Resolution · 2026-10-17Deflection vs resolution: why answered isn't solvedA deflected ticket is one a human didn't touch. A resolved ticket is one that got finished. They are not the same number, and only one of them keeps revenue.Read
Most support bots report deflection: the share of tickets they answered without a human. It sounds like progress. It usually isn't.
What deflection actually measures
Deflection counts tickets a bot answered so a human didn't have to. It says nothing about whether the shopper's problem got solved. A tracking link in reply to where is my order is a deflection, and often a second ticket an hour later.
What resolution measures
Resolution counts the tickets that got finished: the order tracked and explained, the return turned into an exchange, the subscription saved or honestly canceled. That is the number that keeps revenue, and it is the number worth reporting.
Why the gap matters
Read-only bots stall near 35 to 40% deflection as a published benchmark, and the rest still hits a human. An agent with write-access can finish the task, so it reports resolutions and the revenue they saved, not a deflection rate that hides the repeat tickets.
See WISMO on the product pageWhat a Churn & WISMO Audit shows
Strategy · 2026-10-17Resolution jobs are modular: apparel vs supplements vs furnitureThe same agent runs different jobs in different categories. What fires in apparel is not what fires in supplements or furniture, and pretending otherwise breaks things.Read
A resolution agent that runs the same three jobs everywhere is wrong somewhere. The jobs are modular, and the category decides the mix.
WISMO is the constant
Where is my order fires in every category, because every store ships something and every shopper wants to know where it is. For food and furniture, WISMO even leads, because perishables and freight need narration, not a tracking link.
Returns and subs depend on the goods
Returns-to-exchange leads in apparel, where fit drives returns. Subscription-save leads in supplements and beauty, where replenishment runs on subs. Furniture has neither paid subs nor easy returns, so it runs two jobs, not three.
Why the discipline matters
Running subscription-save on an apparel brand with no paid subs is noise. Leading beauty with exchange when hygiene rules block most returns loses trust. The audit maps the exact mix, so TX Agents runs the jobs that fit, and only those.
See the five category decksReturns to exchanges
Returns · 2026-10-17Turn returns into exchanges without hiding the refund buttonMost returns are a fit problem a swap would fix. You can protect the margin by offering the better fit first, and you can do it honestly.Read
Returns are 30 to 50% of DTC tickets, and most of that revenue walks out as a cash refund. It doesn't have to, and you don't need a dark pattern to keep it.
Why returns become refunds
By the time a shopper reaches the returns portal, the default path is a cash refund. The better-fit swap, the one that actually solves their problem, is a few extra clicks they never take.
Intercept before the portal
If you catch the refund intent in the support conversation, you can offer the right-size swap or store credit first, framed as the easier outcome. Most shoppers wanted the product to fit, not their money back.
Keep it honest
The refund stays one click away. Hiding it produces chargebacks and one-star reviews. Offering a genuinely better option first, then honoring the refund if they still want it, keeps both the margin and the customer.
See returns-to-exchange on the product pageModular jobs by category
Subscriptions · 2026-10-17Subscription-save that's compliant: pause-first, honest cancelYou can cut subscription churn without trapping anyone. Offer a pause or swap matched to the real reason, and keep the cancel button one honest click away.Read
Subscription cancels are recurring revenue walking out, usually over something a pause or a swap would fix. The trick is saving the subscriber without burning the relationship.
Catch the reason, not just the click
Too much stock, a tight month, a product that didn't suit: each has a different save. A pause fixes over-stock, a cadence swap fixes too much too often, a product swap fixes fit. Guessing one offer for all of them fails.
Pause-first, swap-second
Offer the save that matches the reason before the cancel goes through. A subscriber who pauses for 60 days is retained revenue; a subscriber who cancels is gone and often gone for good.
Honest cancel, always
If they still want out, cancel in one step and never hide the button. Regulators and shoppers both punish dark-pattern cancel flows. An honest save keeps the ones you can and lets the rest go cleanly.
See subscription-save on the product pageModular jobs by category
Audit · 2026-10-17What a Churn & WISMO Audit actually shows youThe audit reads your own data and hands you a one-page picture: your ticket mix, your leaks, and the dollars you could recover, before anything is built.Read
Before any build, the free Churn and WISMO Audit answers one question honestly: what is your support queue actually costing you? The number comes from your data, not an average.
Your ticket mix
The audit breaks down what share of your queue is WISMO, returns, refund-status, subscriptions and everything else. Most founders are surprised how lopsided it is, and how much of it is resolvable.
Where the money leaks
Return rate and the refund-versus-exchange split, refund-status volume, subscription churn and cancel reasons, and the gap between deflection and real resolution. Each is a dollar figure, sized from your data.
A one-page recoverable number
The output is a single page: the recoverable-revenue estimate as a range, and which TX Agents jobs fit your category.
Get your free auditWISMO's real cost
Resolution · 2026-10-17Why read-only bots stall at 35 to 40% (and write-access doesn't)A bot that can only read can only answer. The moment a ticket needs an action, it hands off, which is why deflection plateaus.Read
Read-only support bots hit a ceiling. The published benchmark puts it near 35 to 40% deflection. The reason is structural, not a tuning problem.
The ceiling is the action
Most tickets that matter need something done: a reship opened, an exchange written, a subscription paused. A read-only bot can describe the action but not take it, so it hands off, and the deflection number stops climbing.
Write-access changes the math
An agent with scoped write-access can finish the task: track the order and open the reship, intercept the return and write the exchange, catch the cancel and apply the pause. Those move from deflected to resolved.
Safely, not recklessly
Write-access only works behind limits: least-privilege scopes, hard value caps, a full log, and a human for the hard cases. That is the trust gate that lets an agent do the work without exceeding what you granted.
See how TX Agents stays safeDeflection vs resolution
WISMO · 2026-10-17WISMO is 40 to 60% of your tickets. Here's the real costWhere-is-my-order is the biggest, most repetitive slice of a DTC queue. The cost isn't just support time; it's the refunds and churn a slow answer sets off.Read
WISMO is the single largest category in most DTC support queues. It is also the easiest to answer from live data, and the most expensive to answer slowly.
The direct cost
At 40 to 60% of tickets, WISMO is where most support hours go. Multiply by your WISMO volume and the number gets real fast.
The hidden cost
A slow or vague WISMO answer breeds a second ticket, a refund request, or a chargeback. For perishables and freight it is worse: a late box that nobody narrates becomes a spoilage refund or a damage dispute.
The fix
Answer from live order and carrier data, not a tracking link, and catch the late shipment before the shopper writes in angry. That is a resolution, and it stops the chain of costs a slow answer starts.
See perishable WISMO for food brandsDeflection vs resolution
Integration · 2026-10-17The operator-layer model: TX Agents runs on top of Loop, Gorgias and RechargeTX Agents isn't another tool to rip and replace. It's an operator that sits on top of the stack you already run and actually operates it.Read
The fastest way to kill a DTC integration is to ask a brand to rip out a tool it depends on. TX Agents doesn't. It operates on top of what's already there.
On top, not instead
Loop or Redo stay your returns app. Gorgias or Zendesk stay your helpdesk. Recharge or Skio stay your subscriptions tool. TX Agents runs on top of them, so they remain the system of record and nothing you depend on gets torn out.
Intercept pre-portal
TX Agents catches resolution intent in the support conversation, before the shopper reaches a returns or cancel portal, and takes the action there. That is where intent is highest and the best outcome is still possible.
Near-zero switching risk
Because nothing is replaced, switching risk is near zero. The operator layer turns on the outcomes your existing tools are capable of but rarely configured to produce.
See how TX Agents wires inPartners: TX Agents operates the tools you sell