When Data Flows, Rubbish Floats Downstream
Enterprise architectures resemble vast river systems: many sources feeding into a single big flow of information. Leads bubble up from fresh springs, while opportunities and orders join from deeper underground channels. Legacy data seeps in like meltwater from distant glaciers, slow yet persistent, carrying sediment from systems long forgotten. Cases, tickets and issues surge downstream after heavy rainfall: bursts of customer activity that swell the current. And through it all, the data keeps flowing and brings value to applications, platforms and people.
But with any flowing system, rubbish may enter the stream at any point. Pollutants may seep from the melting glaciers: remnants of legacy systems carrying outdated formats or incompatible processes. The rainfall may wash in loose debris: sticks, stones and leaves from unfiltered human inputs that clog drains. Tributaries may bring in cloudy water from poorly governed integrations, where field mappings or data ownership are unclear. Even the once-pure springs can become tainted when validation is bypassed or quick fixes are poured in without filtration.
All this rubbish does not stay in one place. It moves with the current, through your teams and departments. As it accumulates, it pollutes the data it encounters. Over time, this build-up may lead to blockages: integration errors or failed synchronisations, which contaminates any data it touches.
How to regulate the flow?
Engineering effort is needed to build pumps and pipes that connect seamlessly to every source feeding into the system, to install grates and filters that catch duplicates and to construct dams and locks that regulate the flow of data. ‘Cleaning plants’, governance frameworks that scrub invisible pollutants from the current, are essential to prevent corruption and loss before they travel downstream. By installing measuring stations and monitoring the flow, emerging blockages can be detected before they spread. Without these safeguards, even the clearest streams eventually turn muddy.
Ultimately, all rivers converge somewhere. In enterprise IT, that destination is the data lake. This is where the countless streams of operational data settle, forming the foundation for analytics and insight. But without circulation and care, even a lake becomes stagnant. A data lake without governance becomes a data swamp: wide, deep and murky, where potential insights sink beneath layers of unmanaged debris.
Governance isn’t just a technical framework, it’s stewardship. Every team that touches data plays a part in keeping the current alive and the water clear. Because once trust in your data is lost downstream, no filter can bring it back.
Five starting points that help keep your company’s data flow clean and strong
- Establish a baseline governance framework. Start by describing your data architecture: which systems hold which data, how do systems connect, what information flows between them. This creates a shared understanding of the landscape and provides the foundation for every following improvement. Identify different data domains and their criticality, which allows prioritisation of next steps by focussing on key areas.
- Define data quality and duplication metrics. Agree on how you will measure duplication, cleanliness and completeness. Define what “good” and “good enough” looks like. These definitions prevent confusion later and may guide your initial cleanup work.
- Set realistic quality targets and make them visible. A 0% duplication rate is unrealistic and often unnecessary. Start by working toward something achievable, such as reducing account duplication to 10%. Also, not all data is equal. Identify and classify data that are more important to clean and focus efforts there. Publish the metrics in visible dashboards and schedule regular check-ins (monthly, quarterly, or tied to release cycles). This makes data quality a continuous discipline rather than a one-off project.
- Identify and implement quick fixes. Add data validation rules, set required fields, enable duplicate detection and improve error logging. Focus first on systems and integrations that generate the most errors, as these are often the highest-value areas to address. Errors signal active business usage and failing business processes. And don’t forget the human aspect: by refreshing your end user’s knowledge you can avoid messy data to begin with.
- Assign clear ownership. Define who is responsible for each data domain, each process and each metric. When deterioration occurs or changes are needed, ownership ensures accountability and prevents issues from falling between teams. Make sure there is a clear system onboarding process so new systems and integrations are held to the same standards and have their own metrics.
Ready to talk about it?
The above points should give you guidance on how to start. We understand this can still be a daunting task. At Gen25, we help organisations keep their rivers clear: connecting systems, filtering inputs and ensuring that what flows downstream can be trusted to deliver real insight.
Curious to hear how we could help your organisation more specifically? We would be glad to further discuss it with you!
