. The agent closes its own loop, and every other agent inherits the result.

  1. 01

    Watch

    Orders, browsing and email behaviour are read as they land.

    Continuous
  2. 02

    Segment

    Customers are grouped by value, frequency, affinity and interest.

    Self-updating
  3. 03

    Test

    Each micro-segment gets a small experiment on one variable.

    Per segment
  4. 04

    Apply

    Winning angles and offer depth become the default for the next send.

    Automatic

Get started

See what your data already knows

Connect Shopify and Klaviyo, and Nexie comes back with your segments, what each one responds to, and the campaigns worth running next.