Skip to content

Marketing engineering

Your marketing
know-how.
Built to scale.

Turn what makes your business different into intelligent marketing systems that work at scale.

We find the parts worth engineering, combine your knowledge with data and software, and prove whether they help the business.

Find what is worth engineering See how we work

Free discovery. A useful first conversation with Adam.

A system with your business in mind
01 / THE INGREDIENTSYour business signals
+ The things only
your team knows.Yes, even that spreadsheet.
02 / THE DECISIONMake the right call.Your context + the right software.
03 / THE POINTA better business outcome.Act. Measure. Learn. Improve.
Keep learning from what happens next.

Marketing and data expertise
trusted by teams at

The work behind the logos

More possible. More discerning.

Not everything is
worth automating.

Technology has made far more possible.
That makes judgment more important, not less.

Some work should stay human. Some is already solved by good software. Some ideas are technically impressive and commercially pointless.

The opportunity is to find the few places where custom engineering makes a real difference. Including things your team could never realistically do before.

What earns the right to become a system?

Four questions before we write code.
01

A meaningful outcome

Profit, valuable customers or a decision that materially changes the business.

02

A reason to make it yours

Company knowledge and custom logic that off-the-shelf software cannot use well.

03

Enough scale to matter

A recurring decision, a growing workload or a new capability that earns its keep.

04

A credible way to prove it

Useful data, feasible delivery, measurable progress and an acceptable level of risk.

Your unfairly specific advantage

Your company knows
things AI doesn’t.

How your customers behave. When a campaign usually picks up. Which sales are actually profitable. The exceptions your best people spot instinctively.

The valuable part is knowing which of those things changes the decision.

01 / Promotion response

More sales. But more profit?

Promotion response curveA conceptual comparison: the generic assumption is that deeper discounts keep increasing profit. With your margins and customer response included, profit peaks and then falls. The decision is where to stop discounting.Profit contributionLowerHigherDiscount depthThe turning point
The best discount isn’t the biggest one.

Your margins, stock levels and customers’ response change where a promotion stops paying. A generic rule cannot tell you where that point is.

What changes the decision

Optimise for the sales worth making, not just more sales.

02 / Product trajectory

A slow start. Or a real problem?

Product trajectoryA conceptual product launch: the generic benchmark grows steadily. Your expected pattern starts more slowly, then accelerates. A highlighted early gap is expected, so it is not automatically a reason to change course.Sales momentumLaunchEstablishedTime since launchBehind the benchmark.On your own curve.
A familiar pattern needs a different response.

Your product may take time to find its customers. A generic alert sees a slow start. Your team recognises a pattern from previous launches.

What changes the decision

Compare performance with your own expected curve. Investigate meaningful deviations.

General-purpose models bring broad knowledge. Your business brings the context. We work out how to put the useful parts together.

Marketing people. With engineering depth.

We don’t replace your team.
We give its know-how more reach.

Working alongside your marketing leaders, channel specialists and data team, we connect the business question to the system that can answer it.

01 / Marketing judgment

Find the right opportunity.

Marketers who understand what matters, which exceptions count and what your team will actually use.

02 / Data & measurement

Make it accountable.

Data people who know which signals to trust, what needs fixing and how to connect an action to an outcome.

03 / AI & software engineering

Make it work at scale.

Engineers who can integrate, deploy and maintain a working system, with the right boundaries built in.

From a good question to a working system

Prove one valuable loop.
Then build on it.

Start with something valuable, achievable and measurable. The biggest possible project can wait.

  1. 01Free

    Find the opportunity.

    A conversation with Adam about value, scale, your knowledge and readiness. Is custom engineering the right answer at all?

    A reason to proceed. Or not to.
  2. 02First paid engagement

    Audit + working POC.

    Map decisions, data and internal knowledge. Compare possible workflows, including new possibilities. Build one narrow proof of concept with real data and an agreed measure of success.

    A working loop and a production roadmap.
  3. 03Production

    Make it dependable.

    Integrate it into daily work. Add monitoring, approval boundaries, failure handling and a clear owner.

    A system your team can rely on.
  4. 04What comes next

    Extend. Or take the reins.

    Improve the system, explore the next opportunity or transfer the knowledge and tools to your team.

    A growing capability, on your terms.
Built into every step

“If this works, how will we know?”

Incremental profit where we can credibly measure it. Customer lifetime value and qualified leads where they fit. Otherwise, an honest measure of decision quality, testing capacity, errors or consistency. We agree the evidence before the build.

How we measure value

As much intelligence as the job needs

Not everything
needs an agent.

Sometimes rules, SQL and ordinary software are the right answer. Sometimes AI makes something new possible. Autonomy is a design decision, not the goal.

01

Deterministic software

Clear rules. Predictable execution.

02

AI-assisted work

Help your team investigate and decide.

03

Human approval

Prepare the action. Let a person approve it.

04

Bounded autonomy

Act within agreed limits. Escalate exceptions.

These are design choices, not levels to climb. Even a simple rules-based system can act autonomously.

What if our data has to stay inside our environment?

We can use your infrastructure, local processing, deterministic software or local/open models, with restricted or no external AI calls. Your security, privacy and regulatory constraints shape the architecture.

Local and tightly constrained models offer more control over data and can solve many useful tasks. For demanding reasoning and complex agentic work, frontier cloud models generally offer stronger capabilities. We evaluate the tradeoff for your task, alongside cost, confidence, business risk and reversibility.

Built to survive past the POC

A demo works once.
Your system has to keep working.

Data changes. Platforms change. Business rules change. We design for those changes, with monitoring, cost controls, evaluation and a clear response when something breaks.

You should know whether it is working, when it is not, why it failed and what needs to change.

Signals → Context → Decision → Action → Outcome → Evaluate & improve ↺

The experience behind the approach

16 years learning the ingredients.
A new way of combining them.

Our marketing engineering approach is new. Our experience in marketing, analytics, measurement and engineering is not.

These published references show the disciplines and people behind it.

Customer value & measurement

What can you afford to spend on a customer?

For Kiwi.com, we worked on customer value and profitable acquisition, incrementality experiments and marketing orchestration, alongside its internal teams.

Read the published story
Data warehouse & semantic layer

Reliable data foundations for people and AI.

For Mixit, we built the data warehouse in BigQuery from scratch, with a semantic layer, a knowledge catalogue and instructions for agents. We keep strengthening the foundations and supporting measurement, reporting and attribution.

Read the story
Product data & B2B automation

Turn product knowledge into a working system.

For Royal Canin, we built a product database, B2B ordering and data feeds that keep partners supplied with current product information. The published results include 35% lower call center costs and 10% more orders from customers.

Read the published story
Hands-on marketing partnership

Know the business. Stay for the long run.

Our work with Snowboard Zezula spans digital marketing, analytics and international expansion. Practical operating experience, built over a long-term relationship.

Read the published story
Adam is a leading expert and also a trustworthy, organized discussion partner. I highly value the cooperation with him.
Lucie PezlarováHead of Analytics, Kiwi.com · published testimonial

Your system. The right home for it.

We can run it.
We can teach your team to run it.

Both are good outcomes. Choose the operating model that fits your business.

Managed by igloonet

A long-term engineering partner.

We build, host or run, maintain and support the system. Together, we keep improving it as your business changes.

You own the business decisions. We look after the system.
Built inside your business

A capability that becomes yours.

We build in your infrastructure, transfer the code, document the system and train your technical team to run and extend it.

Build together. Grow your own marketing engineering capability.

What is underneath

One outcome.
The capabilities to get you there.

We bring these together around the problem. And if one part needs fixing first, we can start there.

Data foundations

Give the system a reliable picture of your business.

BigQuery · CRM & ERP · integrations

Measurement & experimentation

Find out whether a better system creates a better outcome.

Incrementality · experiments · attribution

Marketing analytics

Connect marketing decisions to customer and business value.

LTV · profitability · customer quality

AI & engineering

Turn the useful idea into something your team can rely on.

Software · agents · guardrails · monitoring

A little curiosity goes a long way

Thinking, tinkering
and useful questions.

Our working principles, experiments and small tools. A place for the ideas worth exploring.

Step inside Thinking / Labs

Let’s find the useful part

What is worth
engineering
in your business?

Start with a free conversation with Adam. Bring a business problem, a hunch or a workflow that ought to work better.

We’ll work out whether custom engineering makes sense. If it does, the first paid step is an audit and working proof of concept.

Free discovery. By submitting, you agree to the processing of personal data.

Adam Šilhan

Adam Šilhan

Marketing engineering · discovery & delivery

adam@igloonet.cz +420 771 229 209Adam on LinkedIn Your first conversation is with the person
who helps shape the work.