Stop the RTO bleed —
make couriers pay.
Every RTO is double freight and working capital locked in a truck for two weeks. ShipSense scores the risk before you ship, flags "customer unreachable" NDRs the record does not support, and builds the evidence to claim back what couriers owe you.
THE TRIALone parcel · one night · scroll the whole storyRTO is the silent tax
on Indian D2C.
Every RTO means double freight, two weeks of blocked working capital, and dead packaging — with the courier writing their own failure report.
Drop your shipment CSV.
See your RTO bleed in 10 seconds.
A quick estimate from your own orders — no number tweaking, no sales call. Parsed entirely in your browser. Nothing is uploaded or stored.
This is the rough cut. The free audit returns your exact number — and which of these NDRs don't survive a timestamp check.
Get my exact number — free audit →Four moments.
One outcome: your money back.
No dashboards. No manual work. ShipSense runs quietly alongside every order — and acts when it matters.
The courier says one thing. Gawah asks your customer.
Gawah — your customer becomes the witness. On a suspicious failed delivery, ShipSense sends one WhatsApp question. The one-tap reply is recorded with its timestamp — the customer's own account of whether anyone came to the door.
The cheapest RTO is the one that got paid for before it shipped.
COD orders come back at a multiple of the prepaid rate. Rather than quote you somebody else's festive-quarter statistic, the free audit shows you your own COD-versus-prepaid split first. Then this engine goes after the gap: it moves the riskiest COD orders to prepaid before dispatch, and measures whether that actually worked.
Gawah asks. Hazri proves. A reply the courier cannot dispute.
A "no one came" is a statement. Hazri turns it into evidence: a one-time word is committed to a sealed ledger, sent to the consignee, and echoed back — by text, and on a recorded call. Each echo is matched to the commitment and timestamped. The courier's "customer unreachable" now sits next to a reachable customer, on the record.
A delivery check written in English gets ignored in Kanpur. ShipSense writes its messages in English, Hindi and Marathi. The engine picks the language from the delivery state, and you can override it per order. Each language is its own Meta-approved template — which is what lets Hazri open a conversation instead of only replying inside one.
A human-sounding nudge, before the parcel moves
ShipSense sends a short WhatsApp voice note in your brand's name — confirming the address, checking availability, or nudging COD to prepaid. Three real samples below.
Samples for illustration. Voice notes stay switched off until your pilot starts and you enable them — nothing is sent to your customers before then.
Follow the money in your logistics stack. ShipSense is paid by you, never by a courier.
Follow one rupee. Where it's paid decides whose side each player is on.
Your aggregator takes a cut of every courier shipment. Exposing a courier's errors would shrink its own revenue — so nobody in that chain is paid to check them, and you eat the RTO.
A tool is on the side of whoever pays it. ShipSense is paid by you, never by a courier.
Starline Exports is our founder's own brand. It is permanently disclosed here, and its shipments are permanently excluded from every published ShipSense statistic, index and certificate. We test the engine on it; we never count it. If any published number would change when Starline rows are removed, we retract it, recompute it, and publish the correction.
A courier honesty score that gets worse when they lie.
A courier can dodge accountability by filing a failure that did not happen — "customer unavailable", "address not found". ShipSense scores every one of those claims against its own delivery evidence, so a courier that fabricates a failure lowers its own honesty number, and can't buy it back by filing more. It's not a dashboard opinion — it's a rule built so telling the truth is the only way to score well.
Describes reporting accuracy only — not liability, fault, or any recovery outcome. Illustrative sample on anonymised couriers; a real pilot's own couriers replace this once they ship.
Every 'delivery failed' claim is checked against our own delivery evidence — a reply, a call answered, a payment made. A false claim scores zero and drags the courier's number down. Filing more claims can't raise it.
This measures one thing: how truthfully a courier reports why a delivery failed. It is not a verdict of liability, fault, or money owed — those live in the evidence pack, separately.
Every number here is labelled sample and runs on anonymised couriers. The day a pilot brand ships real orders, the same score fills with their real, privacy-safe couriers — and only then names them.
When we say it didn’t rain, we can tell you how often we’re wrong.
“Heavy rain” is the easiest failure reason to write and the hardest to check. ShipSense checks it against two independent public records for the exact destination PIN code on the exact date — then checks itself against the India Meteorological Department’s own rainfall grid, and publishes the error rate.
Measured, not claimed: IMD 0.25° gridded rainfall 2023–2025 against NASA POWER and NOAA CPC on the same cells and days — Delhi, Mumbai, Pune, Bengaluru, Hyderabad, Kolkata, 1 Apr–31 Oct, 1,952 both-dry days. This is the accuracy of the METHOD. It is not a dispute win rate and not a recovery figure.
NASA Langley Research Center. United States Government public domain, freely available with no restriction on use. A physics-based reconstruction of the atmosphere on the day in question.
NOAA/OAR/ESRL Physical Sciences Laboratory, Boulder, Colorado. United States Government public domain. Built from instruments on the ground — a different institution and a different kind of evidence from a model.
IMD's own 0.25° daily rainfall grid, 2023–2025. We do not print IMD figures in a pack — IMD is what we TEST OURSELVES AGAINST, and it defines the 2.5 mm rainy day.
The rain explanation is not supported. This is the only branch that becomes an accusation.
The day was borderline, or heavy rain fell the next morning inside the window IMD counts against this date. Shown, never filed.
It rained. The exhibit says so and corroborates the courier. Exoneration is symmetric on purpose.
No verdict is drawn. Absence of a rainfall record is not evidence the day was dry.
Why a claim gets withheld: IMD measures rain from 08:30 to 08:30 and stamps the day it ENDS, so its record for a date overlaps ours for the day before. Across six cities that shift is worth a correlation of 0.62–0.96 against 0.23–0.61 unshifted. A dry day that is only just dry, or one followed by a downpour, is not a safe accusation — so we do not make it.
The method is measured; the book is not yours yet. Every failed delivery in our system today is simulated, so we quote no win rate and no recovery figure. When we show you a rupee number, it will have come from your shipments.
Rainfall data courtesy of the NASA Langley Research Center POWER Project, and of NOAA/OAR/ESRL PSL, Boulder, Colorado, USA — both United States Government public-domain records. Accuracy tested against India Meteorological Department gridded rainfall, IMD Pune. Every exhibit prints its source, its licence, and the moment the record was fetched.
Certified, not alleged.
Anyone can point at a courier and call a failed delivery fake. ShipSense issues a certificate instead: a sequential test that accuses only when the evidence crosses a line pre-committed to a false-accusation rate of at most 1.05%, plus a 95%-confidence floor on how much that courier actually fabricates. Too few reports, and the courier stays in Monitoring — we never accuse on a hunch.
Describes reporting accuracy only — not liability, fault, or any recovery outcome. The 95% floor is a lower bound from each courier's own record and assumes independent reports. Illustrative sample on anonymised couriers; a pilot's own couriers replace this once they ship.
Each line is one courier's cumulative evidence (log-likelihood), case by case. The red walk is accused at the exact case it crosses the pre-committed line — early, because certainty arrived early. The amber walk never crosses either line and stays in Monitoring. The lime walk sits below the clear line today — and keeps being tested as new reports arrive. Conflicting reports move no walk. Sample drawing; a pilot sees this chart on their own couriers.
We accuse a courier only when a sequential test crosses a line pre-committed to a false-accusation rate of at most 1.05% — about one in a hundred. It reads the evidence as it arrives, so we accuse the moment it is certain and not before.
Every certificate carries a 95%-confidence lower bound on how much the courier actually fabricates, computed from that courier's own corroborated record. Smoothing never inflates the floor, and conflicting reports carry no accusation weight.
A courier with too few corroborated reports stays in Monitoring — never accused on a small sample. That restraint is the whole guarantee, and it is tested: a three-of-three fabricator is not certified.
The rate card is not the price.
A courier that fabricates failed deliveries charges you twice: once on the invoice, once in fake RTOs. ShipSense multiplies each courier's certified fabrication floor by its own failed-delivery rate on your book and a labelled NDR-to-RTO conversion assumption — giving a minimum number of fabricated RTOs per 1,000 shipments, part measured and part a stated assumption. Compare couriers on that, not on the quote.
Same anonymised couriers and certified floors as the certificate above. The 0.5 conversion factor is a labelled, deliberately conservative assumption — only half of fabricated attempts are assumed to become an RTO — replaced by your measured rate on a real book. Rupee figures come only from the number you typed; we never invent one. Describes reporting-accuracy economics — not liability, fault, or any recovery outcome. Illustrative sample; a pilot sees this on their own couriers and their own rate cards.
Three tools fight RTO.
ShipSense goes after the courier.
Checkout tools work before dispatch. Workflow tools work after NDR. Neither touches the gap between — where couriers write their own failure reports and your money disappears.
The gap between dispatch and resolution is where your money disappears. That gap is where ShipSense does its main work.
See what RTO is quietly costing you.
Move the sliders. The bleed is your own arithmetic on industry benchmarks. What ShipSense can claw back is a question for your audit — not something we'll guess at on a landing page.
Wemeasuresavingsthehonestway.
A random holdout group ships exactly as before. The rest get scored, held, or converted. We measure the gap. That gap — once it clears 95% confidence — is the only number we will ever report as your savings. We have not run a cohort yet, so there is no number here to show you.
One price. It moves with your volume, not with a sales call.
There are no plans to compare and nothing to negotiate. You pay a flat monthly fee, a small rate on the shipments we handle, and a tenth of the money we provably get back for you. Recover nothing and the third line is zero. Prices exclude GST. Fair-use limits are included: 100 verification calls and ₹300 of AI compute per store per month — we flag you well before you'd reach either. The free audit comes first either way, so you see your bleed before you pay a rupee.
Straight answers.
Stop guessing. Get your real RTO number.
A free 7-day audit on your own shipment data. No credit card, no rip-and-replace — we read alongside your existing couriers.
A math-first logistics intelligence layer for Indian D2C. We score RTO before you ship, flag NDRs the record does not support, and build the case to recover what couriers owe you.
Founder & Creator, ShipSense
Geist
Inter
Mono