More data has not meant more certainty. That was the uncomfortable premise of a September session at the MarTech Conference titled “The data trust crisis: Why your customer data is getting worse” — a discussion that will feel familiar to anyone who has stared at a CRM dashboard and quietly wondered how much of it is still true.
The panel featured Ana Mourão, founder and author of the Experimental Marketer Framework; Ryan Warren, chief CRM officer at Razorfish; and Zack Wenthe, director of product marketing and customer data evangelist at Tealium, with Craig Howard moderating. Their shared diagnosis: tracking changes, shifting consumer behaviour and simply ageing records are eroding data quality faster than most teams are repairing it.
Permission first, purpose second
The panel’s starting point was that the most valuable data is zero- and first-party information customers knowingly hand over — through explicit sign-ups or through digital behaviour on your own properties. Everything else is increasingly fragile.
The bigger trap is hoarding. Storing every conceivable attribute “just in case” creates operational overhead without an activation path, and those unused fields quietly rot. The panel’s advice was to concentrate on a small set of core attributes that actually drive decisions, rather than maintaining a museum of customer trivia.
Your own campaigns may be degrading your database
This was the most counter-intuitive point, and the one Indian growth teams under monthly targets should read twice. When performance dips, the reflex is to push more messages across more channels. The panel described why that becomes self-defeating:
- Over-messaging causes audience fatigue and pushes open rates down.
- Disengaged contacts stop generating behavioural events.
- With no fresh interaction signals, profiles decay faster across the entire stack.
No amount of platform tuning fixes an audience that has stopped listening. Aligning send frequency with genuine audience intent, the panel argued, is a data-hygiene decision as much as a creative one.
AI raises the cost of bad data
As teams hand campaign routing and decisioning to AI agents, weak inputs stop being a reporting nuisance and become a live financial risk. Algorithms fed unfiltered data make real-time targeting mistakes, and teams then burn hours trying to reverse-engineer what went wrong.
The panel outlined three types of context automation needs before it can be trusted:
- Identity: who is this user or account, really?
- Engagement history: what verified actions have they actually taken?
- Business parameters: what outcome is this workflow supposed to deliver?
The upside is real. Traditional automation forces prospects down rigid, linear journeys; AI-led models can react to live behaviour — but only when anchored to clean first-party inputs.
Don’t wait for perfect data
A useful antidote to analysis paralysis: waiting for a pristine database before launching anything just kills momentum. The panel recommended building targeted proofs of concept using the fields you already have. Small-scale wins deliver business outcomes despite known gaps, and they build the internal case for funding wider data collection later.
Audit at the profile level, not the dashboard level
Aggregate reports hide the problem. The panel pushed marketers to open individual customer records and interrogate them:
- Contact-level audits: compare what real customers say against what their stored profile claims.
- Source tracing: establish where each field came from, who owns it and when it was last refreshed.
- Ongoing governance: cross-functional rules for data entry, routine profile hygiene and activation routing.
Without governance, a cleaned database re-clutters within months. Quality is an operating discipline, not a project with an end date.
How to get the budget: talk in money
Data governance rarely wins funding when it is framed as hygiene. The panel suggested translating it into downstream cost — engineering hours and compute burned repeatedly cleaning bad records; ad spend wasted retargeting people who already bought because profile syncs lag by days; and ops teams manually re-running models broken by bad formats.
Document those inefficiencies and data trust stops being an IT chore. It becomes what the panel positioned it as: the foundation for resilient, genuinely customer-centric marketing.
Source: MarTech
