ExperimentIQ
Agentic experimentation for CRM and lifecycle teams

Your team runs dozens of tests.
Where do the learnings go?

ExperimentIQ turns every experiment into insights your team can act on, then proposes and runs the next test. So you run more experiments and grow the revenue they bring in.

Become a design partner →See how it works
app.experimentiq.ai
ExperimentsPlannerEmail LabLearningsInsights

What we've learned · All experiments

Specific, personal messages beat generic offers. A discount rarely adds lift on top.

23 W · 6 L · 12 inconclusive

Proven impact

+€412k≈ €34k / month
1
Naming the exact item beats a generic offerStrong
Named product or plan out-performed discount-led copy in 8 of 10 tests.
8 W · 1 L · 1 —
+€188k€16k/mo
2
Urgency works only near a real deadline
Helped when the deadline was days away, hurt when it was weeks away.
4 W · 2 L · 3 —
+€97k€8k/mo

The problem

Experimentation shouldn't feel like guesswork

Most CRM teams test a lot and keep very little.

Results are scattered

Tests live in spreadsheets, decks and Slack threads. Nobody can see what has been tried.

Learnings get lost

When people move on, so does the knowledge. The same tests get run again.

The next test is a guess

Ideas come from gut feeling, and every arm's copy is written and translated by hand.

What changes

One place that learns from every test and runs the next one

ExperimentIQ tracks your experiments, turns the results into insights, and launches new tests on its own, across push, email and in-app.

4×more experiments
+50%revenue from experimentation

Early results. Individual results will vary.

Results · Insights

Learnings roll up into insights

The learning library rolls up into themes your whole team can act on, each backed by the tests behind it.

  • What works and what doesn't, with the evidence behind each
  • Wins, losses and inconclusive results, counted honestly
  • Proven revenue impact, not just uplift percentages
  • Filter by channel, category or label
Naming the exact item beats a generic offerStrong
10
tests
8 W
1 L · 1 —
+8.4%
median win
+€188k
proven

What works

  • The product or plan name in the title
  • Personal context, like back in stock

What doesn't

  • A discount on top of a named item
  • Generic "don't miss out" urgency
2
Urgency works only near a real deadline
Helped when the deadline was days away, hurt when weeks away.
4 W · 2 L · 3 —
+€97k€8k/mo
3
Social proof lifts first-time customers
New users responded; returning users ignored it.
3 W · 3 —
+€74k€6k/mo

Results · Email Lab

Complex emails, built from your own data

Build or import a template, connect live sources like your product catalogue or recommendation API, and let the AI CoPilot edit, translate and check every arm. Send it to your CRM when it's ready.

ExperimentsLearningsInsightsCampaign PlannerEmail Lab
Import Open Connected data3 examples · 12 blocks CoPilot English (source) Preview dataSaved Export
Subject{{first_name}}, picked for you this week39/90PreheaderNew in, back in stock and on your wishlist
BlocksBasicsIssues OK

Header

Hero

Product row Connected

Footer Required

DesignPreviewSample data
BRANDView online

Picked for you, Sam

Shop the edit

Back in stock and on your wishlist

Trail runner

€89

Rain shell

€120

Day pack

€65
See all picks
AI CoPilot

Editing: this email

Add a row of 3 recommended products under the hero, with prices
Added a product row filled from recs. It shows each customer's top 3 picks.
Fetched connected dataChecked catalogue
Applied to the email · Undo
Translate it into French and German
Done: 14 slots in 2 languages.
Translated
Whole emailHeroProduct row
e.g. shorten the hero headline
Connected data

Live values this template pulls from an API

recs

Recommendations API

6 fieldsuserId = {{id}}
catalogue

Product catalogue

8 fieldsmarket = {{locale}}
Connected data

Pull products, prices and recommendations live from your APIs.

AI CoPilot

Describe a change; review it before it is applied.

Translations

Every slot, every language, checked for length.

Checks before export

Structure and content checks, then send to your CRM.

Why ExperimentIQ

How teams work today, and with ExperimentIQ

TodayWith ExperimentIQ
TrackingSpreadsheets, decks and Slack threadsEvery experiment in one place, results synced live
Calling a winnerManual significance checks, if anySignificance calculated, with an alert when a test is ready
LearningsLost when people move onWritten for every test, in a searchable library
Next testPicked by gut feelingHypotheses generated from what has already worked
Copy and translationsWritten by hand, per arm and languageWritten for every arm, in every language
Running testsSet up one by oneAuto-mode runs them and promotes the winners

Track

Every test becomes a learning

Results sync from your data and significance is calculated for you. When a test concludes, ExperimentIQ writes the learning into your library.

PushPositive

Name the item, not the offer

3 Sep – 17 Sep (14d)|A/B Test|2 variants| Winner: Variant 1

Re-engagement# Personalisation

Revenue Uplift

+12.4%

Attributed Uplift

+€41k

Monthly Run-Rate

+€16k
Results
MetricControlVariant 1 Winner
Send Volume120,400120,180
Conversion rate96% significance2.04%2.24%+9.8%
Click rate88% significance6.1%6.6%+8.2%
Attributed Revenue€182k€223k
Push

Naming the item lifts conversion

18 September 2026

Naming the product the customer last viewed outperformed a generic reminder. The team could test the same angle in email.

Conversion rateStrong

+9.8% uplift vs control

Evidence2 linked experiments · Concluded

Linked experiments

Name the item, not the offerConcluded
Back in stock, with the item nameConcluded

Target metric

Click rate31 testsConversion rate24 testsRevenue per user19 testsOrders12 tests
Experiments it can run Conversion rate
Name the item, not the offer

Naming the last-viewed item should lift conversion over a generic reminder.

Push · 3 arms · 6 languages · builds on 8 wins in 10 tests

Run in auto-mode
Send closer to the deadline

Push · control + 1 challenger

Run
Picked-for-you weekly email

Email · built from your product catalogue

Run

Running · Push

Name the item, not the offer
Auto-mode
Window 3Sending now
Window 2Ended

Next challenger is generated from the winner and your learning library

Launch

Pick the target metric. It runs the experiments.

Choose what you want to move. ExperimentIQ generates hypotheses from your learning library, proposes the experiments it can run, writes the copy for every arm and runs them in auto-mode. Each window's winner becomes the new control.

  • Hypotheses generated from your learnings
  • Copy written for every arm: push, email and in-app
  • Personalised and translated into every language
  • Every result goes back into the learning library

Agents

For your team, and for your agents

Everything in the app is also available through the API and an MCP server. Your team uses the UI; Claude and other agents can plan, launch and read experiments directly.

  • The same operations in the UI, API and MCP
  • Scoped API keys with read and write access
  • Your experiment history stays the single source of truth
# In Claude, connected to ExperimentIQ › What should we test next to lift conversion? campaign_experiment_suggest(targetMetric: "conversion_rate") → 3 hypotheses, built on 24 past tests › Go with the first one. Schedule it for Monday. experiment_upsert(…) → draft created window_schedule(start: "Mon 09:00") → scheduled

How it works

Every test feeds the next one

Track, learn, find the patterns, launch the next test, then let auto-mode repeat it. Your agents can run the same loop through MCP.

1

Track

Every experiment and its results in one place, updated live.

2

Learn

Each concluded test becomes a learning in your library.

3

Insights

Learnings roll up into themes, with evidence and proven revenue.

4

Launch

Pick a metric. It generates hypotheses, writes the copy and runs the experiments.

app.experimentiq.ai
ExperimentsPlannerEmail LabLearningsInsights

1 · Track

PushLaunched

Name the item, not the offer

MetricControlVariant 1 Leading
Conversion rate96% significance2.04%2.24%+9.8%
Click rate6.1%6.6%+8.2%

2 · Learn

Push

Naming the item lifts conversion

Conversion rateStrong

+9.8% uplift vs control

Evidence2 linked experiments

3 · Insights

1
Naming the exact item beats a generic offerStrong
8 W · 1 L · 1 —
+€188k
2
Urgency works only near a real deadline
4 W · 2 L · 3 —
+€97k

4 · Launch

Target metric

Click rate31 testsConversion rate24 tests
Experiments it can run
Name the item, not the offer
Run in auto-mode
Picked-for-you weekly email
Run

Then auto-mode repeats it

Name the item, not the offer

Auto-mode
Window 3Sending now
Window 2Ended
Ready to call

Become a design partner

We're building ExperimentIQ now and looking for a few CRM and lifecycle teams to shape it with us, with early access, hands-on onboarding and direct input into the roadmap.

We only use your details to reply. See our privacy policy.