We build the data foundation your marketing needs
No warehouse yet? That is the normal starting point, not a blocker. We build it and we run it.
30 minutes · Questions, not a pitch
A marketing data foundation enables both reporting and activation
When every source lands in one modelled warehouse, the same trusted numbers do double duty. Your reporting reads from it, and the tools that spend your budget are fed from it. You model the customer journey once, then use it everywhere.
Measure ROAS
Ad platform numbers joined to real revenue. Biggest gap for subscription, where the money arrives over months.
Signal engineering
Automated buying like PMax leaves the signal as the main thing you control. Predicted values need somewhere to run.
Stitch web-to-app
Journey starts on web, conversion happens in the app. Identity breaks at the join and one user becomes two.
Measure long sales cycles
The information lives in the CRM. ROAS stays wrong until closed revenue connects back to paid.
Run MMM
MMM has real data requirements. We build what it needs so you can start your MMM project.
Query it over MCP
Your team asks the marketing data questions in plain language, from Slack or an AI assistant, instead of waiting on a dashboard.
Triangulated attribution
Attribution modelling, incrementality tests and MMM cross-checked against each other. Sits on top of this, not instead of it.
This is only possible by centralizing all your relevant data sources in one place
Propel ingests and stitches these sources together to create a unified model for the customer journey.
There are no pre-requirements to working with Propel
Those are the things we build.
Teams put this off because they think the prerequisites are on them. A warehouse. A data engineer. A CDP someone configured properly. UTMs that make sense.
We build toward a goal you name first, not a warehouse for its own sake.
If you already have a warehouse and your marketing data has never reached it, that is also a normal starting point. We usually build the marketing warehouse alongside your existing one rather than inside it.
We run it as a retainer. We build the foundation, then own and run the pipelines so your team can use the data without maintaining it. If you ever want to take it in-house, everything transfers to you.
How we get from your sources to a foundation you can use
Retainers start with a 12-month period. The warehouse itself is set up inside the first two to three months, because first we have to know what to build.
Marketing workshop
What marketing is trying to achieve, and what decisions are waiting. This sets what the data has to answer, and nothing gets built before it.
Data workshop
With your data owners. What exists, where it lives, what state it is in, and who grants access to each system.
Define the sources to ingest
A prioritised list, ordered by what the workshops said you need answered rather than by what is easiest to connect.
First ingestion
The sources land in the warehouse. This is also the step that tells you what is genuinely broken upstream, which no audit predicts from outside.
Fix what the data exposed
UTMs that do not describe themselves, survey questions missing channels, form fields that lose the answer. These get fixed at the source.
Second ingestion and modelling
Re-ingest with the fixes in place, then build the modelled marketing tables in dbt, refined over the engagement rather than declared finished.
We own and run the pipelines throughout, so nobody on your side inherits maintenance work they did not ask for.
What happens on a call
The first call is a diagnostic, not a pitch. One question:
What is marketing trying to achieve?
Both the decisions waiting on better numbers, and what you want to do in the campaigns: bid on predicted value, feed sharper signals, run MMM.
One of two things happens after it. If we do not think you need this, you get an email saying so. If we do, we book a presentation within 7 working days on what you need, why, and what it would cost.
FAQs
Do I really need a data warehouse for marketing?
Sometimes yes, sometimes no, and that is exactly what the diagnostic call is for. A single Google channel with an in-session conversion can often get by without one. Predicted-value bidding, MMM, and revenue that only closes in your CRM cannot. We will tell you which case you are in, even when the honest answer is that you do not need us yet.
Is the foundation the whole project?
No, and we do not build one on spec. The foundation is always built toward a goal you name first, whether that is triangulated attribution, signal engineering, MMM or something else.
We have no data engineer. Can we still do this?
Yes, and that is the usual case. We build and run the whole foundation, so you do not need to hire a data engineer to get started. You bring the marketing questions and access to your systems, we bring the engineering.
Do we need to replace our current stack?
No. We do not replace your data sources, so your CRM and your event analytics stay exactly as they are. We usually build the marketing data warehouse alongside your existing one rather than inside it, so nothing you already rely on has to move.
Can you work with the warehouse we already have?
Usually we create a separate marketing warehouse, because it keeps the marketing models clean and easy to hand back later. If you would rather we build inside the warehouse you already run, we look at that case by case and tell you honestly whether it is the better call.
Can you work with my data team in building this?
Yes. We will need permissions from your data team and from the owners of each source, so they are involved either way. Building the warehouse and the models jointly does not usually make the project faster, but if you have reasons to set it up that way, we can accommodate it.
How long does it take?
Most of our retainers start with a 12-month period. The warehouse setup is done inside the first two to three months, because first we have to know what to build. From there we keep refining the models and the data we capture for the rest of the engagement.
Do we have to commit to a 12-month retainer?
Twelve months is our default because of how the work is shaped. The warehouse itself goes in during the first two to three months, but most of the value comes after that, when we refine the models against real decisions and add sources as new questions come up. A shorter engagement tends to leave you with a foundation that was never tuned to how your team actually uses it. The retainer is not the only way to work with us, though. If you have a data team that will take the pipelines in-house once they are built, or you want to structure this differently for another reason, raise it on the call and we will lay out the options.
Can you set up MCP access to the data?
Yes. Once the modelled tables exist, we can put an MCP server on top of them, so your team can ask questions of the marketing data in plain language, from Slack or an AI assistant, instead of waiting on a dashboard. It answers questions about the data. It does not replace deciding what the numbers mean.
Can you set up offline conversions and CAPI events?
Modern ad platforms buy on the signal you send them, so a conversion that only lives in your CRM, or that happens offline, is invisible to them until you pass it back. Once the foundation is in place, we build and run the server-side pipelines that do this: CAPI for Meta and the equivalents for Google and other platforms, sending each conversion back with the right value attached. That is what lets the platforms optimise toward your actual customers rather than the cheapest lead.
Can you provide different attribution models?
No single attribution model is correct, so we do not hand you one and call it done. With the modelled data in one place, we can build several views, from first and last touch to position-based and data-driven, and show you where they agree and disagree, which is usually more honest than one number. Attribution modelling is also just one of the lenses we triangulate, alongside incrementality tests and MMM, rather than the whole answer. Which models are worth building depends on your channel mix, and we work that out with you.
Can you stitch web-to-app journeys and measure long sales cycles?
Both are common reasons teams come to us, and both come out of the same foundation. When a journey starts on the web and converts in the app, identity usually breaks at the join and one person gets counted as two, which quietly inflates your numbers; we stitch that identity back together in the warehouse. Long sales cycles are the opposite problem: revenue closes in the CRM weeks or months after the click, so ROAS looks wrong until you connect the two. Once the foundation exists, both are modelling work rather than new plumbing.
What does a warehouse cost to run?
On our retainers, the storage and compute sit with us, so it is part of what you already pay rather than a separate bill to manage. Marketing data is small next to product or event data, so the warehouse itself is rarely the expensive part. If you would rather run it in your own cloud account, we can set it up that way and pass the costs straight to you.
Who maintains this after the project is done?
Most of our projects are structured as retainers, so we own the pipelines and the ongoing changes for as long as we work together. Nobody on your side inherits maintenance they did not ask for. If you would rather structure it differently, we can talk that through.
What happens after the retainer is done?
Everything belongs to you. We use the last month as a handover: we onboard your data team, transfer the data to a warehouse you own, and share all the dbt models. If you ever want to take it in-house, nothing is locked to us.
