ReceivePoint

Ask whether the door will be open, not just where it is.

You already have address data. What you don't have is next Tuesday — the recipient telling you ahead of time that nobody will be in. One call gives you those dates, the address they go with, and when the recipient last confirmed it.

ReceivePoint isn't launched yet. There's no pricing here and nothing to sign up for. We want to talk to a few people who route parcels for a living and find out whether the data below is shaped the way you'd actually use it.

You already solved the address. You have not solved next Tuesday.

Two of these three you already have. The third one is the part we're building.

What you already buy

NCOALink and the validation vendors. They tell you an address exists, is formatted properly, and was filed as a move at some point. All of it looks backwards.

What you already build

Scoring from your own delivery history — which addresses have gone wrong before and how often. Nobody outside your company can match that, and we're not trying to.

What nobody has

The recipient telling you the door is shut from the 24th, before it happens. Not guessed from history or pulled out of a move filing. They typed it in, and we know when.

So this isn't competing with your own scoring. However good your model is, it can't know about a closure that hasn't happened yet. No history can.

You already pay for bad address data. Just not in one line item.

It's spread across re-attempt labour, fuel, call centre time, claims and reverse logistics, so it never adds up to one number anybody owns.

$17.78

per failed delivery attempt

What a first attempt costs when it doesn't land, before anybody tries again.

SmartRoutes
$216B

lost annually across US retail

Failed, delayed and misdirected shipments. Address accuracy is the most direct lever on it.

SmartRoutes
$24.63

average address correction fee

Revenue for you on paper, and a reason the shipper shops elsewhere next quarter. UPS averaged $25.25 in 2026, FedEx $24.00 in 2025.

Refund Retriever and Reveel
$10 to $40

per item in reverse logistics

Processing cost to handle a return, on top of the transport you have already spent.

nShift

One endpoint. Two outcomes worth planning for.

It's a POST so addresses never end up in a URL, a proxy log or a referrer header. Everything comes back in one response.

The recipient is on the registry and is holding deliveries

POST api.receivepoint.com/v1/query
{ "address": "1420 Pearl St, Boulder CO 80302" }

200 OK
{ "address_found": true,
  "deliverable":   false,
  "reason":        "recipient_hold",
  "hold":          { "from": "2026-12-24",
                     "to":   "2027-01-02" },
  "confidence":    0.94,
  "confirmed_at":  "2026-11-02" }

Hold it at the depot or send it to a pickup point, without spending an attempt. The confidence figure only reflects how recently the recipient confirmed the record — it isn't a guess at whether the delivery would have worked. confirmed_at gives you the raw date if you'd rather set your own threshold.

We have never seen this address

POST api.receivepoint.com/v1/query
{ "address": "99 Nowhere Rd, Boulder CO 80302" }

200 OK
{ "address_found": false }

A miss comes back as a 200, not a 404. “We don't know anything about this address” is a real answer, and it shouldn't set off your error handling or your retries.

You don't negotiate access per recipient

When somebody signs up, USPS, FedEx and UPS can read their record. That's settled once, at signup, so you can query any address in the registry without a handshake for each one. Stores work the other way round — the recipient grants each one separately and can take it back.

Nothing is computed while you wait

Confidence is recalculated on a schedule and cached, so a query is just a read. It runs at the edge against a read replica.

Query per parcel or per manifest

One address or a batch, whichever fits where address validation already sits in your pipeline.

Every read is logged

Append-only, tagged with your key, and the recipient can see it. That's part of why they're willing to keep the record accurate.

Why this record stays fresh when the ones you buy go stale.

This is the question that actually matters, and it's about incentives rather than technology. Every address file you can buy decays the same way. It comes from a one-time event, filed by somebody who gets nothing out of filing it.

  1. 1

    The calendar is what brings them back

    Most people move every five to seven years, so a record built around moving gets one visit and then rots. Travel, holiday closures and a standing day off the dock happen several times a year. People come back for the calendar, and they confirm the address while they're there. That's the difference between this and a move file.

  2. 2

    One update instead of thirty

    Somebody looking at thirty accounts updates two and gives up. Make it one action and more of them finish, because the work finally matches what they get for it.

  3. 3

    The recipient gets something out of it

    Their parcel turns up, they avoid a correction fee, and they stop finding cards through the door. They're not keeping it current for you, which is exactly why they'll keep it current. Every confirmation is timestamped, and that's the date you'd set a threshold against.

Where this actually is right now.

Built

The registry schema, the recipient account model, and the rules about who can read what.

Being built

The query endpoint above, the confidence job behind it, and partner key management around both.

Not there yet

Coverage. The waitlist is open but the registry is empty until people are on it, so a query today comes back address_found: false every time.

What we're asking for

An hour. Whether these are the fields you'd branch on, where this would sit in your pipeline, and what would stop it being worth the integration.

A registry like this is worth very little to the first carrier that reads it and a lot to the tenth. Early on that makes it a hard sell, which is why we're starting with conversations instead.

Tell us where this would break.

Leave an email and we'll come back to you ourselves. If you'd rather pull the response apart than talk about integrating, that's the more useful conversation anyway.

Which are you?

Encrypted in transit and at rest, and never sold or passed on.