A system that never stopsexploring your growth.
Continuous exploration, turned into action across campaigns, product, landing and paid. Everything else becomes a sized recommendation.
The actual difference
Markin is not another decisioning engine.
Markin is not a decisioning engine. A decisioning engine ranks actions a human already defined. Markin works like a data science and growth team: it forms its own hypotheses about why ARPU is stuck, marketing, product, pricing or technical, sizes them, executes them inside the systems you already run, and reads each one against a holdout.
| A decisioning engine | Markin | |
|---|---|---|
| Where the hypothesis comes from | A human authors it. The engine chooses between options someone already approved. | Markin authors it. It reads the base, finds where revenue is leaking or unclaimed, and writes the hypothesis itself. |
| What it is allowed to question | Message, offer, channel, timing, inside the campaign surface it was given. | Anything that moves ARPU: onboarding friction, pricing and packaging, a feature nobody adopts, a payment failure spike, a broken deeplink. |
| Who does the analysis | Your analysts, before and after. The engine optimises; it does not investigate. | Markin does the analysis. Sizing, segment definition, experiment design and readout are automated end to end. |
| Where it stops | At the recommendation. Someone still has to build and launch it. | It launches. Markin executes inside your existing platforms and product surfaces, then closes the loop on the result. |
| Throughput | As many hypotheses as your roadmap has room for, typically a handful per quarter. | Hundreds in parallel, every one carrying a control group. |
| What happens when it is wrong | The programme keeps running until someone reviews it. | It is retired automatically. Failing to beat control is a normal, cheap outcome. |
A decisioning engine picks the best action from a list you wrote. Markin writes the list, and runs it in your stack.
Every surface of growth
Exploration never stops at the campaign.
Markin explores your data continuously and turns what it finds into action wherever growth actually happens. The loop keeps learning and keeps launching, across every surface your team operates.
Campaigns & CRM
Offers, lifecycle, win-back and contact pressure, executed per customer inside Braze, Iterable, Salesforce or your existing stack.
Product experiments
Onboarding steps, feature discovery, paywall placement and in-product treatments, run as controlled experiments like anything else.
Landing & web
Journey and page tests routed to your web experimentation stack, with the same holdouts and the same causal read.
Paid media
Audience and budget shifts, including suppressing customers who would convert anyway, measured on incremental ARPU rather than clicks.
Pricing & offers
Discount depth, plan framing and bundle structure tested by cohort against margin, not argued about in meetings.
Fallback
And what Markin cannot execute
It does not disappear. It arrives as a sized recommendation: the evidence, the expected incremental ARPU, the suggested owner and the first step, ready for a human to pick up.
Surface 01 · Context & exploration
Exploration runs continuously, not when someone asks a question.
Markin keeps reading the whole picture: behaviour, transactions, product usage, support, payments and channel history. It does not wait for a brief. It investigates on its own, opens a live case for anything that looks like money left on the table, and shows the evidence, the cohorts, the tables consulted and the reasoning behind each conclusion.
markin · investigations / OPP-4821
Investigation · OPP-4821
eSIM plans undersized for long-trip travelers.
Travelers heading to destinations with average stays > 21 days are buying 7-day plans and topping up mid-trip, at a worse unit price.
Signal detected
Top-up rate for LATAM & SEA destinations rose 34% MoM, concentrated on 7-day plans purchased 0–2 days before departure.
Supporting evidence
Top-up frequency by destination avg stay, last 90 days.
Customer & cohort context
48,210 customers · avg trip 24d · 71% first-time long-trip · 62% Android · destination clusters: MX, BR, TH, ID, VN.
Data consulted
Hypotheses generated
- 1. Plan ladder misaligned with true trip length.
- 2. Purchase UX anchors on 7-day default.
- 3. Long-trip travelers unaware of 30-day option.
Economic sizing
+ €0.00
ARPU / eligible customer
Range €1.40 – €2.80 · margin band +2.4 to +3.9 pts.
Surface 02 · Opportunity feed
Everything worth pursuing, sized in money, in one queue.
Pricing friction, an onboarding step that loses people, a feature nobody finds, a paid audience being paid for twice, a churn risk cohort. Each arrives with estimated ARPU and margin impact, confidence, affected population and a recommended next step. The queue is ordered by expected value, so the team works on money rather than on requests.
markin · opportunities
Opportunity feed
127 open · by expected ARPU
- Revenue expansion+ €2.10 ARPU
eSIM plans undersized for long-trip travelers
OPP-4821 · 48,210 customers
86%Ready→ Design experiment - Margin recovery+ 3.8 pts margin
Discounts leaking to guaranteed buyers
OPP-4817 · 12,904 customers
79%Investigating→ Sharpen eligibility - Product issue− €0.90 ARPU risk
Failed activations after v4.2 release
OPP-4809 · 6,441 customers
93%In review→ Route to engineering - Retention+ €4.60 ARPU
Usage decay in top-decile customers
OPP-4802 · 3,120 customers
71%New→ Investigate cohort - Cross-sell+ €1.40 ARPU
Combo A+C associated with +18% retention
OPP-4791 · 22,506 customers
68%Ready→ Design experiment
Hypothesis space
Everything a human growth scientist would look at.
Most growth problems are not message problems. Markin is not restricted to the campaign surface: if something is holding ARPU back, it is in scope, and it gets tested the same way.
Marketing
The classic surface, but chosen per customer rather than per segment, and always against a holdout.
- Which offer this specific customer is worth making
- Channel and timing chosen per person, not per campaign
- Contact pressure and fatigue arbitrated across every programme
- Win-back economics: who is worth a discount and who is not
Product
Where the customer actually experiences the value, and where most silent revenue loss happens.
- Onboarding steps that lose customers before first value
- A feature with high retention correlation that half the base never discovers
- Paywall and upgrade prompt placement
- In-product surfaces used as a treatment arm, not just email and push
Commercial
Pricing, packaging and the shape of the offer itself, tested rather than argued about.
- Plan and bundle structure by cohort
- Discount depth against margin, not against conversion alone
- Annual versus monthly framing per customer
- Dunning and involuntary churn recovery sequences
Technical health
Anomalies nobody asked it to look for. This is the category no decisioning engine covers.
- A checkout error rate that rose on one device and one region
- Payment failures concentrated in a single issuer or method
- A broken deeplink quietly killing a high-value journey
- Latency or delivery degradation eating conversion before any message does
Think of Markin as a data science and growth team that never sleeps: it investigates, forms hypotheses, ships them into your own stack and proves each one against a control group, at a volume no human team can reach.
Surface 03 · Candidate actions
The answer is rarely another message.
For each opportunity Markin proposes candidate actions wherever the fix actually lives: a price or packaging change, a product or onboarding change, a landing test, a paid audience shift, a recommendation model, a CRM journey, a support intervention, an engineering ticket, or no action at all. Each option is sized by expected incremental ARPU and margin, with the system that must run it and the guardrails already attached.
markin · investigations / OPP-4821 / candidate actions
Candidate actions
7 candidates · ranked by expected incremental ARPU
- Pricing+ €2.40/ + 3.1 pts82%
Introduce 30-day plan tier at €39
Billing · Stripe · Discount ≤ 12% · 48,210
Approved - Product+ €1.60/ + 2.2 pts74%
Change checkout default to trip-length picker
Product · Feature flag · iOS + Android · All new
Pending - Recommendation+ €1.20/ + 1.8 pts71%
In-app 30-day upsell during activation
App · Native · Frequency cap 1/trip · 31,004
Auto - CRM+ €0.80/ + 0.9 pts62%
Pre-departure long-trip nudge
Braze · Push + Email · Quiet hours · 22,410
Pending - Engineering+ €0.60/ + 0.4 pts90%
Fix plan-ladder rendering bug on Android 12
Jira · Ticket · P1 · 6,441
Auto - Human review, / , 99%
Legal review of auto-renewal copy (EU)
Internal task · Blocking · n/a
Pending - No action, / , 65%
Hold on France · low expected uplift
, · Revisit in 30d · 3,904
Auto
Surface 04 · Experiment design
Anything that ships, ships as an experiment.
Treatment and control, eligibility, primary KPI, measurement window, margin and communication guardrails, and required approvals. An experiment runs wherever the action lives: a campaign, a product change, a landing test, a pricing move or a paid audience. The design is the same in every surface, which is what makes results comparable across the whole growth operation.
markin · experiments / EXP-2214
Experiment · EXP-2214
30-day plan tier · long-trip travelers
Treatment groups
· T1 · Show 30-day tier at €39
· T2 · Show 30-day tier + pre-departure nudge
Control group
Current plan ladder · 20% of eligible traffic
Eligibility
First-time long-trip customers · avg stay > 21d · ≤ 1 push + 1 email per traveler · quiet hours local time.
Primary KPI
Incremental ARPU / activated customer @ trip-end
Guardrail · margin ≥ baseline − 0.5 pts · 45d window, MDE 3.2% @ 80%.
Approvals
· Growth lead · approved
· Pricing · approved
· Legal (EU) ·pending
Rollout plan
Ramp 5% → 20% → 50% over 3 weeks · auto-stop on guardrail breach.
Surface 05 · Execution & handover
Markin executes what it can and hands over what it cannot.
Most decisions leave Markin automatically into the CRM, channels, pricing, product, web or paid systems you already run. When something needs a human, an engineering change or a system Markin does not touch, it does not disappear: it becomes a sized recommendation with the evidence, the expected impact, the suggested owner and the first step. Your existing systems stay the systems of record.
Where your data lives
Warehouses, CDPs, product analytics, CMS.
Segment
Customer Data Platforms
mParticle
Customer Data Platforms
Rudderstack
Customer Data Platforms
Snowflake
Data Warehouse & Storage
BigQuery
Data Warehouse & Storage
Databricks
Data Warehouse & Storage
Redshift
Data Warehouse & Storage
Contentful
Content Management
Sanity
Content Management
Contentstack
Content Management
Amplitude
Product & Behavior
Mixpanel
Product & Behavior
PostHog
Product & Behavior
Segment
Customer Data Platforms
mParticle
Customer Data Platforms
Rudderstack
Customer Data Platforms
Snowflake
Data Warehouse & Storage
BigQuery
Data Warehouse & Storage
Databricks
Data Warehouse & Storage
Redshift
Data Warehouse & Storage
Contentful
Content Management
Sanity
Content Management
Contentstack
Content Management
Amplitude
Product & Behavior
Mixpanel
Product & Behavior
PostHog
Product & Behavior
Markin
Growth Agentic OS
- 01Data & Context
- 02Intelligence Core
- 03Decisioning Agents
- 04Open Web
- 05Channels & Delivery
- 06Experimentation
- 07Measurement
- 08Governance
Tenant-isolated. Always on.
Where growth happens
Messaging, ads, delivery, observability.
Firebase
Delivery Infrastructure
Twilio
Delivery Infrastructure
AWS SNS
Delivery Infrastructure
SendGrid
Delivery Infrastructure
Braze
Engagement Tools
Iterable
Engagement Tools
Klaviyo
Engagement Tools
Salesforce
Engagement Tools
Meta
Ads & Audiences
Google Ads
Ads & Audiences
TikTok
Ads & Audiences
Slack
Observability & Teams
Jira
Observability & Teams
Linear
Observability & Teams
Looker
Observability & Teams
Firebase
Delivery Infrastructure
Twilio
Delivery Infrastructure
AWS SNS
Delivery Infrastructure
SendGrid
Delivery Infrastructure
Braze
Engagement Tools
Iterable
Engagement Tools
Klaviyo
Engagement Tools
Salesforce
Engagement Tools
Meta
Ads & Audiences
Google Ads
Ads & Audiences
TikTok
Ads & Audiences
Slack
Observability & Teams
Jira
Observability & Teams
Linear
Observability & Teams
Looker
Observability & Teams
Surface 06 · Incremental impact & learning
Every result feeds the next exploration.
Impact is read as incremental ARPU, revenue and margin against a holdout, per experiment and across the portfolio: opportunities found, experiments live, actions scaled, actions stopped. What beats control scales, what does not is retired, and both outcomes update what Markin looks for next. The system gets better at your business the longer it runs.
markin · impact · last 90 days
Incremental impact
Causal · treatment vs control · confidence-adjusted
Incremental ARPU
+ €0.00
rolling 30d · 92% conf
Incremental revenue
+ €0.00M
last 90d · attributed
Incremental margin
+ 0.0 pts
net of guardrails
Treatment vs control · ARPU
Portfolio
- Opportunities discovered312
- Experiments live48
- Actions scaled27
- Actions stopped14
Job to be done
The same work, at a different throughput.
Nothing below needs a tool that does not exist. It needs the work to happen continuously instead of once a quarter, and to be proven against a holdout instead of argued about.
| Job to be done | Your stack today | With Markin |
|---|---|---|
| Notice that revenue per customer is drifting in a segment | Someone spots it in a dashboard review, weeks after it started. | Detected as a signal the day the drift clears noise, with the segment already sized. |
| Explain why it is happening | An analyst is pulled off the roadmap for a two-week investigation. | An investigation runs automatically and returns the drivers with their evidence. |
| Come up with hypotheses worth testing | A workshop produces the handful of ideas the room happened to think of. | Hypotheses are written continuously across marketing, product, pricing and technical health. |
| Decide which hypotheses deserve budget | Prioritised by seniority and gut feel, with no size attached. | Each one is sized in revenue and ranked before anything is built. |
| Choose the next best action for one customer | Segment rules and campaign calendars decide, refreshed when someone has time. | Chosen per customer, per moment, against everything else competing for that customer. |
| Actually launch it | A ticket to the lifecycle team, then a slot in next month's calendar. | Executed inside the systems you already run, with no new channel to adopt. |
| Prove it caused the revenue | Reported against non-qualifiers or a global holdout, if at all. | Every decision carries a randomised control group; uplift is read against it. |
| Kill what does not work | Programmes survive because nobody owns retiring them. | Failing to beat control retires the programme automatically. |
| Do all of it again next week | Capacity-bound: four to eight tests a quarter. | Hundreds of hypotheses in flight in parallel, continuously. |
Architecture
Four areas, one product.
Context, Intelligence, Execution, and Learning & Control. Every capability in Markin lives under one of the four.
Context
Customer, product and transaction signals, identity resolution, external intelligence.
- Customer data
- Identity
- Product & transaction signals
- External intelligence
Intelligence
The methods used to understand behavior and estimate impact.
- Predictive models
- Uplift & causal models
- Research agents
- Hypothesis generation
- Decisioning
Execution
Where decisions leave Markin and reach your customers, systems or teams.
- Candidate actions
- Integrations
- CRM & channels
- Product, support & internal routing
Learning & Control
How Markin stays measurable, safe and trusted.
- Experimentation
- Measurement
- Governance
- Auditability
- Human approval
Applied research
The method that fits the problem.
No single model solves every decision. Markin combines the families below and picks the right one per problem. Predictive models estimate behavior; causal methods establish incremental impact.
Governance
Enterprise controls, on by default.
Permissions
Role-scoped access, per workspace and per action.
Eligibility rules
Customer, cohort and market constraints applied upstream.
Margin & discount limits
Guardrails enforced at the decision layer.
Communication limits
Frequency caps, quiet hours, channel budgets.
Human approval
Configurable approvals per action type and risk band.
Audit trails
Every decision, model version and outcome is traceable.
Model & experiment versioning
Every artifact is versioned and reproducible.
Emergency pause
Kill switch per action, agent, experiment or tenant.
Research
Further reading
How we calculate ARPU, how next best action marketing is measured, and where the 2026 benchmarks sit.
- Read the guide
Guides
What is ARPU? Definition, formula and how to calculate it
ARPU calculation explained: the formula, a step-by-step method, variants by business model and benchmark levels from 147 B2C operators.
- Read the guide
Guides
Next best action marketing: what it is and how it works in 2026
Next best action marketing explained: how it works, real B2C examples, how it differs from segment campaigns and marketing automation, and what it is worth on a 2M base.
- Read the guide
Research
ARPU benchmarks by industry 2026: median vs top-quintile
ARPU benchmarks by industry for 2026: median and top-quintile monthly ARPU levels and growth rates across nine B2C categories, with the method.
See Markin working on your ARPU.
30 minutes with our team. Bring one growth question, leave with an opportunity feed scoped to your business.