At a time when information flows freely and expertise has made accumulating and processing data nearly second nature, publishing that information in a significant means stays a problem. It’s, in some ways, an artwork. All too typically, information is just offloaded to purchasers, however not really delivered. Efficient information publishing is a type of storytelling, one which calls for context, consistency, and readability. It’s what transforms uncooked numbers into insights that permit folks to behave on with confidence in enterprise contexts.
Ravish Mishra, VP of engineering (information) at Angel One, addressed this important facet of knowledge analytics throughout a session titled ‘The Artwork of Information Publishing: From Noise to Narrative’ on the DES 2025 summit. Chatting with a room full of knowledge engineers and tech fanatics, Mishra navigated the historical past, challenges, and future of knowledge structure, encouraging firms to maneuver past storage in the direction of significant storytelling.
From Warehouses to Narratives
Mishra opened his speak by tracing the evolution of knowledge administration—beginning with information warehouses, progressing via lakes and lake homes, and now advancing in the direction of extra AI-integrated architectures, together with generative AI stacks.
“At Angel One and throughout India, we’re actively constructing processes on prime of GenAI and associated AI applied sciences,” he stated. He emphasised the classical ‘5 Vs’ of massive information: quantity, selection, velocity, veracity, and worth. Whereas organisations have tackled the primary three by way of fashionable architectures like information lakehouses, he identified that worth and veracity proceed to be difficult for a lot of organisations.
“What number of of you suppose your organisation is utilizing information to its full potential?” he requested the viewers. “In case your platform allows storytelling that drives enterprise selections and enhances buyer expertise, then you definitely’re heading in the right direction.”
He raised considerations over the rising quantity of knowledge, noting, “Over the previous 5 years, world information quantity has grown by 5%. As [the amount of] obtainable data rises, organising it turns into critically vital. With out efficient administration, we danger creating methods full of hard-to-interpret, low-quality information [that is] unreliable for enterprise selections.”
The Noise in Information
A recurring theme in Mishra’s session was the idea of “noise” in information—inconsistencies and errors that dilute its usefulness. He highlighted a number of key varieties: high quality noise, semantic noise, temporal noise, and structural noise.
“The true worth of knowledge is unlocked when it allows knowledgeable decision-making. Ideally, these selections needs to be encoded inside the information mannequin itself. It’s not nearly making selections, however understanding why they’re made,” he stated.
In keeping with Mishra, the true worth of knowledge doesn’t lie within the warehouse or lake; it’s within the narrative layer. He advocated for constructing dimensional fashions inside the “gold layer” of knowledge structure, which helps simpler querying and quicker interpretation. Nonetheless, even these fashions typically depend on analysts to derive that means manually.
“We must always purpose to encode selections inside information fashions,” he stated, criticising the frequent anti-pattern of treating information as an afterthought.
“Companies typically fail to plan how they’ll measure a brand new characteristic’s success throughout its design. Instrumentation is added post-release, resulting in fragmented information and missed insights.”
He additionally highlighted the inefficiency of duplicating enterprise logic throughout each the supply and analytics layers—a apply he strongly suggested towards.
Rework Information from Noise to Narrative
Mishra outlined a framework for constructing decision-driven information methods rooted in key rules: stability and immutability, fastened contracts, service-level agreements (SLAs), and semantic readability. Drawing from his expertise, he burdened the significance of immutability as soon as information is printed, making certain that downstream methods don’t alter the integrity of occasions.
“Stability means preserving the temporal context, particularly when syncing normalised tables like orders and order gadgets, which arrive at completely different speeds. Failure to take action leads to deceptive insights,” he defined.
To deal with semantic noise, he known as for a strong semantic layer that defines information that means clearly and constantly. “Good information merchandise should not solely encode details but in addition help narratives and explainable tales that reveal why selections had been made,” he stated.
This consists of enriching information fashions with diagnostic and prescriptive intelligence, constructing actionable interfaces, and protecting narratives up to date as information sources change. “Mounted contracts and SLAs are essential to make sure constant analytics and dependable insights,” he added.
Because the session concluded, Mishra reminded attendees that the last word purpose of a knowledge system isn’t simply analytics; it’s narrative. “Selections are encoded in information. Your fashions shouldn’t solely reply questions but in addition inform tales,” he stated. “That’s how we remodel information from noise to narrative.”
The publish What Each Enterprise Should Know In regards to the Peril of Information Noise appeared first on Analytics India Journal.
