Table of Contents
Every enterprise personalization deck opens with a slide labeled “360-degree customer view.” Almost none of the brands presenting them actually have one. The gap between the picture on the slide and the reality of the data infrastructure is where most personalization programs quietly fail.
The data on the gap is stark. Research from the CDP Institute finds that 68% of brands report they are still struggling with siloed data preventing a unified customer view. Salesforce’s connectivity research shows that UK enterprises use, on average, 796 different applications, and only 33 percent are integrated. Nearly half of marketers say some part of their customer data is siloed and difficult to access. Ninety-seven percent of executives say data silos are actively hurting the business.
These are not new numbers. They have been broadly consistent for a decade. What has changed is the confidence with which brands claim, in vendor decks and quarterly reviews, to have solved the problem. The claims have gotten better. The infrastructure has not.
The phrase “we have the data” collapses at least four distinct stages of data maturity, and the dropout between them is enormous.
Data can be present collected somewhere, perhaps in an analytics tool or an ad platform or a support system, without being connected to a customer identity. Anonymous session data, unattributed email opens, uncorrelated support tickets. Present, but strandable.
Data can be connected and attached to a customer profile without being unified across systems. The customer’s Shopify identity, Klaviyo identity, and support-desk identity may all exist and all be linked to the same person but sit in three separate stores that don’t talk to each other. Connected within each silo, fragmented across the picture.
Data can be unified and pulled into a single view without being usable by the systems that need to act on it. A CDP that ingests everything and exposes almost none of it in real time to the personalization surfaces of the store is a common architectural failure. Usable data is available at the moment of decision, not just in a batch export.
Data can be usable and still not be acted on, either because the personalization logic wasn’t designed around it or because the team never got around to building the surface that would use it. This is the most frustrating category. The signal exists. The infrastructure exists. Nothing changes for the customer.
Most brands claiming a 360-degree view are, on honest audit, operating at 20 to 40 percent coverage on the “acted on” stage. This is the number that matters. Everything upstream of it is preparation.
A personalization architecture that treats every data source as either available or unavailable fails silently in production. A more useful model treats every source as being in one of six states at any given time: not available (the source simply doesn’t exist for this customer or category), declined (the customer opted out), pending authorization (auth is expected but not yet granted), connected (working), connected but degraded (partial data, rate-limited, out of date), or disconnected (broken).
This is not a plumbing detail. It changes what personalization can honestly do. A personalization decision made with the assumption that Klaviyo email history is available will produce garbage when Klaviyo is in a degraded state and the last thirty days of sends are missing. A “recommended for you” surface that depends on browsing history will feel eerie when the source is fully connected and lifeless when it isn’t. The customer doesn’t see the state of the data source. They see the state of the experience.
The discipline is to design every personalization surface with an explicit answer to the question: what does this surface do when the input data is in each of these six states? Brands that skip this end up with personalization that works well in the demo, works occasionally in production, and gets quietly turned off after six months when the operator can’t explain why the outputs are strange.
Consider a mid-market jewelry store on Shopify. The typical inventory of customer data sources looks something like this. Shopify itself: fully connected, mostly usable. Klaviyo email: connected, but subscription and consent gaps mean only 40 to 60 percent of the customer base is in a state where email history is complete. Product reviews: present, rarely tied to customer identity in a way personalization can use. Support tickets and WhatsApp conversations: increasingly common as a customer channel, almost never integrated into the personalization stack. Instagram DMs, where much high-consideration jewelry discovery actually happens essentially never integrated. Search terms typed into the on-site search bar: often not persisted at all. Wishlist behavior: patchy. Returns and their reasons: usually captured, rarely fed back into product surface logic.
The honest audit of this store, the kind that maps every source to its actual capability state, tends to find that meaningful, actionable, real-time first-party signal exists for perhaps 30 percent of the identified customer base and covers perhaps 40 percent of the behaviors the store cares about. The remaining coverage is either aspirational or fictional.
This is not a failure. It is the baseline. What separates brands that build defensible personalization from ones that don’t is not whether they have full coverage. It is whether they know what coverage they actually have and whether they design personalization that degrades gracefully as coverage varies rather than pretending to a completeness they don’t possess.
The strategic move for most brands, right now, is not to buy another data platform. It is to audit the one they have. Not the vendor-pitch version of the audit. The engineering version. Every source, every state, every dropout, every place where a personalization surface is silently receiving degraded input.
The brands that do this discover something uncomfortable: their personalization has been running on a much thinner data spine than they thought. They also discover, more usefully, that the thin spine they actually have is enough to do specific, honest, high-value work if they design for what they have rather than what the deck says they have.
The brand that admits it has 30 percent coverage and designs the customer experience around that 30 percent will out-personalize the brand that claims 360-degree coverage and generates the other 70 percent from thin air. The next article in this series is about what that generation actually looks like and why disciplined observation beats the inventive generation every time.
Fullestop is a custom web and app development company that builds commerce infrastructure beyond out-of-the-box software. From custom Shopify ecosystems to API-first data integrations, our team helps brands close the gap between the personalization they claim and the personalization their data can actually support. We’ve partnered with global brands, including Sony Pictures Networks, Volkswagen, and Adidas, to build ecommerce experiences that are grounded in real, usable signal rather than assumption.
For the past few years, Small and Medium-Sized Enterprises (SMEs) in the US have treated AI like an experimental playground—a place to "see what happens."...
In today's rapidly evolving digital landscape, artificial intelligence is no longer a futuristic concept, but a tangible force reshaping industry worldwide. At the forefront of...
Let’s be honest: relying solely on third-party aggregators is a leaky bucket for your revenue. While platforms like UberEats, DoorDash, or Grubhub are excellent for...
