---
title: "Introducing OptimalDial Identity: Know Which Numbers Aren't Your Contact's"
url: "https://optimaldial.com/blog/introducing-optimaldial-identity"
description: "About one in five numbers on a purchased list belongs to someone other than the contact named on that row. OptimalDial Identity checks every number against the name on its row and flags the ones that don't match — on ~80–90% of a typical list, before anyone dials."
---

Your reps are having conversations with people who were never your prospect.

On one customer’s file, **about a third of connected calls reached the wrong person** — a third of every conversation their team had, gone before anyone said anything useful. Nothing in a connect rate shows you that. A wrong number that picks up looks exactly like a win right up until the rep says a name.

Today we’re shipping **OptimalDial Identity**: a second signal, on the same upload, that checks every number against the contact named on that row and flags the ones that don’t match — before anyone dials.

~1 in 5 numbers on a bought list belongs to someone other than the contact named on that row — around 20% on average across our data

## Why this keeps happening

Lists decay. People change jobs, carriers recycle numbers, and plenty of contacts hand out a main line instead of a mobile. Across our data it runs at **around 20%** on average — roughly one number in five that doesn’t belong to the contact named on that row.

That isn’t a data-vendor problem you can buy your way out of, and it isn’t visible from the outside. Until someone dials the number and hears the wrong voice, every row looks the same.

## Introducing OptimalDial Identity

Identity is a second signal that runs on the same upload as [Answer Intent](/answer-intent). Answer Intent asks whether a number will pick up. Identity asks whether it belongs to the contact on that row.

It’s worth being precise about what that is **not**. Identity doesn’t find you a phone number, append one, or swap in a better one. It judges the number you already have, against the name you already have. There’s no new data source to buy and nothing to reconcile afterward.

The only new input is a name. Map a full-name column, or first and last name columns, and Identity runs. Everything else is derived from the number itself.

## How we tell whose phone it is

Three signals, none of which needs anything from you beyond that name.

**The carrier caller name.** CNAM is the caller-name record carriers attach to a number — the “John Smith” your phone shows on an incoming call. We read it and compare it against your row. The matcher handles nicknames, initials, the 15-character truncation carriers apply, and `LAST,FIRST` ordering, and it can tell a person’s name apart from a city or a company. It’s deterministic: the same file returns the same verdict every run. Mobile only.

**The voicemail greeting.** Most people record their own greeting and say their name in it. We compare what’s said against the name on your row. This is the quietly important one: it works on numbers that never pick up, which is where most of a list actually sits.

**The live call.** When a human answers, what’s said is compared against the contact’s name.

These don’t vote; they accumulate. Each signal contributes evidence for or against “this is the right person.” Two signals that agree raise confidence, signals that disagree partly cancel, and dial-based evidence outweighs the carrier record, which can be years stale. What comes out is a probability, reported as one of five bands — not a yes or no.

## Why coverage is the argument

CNAM on its own is unreliable. Records go stale, come back blank, or return something generic like a carrier name. That unreliability is why approaches built only on what happens when someone picks up skip CNAM entirely — and it caps them at the numbers that actually answered, which is roughly a third of a list.

Weighing three signals probabilistically instead of trusting any one of them deterministically changes the shape of the problem. We put an identity band on **~80–90% of dialable numbers**, consistently.

Channel

Coverage

Likely Answer

~91%

Likely Call Screening

~88%

Likely Voicemail

**~86%**

Non-Mobile

0%

The voicemail row is the one that matters. **~86% coverage on numbers that never picked up** is only possible because greetings and carrier records both carry names. An approach that can only learn from live conversations cannot reach those numbers at all — and on most lists, that’s where the volume is.

~87% of dialable numbers come back with an identity band — including numbers that never picked up

Non-Mobile is 0% structurally, not as a gap. Identity is mobile-only: a list of landlines comes back Non-Mobile on every row and carries no identity signal.

## Two answers, one column

Two signals would mean two columns to reconcile, so we cross them into one. The whole rule is one sentence: **take the Answer Intent value; if identity is Medium or better, insert “Right-Party”; if identity is Low, replace it entirely with “Likely Wrong Number.”**

OptimalDial\_Status

What it means

`1 - Likely Right-Party Answer`

Likely to answer, and likely to be your contact.

`2 - Likely Answer`

Likely to answer. We can’t say whose number it is.

`3 - Likely Right-Party Call Screening`

Likely to reach iOS, Android, or Samsung call screening, and likely to be your contact.

`4 - Likely Call Screening`

Likely to reach iOS, Android, or Samsung call screening. We can’t say whose number it is.

`5 - Likely Right-Party Voicemail`

Likely to go to voicemail, and likely to be your contact.

`6 - Likely Voicemail`

Likely to go to voicemail. We can’t say whose number it is.

`7 - Non-Mobile`

A landline or VoIP number. These carry no identity signal.

`8 - Likely Wrong Number`

Unlikely to be your contact.

Ordering is answer-intent-major on purpose: tier 2 outranks tier 5 because a 2–5× connect multiplier is worth more than the gap in identity confidence. Identity only overrides at the floor — tier 8 sinks from anywhere, because a big connect boost on a number that isn’t your contact’s is worth nothing. Tiers 2 and 4 are rare, about 3% combined, and structurally so: if someone picks up, we have something to compare against, so there’s a verdict. “Someone answered and we still don’t know who” is close to a contradiction.

Tier 7 is the only one without the word “Likely,” because line type is a carrier fact rather than a forecast. And the numeric prefix isn’t decoration — it makes the column sort correctly as plain text in Sheets, Excel, and every dialer import we’ve tried.

## What a real list looks like

Here’s one 2,374-contact list with mixed answer behavior. Yours will differ, but the shape is typical.

Tier

Share

1 - Likely Right-Party Answer

~13%

2 - Likely Answer

~2%

3 - Likely Right-Party Call Screening

~11%

4 - Likely Call Screening

~2%

5 - Likely Right-Party Voicemail

**~37%**

6 - Likely Voicemail

~8%

7 - Non-Mobile

~6%

8 - Likely Wrong Number

**~20%**

Unknown

~1%

The point of splitting a list this way isn’t to shrink it — it’s to work each part through the channel it actually responds to. Here’s where we’d start.

**Tiers 1 and 2 — go phone-heavy.** These pick up, and on tier 1 it’s your contact. Build a call-first sequence around them. This is the part of the list where cold calling works exactly the way it’s supposed to.

**Tiers 3 and 4 — dial them if you have a plan for screeners.** Call screening isn’t a dead end; teams connect with screened numbers all the time. But it needs its own approach, so work these after tiers 1 and 2, when you have the bandwidth.

**Tiers 5 and 6 — go multi-channel.** On most lists this is the biggest group by some distance, and it’s the one people waste the most effort on by treating it as a call-only problem. The pattern our customers get the most out of: send the email first, then call and leave a voicemail that references it — “I just sent you a note about X.” The voicemail does the work the dial couldn’t.

**Tiers 7 and 8 — stop dialing and go get better data.** A landline won’t become a mobile and a wrong number won’t become the right one by calling it again. Look for an alternative number for that contact, or move them to other channels entirely.

Those are recommendations, not rules, and we’d rather you treat them that way. What works against VPs at enterprises isn’t what works against owner-operators, and your sequence, your market and your team’s capacity all move the answer. It’s also why we hand back every signal instead of just the rows we’d dial ourselves — the whole point of scoring a list is that you can see what your own campaign responds to and act on it. You know your ICP better than we ever will. The data is there so the decision is yours to make.

A word on `Unknown`, because it’s the easiest value to get wrong: it is not “clean” and it is not “no data.” It sits between Medium and Low, closer to a coin flip than to a safe number. Dropping it costs about as many good conversations as it avoids bad ones, so we don’t filter it out for you and we don’t suggest you do it by default.

## What it changes on a real list

Measuring this properly needs a file where every row was dialed, so both halves of the comparison are real rather than modelled. We had one: **3,870 dials**, with the rep’s call dispositions returned for each. Here’s what dialing right-party only would have changed.

Dial everything

Right-party only

Wrong numbers per 100 dials

9.4

**4.9**

Wrong-number rate on connects

~34%

**~19%**

Numbers dialed

3,870

2,613

Right-person connects per 100 dials

18.5

**21.6**

~48% fewer wrong-number conversations per 100 dials — 9.4 down to 4.9, on the same effort

The top row is the one that matters. Every 100 dials produced **9.4 wrong-number conversations** before, and **4.9** after — roughly half, for the same effort. On that file about **a third of all connected calls** were reaching the wrong person to begin with, which is a third of a rep’s day spent finding out they’re talking to someone who was never the prospect.

The part people expect to be a trade-off isn’t one. Cutting a third of the list did not cost good conversations: right-person connects per 100 dials went **up**, from 18.5 to 21.6. The dials that got removed had negative expected value — the bottom band produced **zero pipeline outcomes across 411 dials.** No meetings, no activated leads, no referrals.

If you track right party contact (RPC) rate, that’s the number this moves, and it moves it by removing dials rather than adding them.

## One more thing we found

While we were validating this, a result fell out that we didn’t expect and initially didn’t believe: **wrong numbers answer more often than right ones.**

CNAM verdict

numbers dialed

answer rate

Says a different person

3,839

**~20%**

Says the contact’s full name

13,627

**~17%**

We checked it against a second, unrelated file and saw the same reversal. The explanation is obvious in hindsight: a stranger has no reason to screen a call from a number they don’t recognise, while the VP you’ve been chasing has been cold-called for years and screens everything. The people easiest to reach are, systematically, the people you didn’t mean to reach.

It’s a curiosity more than a selling point, but it does have one practical consequence worth stating: **connect rate on its own can’t tell you whether things are going well.** Push it hard enough and you drift toward the numbers that answer most freely, which are disproportionately the wrong ones. That’s the reason we kept the two signals separate instead of blending them into a single quality score.

## Identity is in public beta

We’re shipping it because it’s already good enough to change how a list gets worked, not because it’s finished. What the score is built to tell you is whether a number is live and correctly attributed to your contact — not a guarantee about who picks up on any given dial. Accuracy is still improving, the bands will move as more accounts come back, and there are edge cases we haven’t seen yet.

So: run a list through it and tell us what you got. If a number came back `8 - Likely Wrong Number` and your rep reached the right person, that’s exactly the case we want. Reply to any of our emails or send it to support — real dispositions from real lists are what make the next version better.

## What it costs

Nothing changes. Standard validation is **1 credit per number** and MAX is **1.5**, exactly as before, and the depth you pick affects the answer-intent work only. Identity runs at both and doesn’t consume separate credits. Full detail on the [pricing page](/pricing).

## How to use it

**In the app:** upload as usual. The column-mapping step now auto-detects your phone column and your name columns from the header row and shows them against real rows from your own file. Identity is on by default — a list with no names is the exception, and you can turn it off. When your list finishes you get a breakdown of the tier distribution, and an export builder where you pick which tiers and which columns you download.

**Via the API:** add `full_name_column`, or both `first_name_column` and `last_name_column`, when you create an upload. In JSON mode send `contacts` rather than a flat list of numbers, since a bare number carries no name.

```
{
  "full_name_column": "Contact Name"
}
```

Downloads accept `statuses`, `include_answer_intent`, `include_identity`, and `sort_by_status`. Nothing you already built changes: with no new parameters every response is identical to before, and `/api/v1/contacts` is frozen on purpose so existing filters keep working. The nine-value tier lives on `/api/v2/contacts`. Details on the [developers page](/developers).

Find out how many numbers on your list aren't your contact's.

[Get Started →](https://app.optimaldial.com)

## Frequently asked questions

**What is OptimalDial Identity?**

OptimalDial Identity is a second signal that runs alongside Answer Intent on every list. Answer Intent asks whether a number will pick up; Identity asks whether that number belongs to the contact named on the same row. It judges the number you already have — it does not find, append, or replace a phone number. The only extra input is a name column.

**What does the OptimalDial\_Status column contain?**

One of nine values that cross both signals. The rule is one sentence: take the Answer Intent value; if identity is Medium or better, insert Right-Party; if identity is Low, replace the whole thing with Likely Wrong Number. The numeric prefix — 1 - Likely Right-Party Answer through 8 - Likely Wrong Number — makes the column sort correctly as plain text in Sheets, Excel, and dialer imports.

**Do wrong phone numbers really answer more often than correct ones?**

In our data, yes. Across 17,466 dialed numbers, those whose carrier caller-name record named a different person answered about 20% of the time, while numbers whose record named the contact answered about 17%. We saw the same reversal on a second, unrelated list. A stranger has no reason to screen a call from a number they don't recognise, while the prospect you're actually chasing has been cold-called for years. It's the reason connect rate on its own can't tell you whether a list is going well.

**Is OptimalDial Identity generally available?**

It's in public beta. It runs on every list with a name column today and carries no separate charge, and accuracy is still improving as more accounts return call dispositions. If you run a list through it, we want to hear what you saw — especially cases where a number came back as a likely wrong number and your rep reached the right person.

**Does OptimalDial Identity cost extra?**

No. Identity is part of the product and doesn't consume separate credits. Standard validation is still 1 credit per number and MAX is still 1.5; the depth you choose affects the answer-intent work only.

Keep reading

## More from the blog

*   [
    
    ### Best Time to Cold Call: A 2026 Study of 128,998 Calls
    
    We pulled 128,998 outbound calls to settle it. The hour barely moves the needle — calling the right contacts more than doubles your connect rate. Here's the data, by hour and day.
    
    Neel Rawlani · June 9, 2026
    
    
    
    ](/blog/best-time-to-cold-call)
*   [
    
    ### Introducing MAX Validation: Get More Conversations From Every List
    
    A new validation tier for smaller, harder-won lists. MAX flags close to double the likely answerers — so you get more total conversations from the same list.
    
    OptimalDial · June 5, 2026
    
    
    
    ](/blog/introducing-max-validation)

## Ready to Hit 25–30% Connect Rates?

Start with your own list today, or book a demo to see how it works.

[Get Started →](https://app.optimaldial.com) [Book a Demo](#)

14-Day Money-Back Guarantee · Your Data Stays Yours · US & Canada