Most Shopify Plus stores run four to six tools that all touch the same shopper. Klaviyo sends the welcome series. Postscript or Attentive blasts the SMS list. A win-back tool fires ninety days after last purchase. A browse-abandonment tool watches product pages. Each one runs on its own schedule, its own data, its own voice.
Nobody wired them together.
Why the stack grew this way
Ten years ago, no single platform handled cart recovery, SMS, browse abandonment, and win-back in one conversational layer. Vendors specialized. One company built email flows. Another built SMS blast infrastructure. A third built retention tools for existing customers. Stacking best-in-class point tools was a reasonable move given what existed at the time.
The instinct held up because each tool did its job. Klaviyo's flows are good at what they do. Postscript's blast infrastructure sends at scale. Build quality was never the issue. Stacking single-purpose tools was never going to add up to something a shopper experiences as one brand paying attention to her, no matter how good each individual piece is.
Agencies running fifteen or twenty stores feel this first. Every new client means re-learning a new combination of the same six categories of tool, each configured its own way, each holding its own slice of that store's shopper data. The stack isn't just fragmented within one store. Each store an agency touches fragments in its own way.
What breaks as volume grows
A shopper texts back a sizing question on an abandoned-cart flow. Three weeks later, the same shopper gets a browse-abandonment text about a different product, no memory anywhere that she'd already asked something on the last one. Two tools, two conversations, zero continuity.
Data splits the same way. Klaviyo holds one view of the shopper. The SMS platform holds another. The win-back tool works off whatever CRM export ran last. No single system holds the full picture of what a shopper asked, what she was told, and what she bought.
Tone splits too. Email carries whatever voice got written into the flow two years ago. SMS blasts carry another, shorter and pushier, because open rates run high and marketers assume high open rates mean shoppers want short copy. A shopper moving between channels ends up talking to what feels like three different companies depending on which tool happens to be running.
Picture a shopper who abandons a cart on a Tuesday, texts back asking about return policy, and gets an answer. She doesn't buy that week, but she comes back browsing the same category a month later. The browse-abandonment tool has no record that the return-policy question ever got asked, so it sends its standard script, the one written for a shopper starting from zero. She already asked and already got an answer. The tool just doesn't know that, because nothing connects what happened in the first tool to what fires in the second one.
None of this shows up as a line item. It shows up as a slow leak: replies that go nowhere, discounts issued because nobody could answer the actual question, a shopper who stops responding to texts because the last three didn't listen to anything she said. A brand can be running all four tools well by each tool's own dashboard and still be losing shoppers to a gap none of those dashboards were built to measure.
What infrastructure means instead of another tool
The fix isn't a seventh tool stacked on top of six. It's one conversational layer running underneath every moment: cart abandonment, browse abandonment, win-back, whatever comes after that. The layer is designed to hold the shopper's context across all of them, so a question asked in one moment doesn't disappear by the time the next message fires.
That's the line between a point tool and infrastructure. A point tool solves one moment. Infrastructure is the layer every moment runs through.
CART runs this way today, with one engine live: Abandoned Cart Recovery. Browse Abandonment Recovery and Win-Back and Reactivation come next, built on the same conversational core so context is designed to carry forward as each new engine comes online, instead of every new tool starting the shopper over from zero.
It runs as a managed service. Someone reviews conversations, tunes responses, adjusts for what a specific store's shoppers ask, and does that same tuning work across every store an agency manages instead of leaving each one to configure its own version from scratch. That's a different model than installing an app and hoping the defaults fit your catalog.
Not a replacement, an addition
None of this means ripping out Klaviyo. Email still owns the moments SMS can't reach: welcome series, non-cart campaigns, shoppers who never gave you a phone number. Running alongside existing flows like Klaviyo, CART recovers an extra 10-15% of lost revenue, the slice a scheduled flow was never built to catch. The existing stack keeps doing what it already does well, and the layer underneath it changes.
What's built right now
One engine is live today, with two more on the way. That's the current state of it, not a finished platform described as though it already covers every moment a shopper shows up. The point isn't that every funnel stage already runs through one infrastructure layer. The point is the direction: fewer disconnected tools, one system designed to carry the shopper's context across every moment she shows up, whether that moment is a cart, a browse session, or a win-back message down the line.
That's worth naming while it's still in progress, not after it's done.
Curious what CART can catch that your current stack can't?
Run a Revenue Proof →