Implementation & Partners

A first read in a day. A defensible answer in a week.

We begin with a briefing, not a procurement cycle. Bring two years of weekly spend across three channels — the minimum the math needs to separate signal from noise — and we will tell you honestly whether it holds on your data.

Inside the first business day you will see where your spend is actually landing. The week is for the part no model can know on its own: the hurricane that closed retail for ten days, the competitor who went dark, the price change nobody logged. Those show up in the data as effects the engine can measure but not name, and naming them is a conversation. Most custom marketing-mix engagements deliver their first read in months.

Two ways to work with us

Bring your own team, or borrow ours.

A team gathered around a table covered in printed charts and reports.

Self-directed

For teams with their own data or analytics group. Full access to Insights: load your own data, train and review models, run optimizations, and manage permissions across your team.

Managed, through Clarity i2

For teams that want the certainty without adding headcount. Our implementation partner loads your data, runs the modeling cycle, and delivers board-ready recommendations on your behalf.

clarityi2.com ↗

What each side brings

The modelling engine is ours. The expertise is theirs.

Insights is the non-linear modelling and optimisation engine behind every recommendation a managed engagement produces — it learns how each placement moves revenue, measures how channels move each other, and ranks every dollar by what it actually produced. What turns that into a plan a board will act on is Clarity i2's own judgment: thirty-plus years of brand-side and agency analytics experience, reading the data, setting the right constraints, and knowing which recommendation a client's real business can execute. The platform is only as good as what goes into it.

How the engine respects your constraints

The optimizer works inside your reality, not around it.

A recommendation that ignores a signed contract or a production limit isn't useful, no matter how strong the math is. Every optimization runs inside boundaries you set first.

Commitments

Lock a channel or campaign at its current spend. Upfront contracts and sponsorships stay untouched; the engine optimizes everything around them.

Saturation

Cap how far any channel can grow. Keeps a high-ROI recommendation from running into real inventory limits.

Operational bandwidth

Set a ceiling tied to what your team can actually produce, so a channel never gets a recommendation your creative pipeline can’t support.

Strategic weighting

Protect a channel’s floor for brand-equity reasons, even when its short-term ROI is lagging.

Budget silos

Keep a department’s reallocation inside its own budget, so one team’s optimization never reaches into another’s.

Illustrative. Four of these five rows wanted to move further than they were allowed to. The engine returns the best plan inside your limits, not the best plan in the abstract.

Where the data can live

Runs wherever your data has to stay.

Insights is written from the ground up in native C++, with no heavy framework dependency chain. It compiles for Windows, Linux, macOS, and IBM z/OS, and deploys to AWS, Azure, entirely on-premise, or inside a European data center for teams with residency requirements. Privacy-by-design isn't just a claim here — it shows up in how the stack itself is built.

A person standing before racks of servers in a data centre. — On your infrastructure or ours, in your jurisdiction — the engine has no opinion about where it runs, and no need to phone home.

Why not build this internally

Companies with the data rarely have the years it took to build the method.

Adstock, non-linear response curves, and the constraint logic above took years to get right. An internal tool also inherits internal politics — marketing wants it to look good, finance wants it to look lean. An external, structurally independent auditor can't have that problem, because it doesn't answer to either side.

It has to survive fall planning, every year

An internal build competes for budget annually against everything else engineering wants to ship. Lose that argument once and the roadmap stops — usually right after the first version works well enough to be ignored. The measurement platform your CFO depends on should not be re-justified every autumn.

The roadmap keeps moving without you funding it

Features get built here whether or not any one client asked for them, and every client gets them. An internal tool only ever receives the attention its own budget line survives to pay for.

It is a compiled engine, not a research script

Most internal analytics starts as a notebook that worked, gets promoted to production, and accumulates dependencies until nobody will touch it. This is native C++ built as a product from the start — which is why a full multi-market reallocation returns in milliseconds instead of overnight, and why upgrading it is not a project.

Get started

Request a briefing

A strategic briefing runs about forty minutes: what your data looks like today, whether it clears the gateway, and what a first model would cover.

Prefer email? info@fractal-rock.com · Already a customer? Sign in to Insights