Conversational Commerce: When the SMS Thread Becomes the Store

For most of the last decade, SMS was the highest-ROI megaphone a DTC brand owned. You built a list, you sent a blast, and revenue-per-send did the rest. That playbook still works — right up until it doesn’t. The brands pulling ahead now treat the text thread as something else entirely: a two-way surface where a customer can ask about fit, get a real answer, and check out without ever opening the app.

This is what “conversational commerce” actually means once you strip the buzzword off it. Not a chatbot bolted onto your Shopify theme. A thread your customer already has open, where the reply they get is specific to their order, their size, their cart — and where buying is one message away. This piece is the operator’s map of where that pays: the flows worth automating, the questions worth answering in-thread, and the discipline that keeps the whole channel from burning down.

From blast to conversation: why RPS plateaued

Revenue-per-send is the number every SMS program lives and dies on, and for years it went up because the channel was uncrowded. It isn’t anymore. Every brand your customer buys from now has their number, quiet hours are enforced, and list fatigue is real — send more and unsubscribes climb faster than incremental revenue. The blast model has a ceiling, and a lot of programs have hit it.

The way past the ceiling isn’t more sends. It’s making each send able to hold a conversation. When a back-in-stock alert can answer “does it run small?” and take the order in the same thread, you stop paying for the second and third touch it used to take to close. The economics change from cost-per-send to revenue-per-conversation, and the best-performing flows are already the two-way ones.

The high-value flows: cart, restock, replenishment, win-back

Not every message deserves a conversation. Four flows earn it because the intent is already hot and the objection is usually a single question away from resolved. These are also, not coincidentally, the flows with the strongest published benchmarks.

Those figures come from platform benchmark reports — Postscript and Klaviyo among them — so read them as vendor-published and directionally true rather than gospel. What holds across every source is the ranking: abandoned cart and back-in-stock sit at the top because the customer already wanted the thing. Abandoned cart clears a median $3.94 in revenue per send in those same 2024 reports — a number that only exists because the message catches someone mid-decision.

• Abandoned cart.The objection is usually one thing: a shipping cost, a size doubt, a “let me think.” A thread that can name the objection and answer it recovers carts a static “you left something!” never touches.
• Back-in-stock. The highest-intent trigger in the whole program. The customer asked to be told. Answer the sizing question in the same breath and the alert becomes an order.
• Replenishment.For consumables, the reorder point is predictable. A well-timed “running low?” with a one-tap reorder is close to free money — vendor benchmarks put replenishment flows around 13–14% CTR (vendor-published).
• Win-back. The lapsed customer who once loved you is cheaper to recover than a stranger is to acquire — but only if the message is a conversation, not a coupon shouted into the void. We treat that as its own discipline.

Pre-purchase questions: sizing, fit, ingredients, compatibility

Here is the revenue most brands never see because it never becomes a ticket: the customer with a real question and no fast way to ask it. Does this run small? Is it fragrance-free? Will it fit a 2019 model? Is it safe while pregnant? On the product page these questions send people to a review section, a size chart in a modal, or — most often — to a competitor who answered faster.

In a thread, that same question is a thirty-second exchange that ends in a purchase. This is the part of conversational commerce that doesn’t show up in a “deflection” metric because there was never a ticket to deflect. It shows up as conversion. An agent that can read your catalog, your size guidance, and your ingredient list answers it consistently at 11 p.m. on a Sunday, which is exactly when a human merchandiser is not at their desk.

The most expensive customer question is the one that never gets asked — because the shopper just closes the tab.

Post-purchase: the support tickets that are really revenue moments

The instinct is to file post-purchase messages under “support” and measure them by how cheaply you can make them go away. That’s a mistake. The window between “order placed” and “order arrived” is when a first-time buyer decides whether there will be a second order. Handle it well and you’ve bought a repeat customer for the price of a text.

“Where is my order” is the volume king here, and it deserves its own treatment — automating it end-to-end with live carrier data is one of the clearest wins in the whole channel, and we make the full case for it separately. The short version: two-way order-status flows have been shown to cut inbound ticket volume by roughly 14–22% (a range across vendor deployments, not a guarantee), and every one of those resolved threads is a chance to say “while you’re here” without being crass about it.

Subscription saves: the churn conversation that works

For subscription DTC, the cancel button is the whole ballgame, and most brands defend it with a static “are you sure?” screen that converts almost nobody. The customer clicking cancel usually has a specific, solvable reason: too much product piling up, a pause needed for a month, a cheaper size they didn’t know existed.

A conversation that asks why — and can actually act on the answer by pushing the next shipment, swapping the SKU, or adjusting cadence — saves subscriptions that a cancel form loses. The key is that it has to be able to dothe thing, not just offer a discount. A save offer with no action behind it reads as a hostage negotiation. A thread that says “want me to skip next month instead?” and does it reads as service.

Peak season: BFCM volume without a support meltdown

Black Friday through Cyber Monday is where a support stack either scales or publicly fails. Volume can multiply several times over in seventy-two hours, and the questions are overwhelmingly the same handful: where’s my order, can I still change it, does the promo apply, when does it ship. These are exactly the questions a conversational layer handles best, because they’re high-volume and low-ambiguity.

The teams that survive BFCM don’t out-hire the surge — the seasonal temp you onboard in November is at their least effective during the exact window that matters. They put the repetitive majority on an agent that doesn’t care whether it’s handling ten conversations or ten thousand, and they reserve their humans for the genuinely messy cases: the angry customer, the address that’s wrong, the order that split across three shipments. Elastic capacity with no hiring lead time is the entire point.

Compliance and list health: opt-out discipline as a growth strategy

None of the above survives a sloppy list. SMS is a permission channel governed by the TCPA and carrier rules, and the fastest way to lose the channel is to treat consent casually. Express written consent, honest opt-in language, honored quiet hours, and instant, reliable STOP handling aren’t legal boilerplate — they’re what keeps your messages landing in the inbox instead of the carrier’s spam filter.

There’s a counterintuitive growth argument buried here. A clean list of people who genuinely opted in outperforms a bloated one every time, on every metric that matters: deliverability, click-through, revenue-per-send, and the odds you keep your sending reputation. Treating opt-out discipline as a growth lever rather than a compliance tax is one of the quiet marks of a program that lasts. We go deep on the 10DLC and TCPA mechanics elsewhere; for now, the operator’s takeaway is simple: protect the permission and the permission keeps paying.