Hundreds of millionsof events a day — the scale of pipeline I own the roadmap for
~€1Mannual development budget under my roadmap
7releases from my current role — all linked on this page
About
I run the roadmap for an engineering team at Awin, an affiliate network that large retail and fashion brands run their partner programmes on. The role is customer-facing — from quarterly business reviews with some of the platform's biggest brands to RFP pitches for new ones.
Before Awin I spent two years self-employed in Spain: content portals of my own, built up and run profitably on affiliate commissions and advertising — the publisher side of the very business whose platform side I own today. Alongside them I advised a handful of clients on product development. Before that I was Head of Additional Digital Brands at AIDA Cruises, and earlier I co-founded a marketplace that was acquired in 2019.
I started out writing PHP, and it still shows — when a question about business logic comes up, I open the repository instead of forwarding it to someone else. It is also why I write discovery documents a team can build from directly.
I have worked fully remote since 2020, on Spanish and German contracts, in teams spread across several countries. What interests me most are products where measurement is the hard part — attribution, tracking, data quality: the places where a wrong number costs more than a missing feature.
What I'm working on
An event pipeline on AWS, carrying hundreds of millions of events a day on peaks like Cyber Weekend.
Awin's AI visibility solution — identifying the volume publishers drive through AI assistants, and surfacing it where advertisers make decisions.
Three decisions
Three situations from the last few years where the right direction was not obvious. Pick what you would have done — there is no trick, all three options are defensible. We decided these as a team; what follows is the case I argued for and what came of it.
Decision 1 of 3
What last click cannot express
Last click is the standard in affiliate marketing: whoever delivers the final click before the purchase earns the commission. It is simple, predictable and easy for everyone involved to follow. What it cannot express are campaigns where the decision to buy is shaped much earlier — through content, reviews and recommendations. Advertisers wanted room for that, without making their programme harder for partners to read.
What we did
We built the Attribution Manager as a capability an advertiser switches on per programme: custom rules, earlier touchpoints, defined time windows. The mechanics were not the hard part — legibility was. An attribution model is only useful if everyone involved can follow it, so it is stated per programme: everyone involved can see which model applies. Option A leaves advertisers with the default, which suits many programmes but not content-driven ones. Option C makes the alternatives visible without letting anyone act on them; it followed later as a useful complement.
Outcome Advertisers can weight publishers whose contribution sits earlier in the journey — content, reviews, recommendations — rather than only the last click before purchase. It also became one of the building blocks for the US platform migration.
Social apps open links in their own in-app browser. Cookies live in a sandbox there: they do not exist in the device's default browser, and they are not long-lived either. Anyone who buys later — in the evening, in their usual browser — arrives without the context of their click. That is an industry-wide mechanic. The question was how to solve it without creating work for thousands of partners.
What we did
We built the breakout page and rolled it out as a network upgrade: the visitor moves into the default browser before the journey starts — where cookies are at home and persist. No publisher touched a link, no advertiser integrated anything. Option A shifts the work onto thousands of partners and never reaches the long tail; option B models the gap instead of closing it.
Outcome From hackathon prototype to global rollout, live across social and reward platforms — a higher conversion rate for publishers, without a single change on the partner side.
Awin's tracking is standardised — setting it up on the advertiser side is still its own undertaking: the integration team guides it, and on the customer side engineers have to find time for it. For enterprise setups that path is exactly right, and it stays. But a large share of advertisers run on standard shop platforms, and there the effort was out of proportion.
What we did
We grew the platform coverage: plug-and-play plugins for common shop systems, Salesforce Commerce Cloud among them. For the majority, integration becomes an install — while the custom route remains the right one for enterprise setups with requirements of their own. Option A does not address the actual constraint, which is time on the customer side; option C mainly helps those who built their shop themselves.
Outcome Advertisers on standard platforms go live without an integration project — and the plugin coverage sells on its own: many deliberately pick a network that supports their platform plug and play.
Where a feature has been announced publicly, the source is linked. Numbers on impact I would rather discuss than publish here.
Business caseand prioritisation for a cross-team programme of backend optimisations — no new user-facing surface, purely conversion and attribution mechanics. Happy to talk through what it returned.
Breakout pagethat moves visitors out of a social app's in-app browser and into their real one before the journey starts. Otherwise the tracking cookie stays in that sandbox and the later purchase never returns to it — the sales happen, they just are not counted. Awin spring release 2025
Deferred clickmeasurement that keeps commissions attributable when a browser strips the click parameters — Safari's link tracking prevention was the trigger. Measured on the advertiser's landing page as a second path, not a replacement. Awin spring release 2024
Card-linked offersprivacy browsers and consent banners leave classic tracking with gaps — a bank sees every transaction. The Proof of Purchase API turns that into a scalable base for card-linked-offer campaigns on Awin: full conversion coverage instead of the workarounds that never scaled. Awin API documentation
Attribution controlfor advertisers beyond a fixed last-click model — custom rules, earlier touchpoints and defined time windows, so budget follows contribution rather than position. Also one of the building blocks for the US platform migration. Awin autumn release 2024
AI in productionLLM-based extraction inside a data pipeline — with evaluation criteria, confidence thresholds and a defined fallback. The effort is not in the model; it is in deciding when a result may be trusted, and what happens when it may not.
US migrationof a US platform onto the global one — a large project with several workstreams. Mine: making sure customers could keep using key attribution and commission features without interruption.
E-commerce pluginsintroduced for common shop platforms, Salesforce Commerce Cloud among them. They turn technical onboarding into an install — and they sell, because many advertisers deliberately pick a network that supports their platform plug and play. Awin summer release 2024
Transparency APIexposing which tracking methods each advertiser actually ran over the last 15 days, so publishers can judge a partner on tracking quality before they promote it. Awin winter release 2026
Data accessmade the pipeline's data usable beyond the team: schemas documented, hand-over points provided, and — together with data engineering, who own the access layer — made sure the numbers land where engineering, commercial and product can work with them.
Exitco-founded a cruise marketplace and led it as managing director from zero to profitability; acquired by a leading tourism group in 2019. Cruzero, our cruise metasearch app, came out of the same years. fvw on our cruise metasearch, 2015
Call-back channelin cruise direct sales: a form replaced with a call-back wired by API into the call centre's workforce system. A cruise is a considered purchase — a conversation beats a form.
How I work
Problem first. If I cannot state the problem in a paragraph and say who has it, I am not ready to ask engineers for anything.
Success gets defined up front. A metric, a leading indicator, and the condition under which we stop. Deciding afterwards what counted as a win is how teams talk themselves into shipping things that did not matter.
AI is leverage, not a demo. I use it daily for research, analysis and prototyping — and I have shipped it, including the unglamorous parts: evaluation criteria, thresholds, and what happens when the model is wrong.
I write to think. A decision framed on one page beats a meeting. I put the thinking down early so people can disagree with it before we spend engineering time.
Where I've been
2022 – nowAwin Senior Product Manager
2020 – 2022Self-employed, Spain own affiliate-monetised content portals; product consulting on the side
2019 – 2020AIDA Cruises Head of Additional Digital Brands, direct-sales portals
2010 – 2012AIDA Cruises digital product management
2007 – 2010HSBA Hamburg B.A. Business Administration, dual programme
2003 – 2006Lieske Elektronik PHP developer
Languages. German (native) · English (fluent) · Spanish (conversational, B1–B2)
Side projects
I still run Tech Junkies, the last of the content portals from my self-employed years. Building an audience from nothing keeps me honest about how hard distribution is, and about how little a feature is worth if nobody finds it.
Get in touch
Send a message
Email works just as well: patrick@woessner.me. This form only saves you a context switch. I usually reply within a day.
I am not on LinkedIn — this page is my profile.
Request my CV
Fill this in and I will send you the PDF, usually the same day. I would rather know who is reading it than leave it floating around as a public link.
