Call Disposition Codes: A List for India
Jaseem Abbas
- Published
- Updated

A call disposition is the short code an agent or system records after a call to say how it ended, such as callback, not interested or promise to pay. The code is the shorthand (CB, NI, PTP) and the label is the plain words. This guide gives you a ready list of call disposition codes for Indian sales, collections, service and e-commerce floors, with what each one means, the rules for naming them, how many to keep, and the numbers they feed.
Here is why a missing code costs money. Imagine a customer in Nagpur who says, "Please do not call me again." The agent is in a hurry, finds no code for that, and picks "not interested". The dialer sees "not interested", waits a week, and calls the number again. The customer is now angry, and the call may be a breach of your own do-not-call rule. A single missing code caused all of it.
The Nagpur desk, its numbers and its people in this post are an example we made up. They are not customer data.
What is a call disposition?
A call disposition is the outcome of a call, recorded as a short label after the call ends. Five9, a contact center software vendor, describes dispositions as labels or tags that agents or the system assign to completed customer interactions, used to record the outcome, to trigger actions such as a callback or adding a number to a Do Not Call list, and to track metrics (Five9 on dispositions). Floor teams in India simply say "disposition". "Call outcome codes" means the same thing.
What is a disposition code?
A disposition code is the short form of a disposition: PTP for promise to pay, CB for callback requested, DNC for do not call. The label is the same idea written out for a person, and a good code set has both. Choosing the code is part of after-call work, so a long list slows every call.
Why do call disposition codes matter?
Disposition codes matter because they do three jobs at once. They record what happened, they decide what happens next, and they feed the numbers you manage by. These are the same three purposes Five9 lists.
- Record: a manager can see why 400 calls ended without a sale, not just that they did.
- Trigger: a callback code puts the number back in the queue for the date the customer chose. A do-not-call code takes it off the list.
- Measure: connect rate, promise-to-pay rate and conversion are all counts of codes divided by other counts of codes.
The Nagpur story above shows the cost of getting one wrong. A code is not paperwork. It is an instruction to the next person, or the next dialer run, about what to do with this number.
What is the difference between call status and call disposition?
Call status is the network's answer to the dial. Call disposition is what the conversation achieved. A phone that is switched off has a status and no disposition, because no conversation happened. A call that connects and ends with "I will pay on Friday" has both: a status of connected, and a disposition of promise to pay.
Why keep them apart? A person should not have to type "no answer" 400 times a day, and a dialer cannot know whether a customer said yes. So in our view the system logs status and the agent, or an AI agent, chooses the disposition. Mixing them in one list gives you a 60-line dropdown that nobody reads. Alongside the code, an AI agent can also write a written call summary of each call.
It also gives you a two-level tree. The first level is how the call ended, either not connected or connected. The second level is what a connected call achieved, grouped by outcome. Talkdesk, another vendor, supports nested dispositions in its product help for the same reason: sub-categories inside a main outcome (Talkdesk on dispositions).
The same codes work for inbound and outbound calls, though the mix differs. Our guide to inbound versus outbound calls explains the difference.
What call disposition codes should an Indian call center use?
Start with the list below and delete what you do not need. The codes are our suggestions, built for Indian floors, and they are not a standard. The three letters are short on purpose, so rename them to match your own dialer. Every table has the same four columns: code, label, when to use it, and the next action.
Which codes cover calls that did not connect?
These are statuses, so the dialer or system should set them, not a person. How a system tells these apart is answering machine detection. Switched off and not reachable are worth their own codes on an Indian list, and few global lists name them.
Code: NA
- Label
- No answer
- When to use it
- The phone rang and nobody picked up
- Next action
- Retry at another hour
Code: BSY
- Label
- Busy
- When to use it
- Engaged tone
- Next action
- Retry in a short while
Code: SWO
- Label
- Switched off
- When to use it
- The recorded message says the phone is switched off
- Next action
- Retry later in the day or next day
Code: NRE
- Label
- Not reachable
- When to use it
- Out of coverage, or the network cannot connect
- Next action
- Retry later; check after repeated misses
Code: INV
- Label
- Invalid number
- When to use it
- Not in service or does not exist
- Next action
- Remove from the list, flag for a data fix
Code: VM
- Label
- Voicemail
- When to use it
- A voicemail greeting answers
- Next action
- Retry; do not count as a person reached
Code: DRP
- Label
- Dropped
- When to use it
- The line cut before anyone spoke
- Next action
- Retry soon
Which codes cover any connected call?
These apply to every connected call, whatever your business. Right party contact is the key one, because most rates in this post use it as a base.
Code: RPC
- Label
- Right party contact
- When to use it
- You spoke to the person you meant to reach
- Next action
- Go on to the outcome codes
Code: TPC
- Label
- Third party
- When to use it
- Someone else answered; no account details shared
- Next action
- Ask for a good time to call back
Code: WN
- Label
- Wrong number
- When to use it
- The person says it is not the person or number
- Next action
- Remove or flag the number
Code: CB
- Label
- Callback requested
- When to use it
- The person asked you to call later
- Next action
- Record date and time, then call then
Code: LNG
- Label
- Language needed
- When to use it
- The person asks for another language
- Next action
- Route to a speaker of that language
Code: HUP
- Label
- Hung up early
- When to use it
- The person cut the call before the purpose was said
- Next action
- Retry once; do not log as not interested
Which codes suit sales calls?
A disposition becomes the lead status in your CRM, so name these as the stages you really use. Keep the reasons for a no short and fixed.
Code: INT
- Label
- Interested
- When to use it
- Wants more detail or a quote
- Next action
- Send details, book a follow-up
Code: NI
- Label
- Not interested
- When to use it
- A clear no. Add a reason: price, has one, not now, no need
- Next action
- Close or add to a later campaign by reason
Code: APT
- Label
- Appointment booked
- When to use it
- A demo, visit or meeting has a date
- Next action
- Confirm by message; remind the day before
Code: SALE
- Label
- Sale made
- When to use it
- The customer agreed to buy
- Next action
- Hand over to fulfilment
Code: FU
- Label
- Follow-up
- When to use it
- Interested, but needs time or a decision-maker
- Next action
- Set a follow-up date
Code: NQ
- Label
- Not qualified
- When to use it
- Does not fit your offer or area
- Next action
- Close with the reason
Which codes suit collections calls?
Collections is general advice here, not a product feature of ours. The codes below are what a collections desk usually needs. Record what was said, not what you hope will happen. The RBI rules on calling hours and recording are in our guide to RBI rules for recovery agent calls, and a code never makes a lender compliant on its own.
PTP full form in collections: promise to pay
PTP means promise to pay. A PTP code needs a date and an amount, in fields that can be read later. Our guide to payment reminder calls covers what a good PTP call sounds like.
Code: PTP
- Label
- Promise to pay
- When to use it
- Agreed to pay on a stated date
- Next action
- Save date and amount; call after that date if unpaid
Code: PPTP
- Label
- Part payment promised
- When to use it
- Agreed to pay only part now
- Next action
- Save date and amount; plan the balance
Code: PAID
- Label
- Says already paid
- When to use it
- The person says payment was made
- Next action
- Verify before the next call
Code: RTP
- Label
- Refused to pay
- When to use it
- Clear refusal, with the reason noted
- Next action
- Hand to a senior or the next stage of the process
Code: DSP
- Label
- Dispute
- When to use it
- The person disputes the amount or the loan
- Next action
- Pass to a person; stop reminder calls on it
Code: SKP
- Label
- Skip
- When to use it
- The person has moved and cannot be found
- Next action
- Send to the tracing team
Which codes suit service calls?
Code: RES
- Label
- Resolved
- When to use it
- The issue was solved on the call
- Next action
- Close the case
Code: TKT
- Label
- Ticket raised
- When to use it
- The issue needs work after the call
- Next action
- Give a reference number and a date
Code: ESC
- Label
- Escalated
- When to use it
- Needs a senior or another team
- Next action
- Send with notes; tell the caller the next step
Code: TRF
- Label
- Transferred
- When to use it
- Passed live to another person; see [call transfer](/glossary/call-transfer/)
- Next action
- Note who took it
Code: RPT
- Label
- Repeat contact
- When to use it
- The person called about the same issue before
- Next action
- Link to the earlier case; flag for review
How do you log DND and do-not-call requests?
Give compliance outcomes their own codes, because each one needs a different action from every other disposition. This is not legal advice. Our guide to cold calling in India says to stop, log the request and remove the number from your list that day. A person can register on the national DND list by calling or texting 1909, and can block all commercial calls or only some categories (TRAI on unsolicited commercial communication). Read the glossary entry for DND, TRAI's new rules for business calls and our guide to the DPDP rules for call centers for what applies to you.
Code: DNC
- Label
- Do not call me
- When to use it
- The person asks you to stop calling
- Next action
- Stop; remove from every list the same day
Code: DNDR
- Label
- Says registered on DND
- When to use it
- The person says they are on the DND list
- Next action
- Check the list, end the call politely
Code: CWD
- Label
- Consent withdrawn
- When to use it
- The person takes back a consent they gave
- Next action
- Record it and stop the calls it covered
Code: CMP
- Label
- Complaint
- When to use it
- The person raises a complaint, not a query
- Next action
- Log with detail; send to the complaints owner
Code: ABU
- Label
- Abusive caller
- When to use it
- Abusive or threatening speech to the agent
- Next action
- End the call; note it for the supervisor
Code: REC
- Label
- Objects to recording
- When to use it
- The person does not want the call recorded
- Next action
- Follow your own recording rule and flag it
Which codes suit e-commerce calls?
Use these for order and delivery calls. Our guides to COD order confirmation calls, NDR in e-commerce and RTO in e-commerce explain the flows. The outcome has to go back to the system the courier reads, or the call changes nothing.
Code: CODC
- Label
- COD confirmed
- When to use it
- The buyer confirms they will take and pay for the order
- Next action
- Release for dispatch
Code: CODX
- Label
- COD cancelled
- When to use it
- The buyer cancels the order
- Next action
- Cancel and release stock
Code: ADR
- Label
- Address corrected
- When to use it
- A fixed address or landmark was given
- Next action
- Write the new address back
Code: NDRR
- Label
- NDR re-attempt asked
- When to use it
- The buyer asks for another delivery attempt on a date
- Next action
- Send the date to the courier panel
Code: NDRF
- Label
- No attempt made
- When to use it
- The buyer says no delivery attempt was made
- Next action
- Raise it with the courier
Code: RTOA
- Label
- Return accepted
- When to use it
- The buyer agrees the parcel goes back
- Next action
- Let the RTO proceed
How many disposition codes should you keep?
Keep 8 to 12 outcome codes per campaign, each with at most 3 to 5 reason sub-codes, so about 20 to 30 choices in total. This is our own rule, not a standard. The reason is simple: an agent picks from one short list in a few seconds of wrap-up, and reports roll up cleanly because the outcomes stay few and the reasons stay inside them.
The only published figure we found is a balanced range of 10 to 15 codes, from MightyCall's guide to call dispositions (MightyCall). That is a vendor opinion, not a study. It says too few codes oversimplify the picture and too many slow the agent down.
One warning sign is also ours. When "Other" passes about 5% of your calls, a code is missing. Read a sample of the "Other" calls each month and add the code the notes keep asking for.
How should you name disposition codes?
Name a code for what happened, with one meaning and no overlap. These rules are ours, and they work on a floor of 5 or 500.
- One meaning per code. If two codes can both be true for one call, merge or split them.
- Say what happened, not what you hope. "PTP given" beats "will pay". A broken promise comes later, from payments.
- A short code and a plain label. The code is for reports and the label is for people.
- Attach the data the next step needs. A callback needs a date and time. A PTP needs a date and an amount.
- Keep one code set across languages. The code is the same whether the call was in Hindi, Tamil or English.
- No "Other" without a note. Make the note field required.
- Review monthly. Delete codes nobody uses and add the one that "Other" keeps hiding.
Weak: Will pay
- Why it fails
- States a hope, not an outcome, and has no date
- Better
- PTP given, with date and amount
Weak: Customer issue
- Why it fails
- Covers too many things to act on
- Better
- Dispute, or Ticket raised with a reason
Weak: Not interested / Do not call
- Why it fails
- Two meanings in one code, one needing list removal
- Better
- NI and DNC as two codes
Weak: Other
- Why it fails
- Hides what a missing code would have shown
- Better
- A named code, or Other with a required note
Which metrics come from disposition codes?
Most of the numbers on a floor are counts of codes divided by other counts of codes. Here are the ones that matter, with the formulas. Our call center metrics guide and our guide to payment reminder calls cover connect rate and promise-to-pay rate. Here we count promise-to-pay against right party contacts, so the base is one person, not any answer.
Metric: Connect rate
- Formula from codes
- Answered calls ÷ calls dialed
- What it tells you
- Whether the list, hours and numbers work
Metric: Right party contact rate
- Formula from codes
- RPC ÷ calls dialed (our definition)
- What it tells you
- How many dials reach the person you meant
Metric: Promise-to-pay rate
- Formula from codes
- PTP codes ÷ right party contacts
- What it tells you
- Whether the script gets a commitment
Metric: Kept-promise rate
- Formula from codes
- Promises paid on or before the date ÷ promises
- What it tells you
- Whether commitments are real. Set later, from payments
Metric: Conversion rate
- Formula from codes
- SALE ÷ RPC (our definition)
- What it tells you
- Whether the offer and the pitch work
Metric: Wrong-number rate
- Formula from codes
- WN ÷ connected calls (our definition)
- What it tells you
- How clean the list is
Metric: DNC rate
- Formula from codes
- DNC ÷ connected calls (our definition)
- What it tells you
- Whether the list or the pitch is irritating people
We use right party contacts as the base for promise-to-pay everywhere in this post, because a third party or a wrong number cannot make a promise. Whatever base you choose, keep it the same month to month.
What does this look like on a real desk?
Take a two-wheeler EMI desk in Nagpur, with 25 agents and 3,000 dials a day on overdue accounts. Every number is invented to show the arithmetic.
Code: Not connected (switched off, not reachable, no answer, busy, invalid)
- Calls
- 1,650
- Note
- Status, set by the system
Code: Connected, wrong number
- Calls
- 120
- Note
- List quality
Code: Connected, third party
- Calls
- 180
- Note
- No details shared
Code: Right party contact
- Calls
- 1,050
- Note
- The base for the rates below
Code: Of which PTP
- Calls
- 420
- Note
- A promise with a date and amount
Code: Of which says already paid (to verify)
- Calls
- 110
- Note
- Check before the next call
Code: Of which dispute
- Calls
- 60
- Note
- Goes to a person
Code: Of which refused or callback
- Calls
- 460
- Note
- Mixed outcomes
Here is the working:
- Connected calls: 120 + 180 + 1,050 = 1,350. The not-connected row is 3,000 − 1,350 = 1,650.
- Connect rate: 1,350 ÷ 3,000 = 45%.
- Right party contact rate: 1,050 ÷ 3,000 = 35% of dials.
- Promise-to-pay rate: 420 ÷ 1,050 = 40% of right party contacts.
- Check the breakdown: 420 + 110 + 60 + 460 = 1,050.
The kept-promise rate is missing on purpose. It is set days later, from payments, not on the call.
How do disposition codes get gamed?
Agents pick the fastest code, or the one that protects their numbers. A hard call gets logged as "wrong number" because that is one click. A weak call becomes "callback" so it does not count as a refusal. The fix is not a better dropdown. Audit a sample of dispositions against the recording each month, and compare it to what the agent chose. The scorecard in our guide to call audit in BPO has a line for exactly this, "outcome code and notes correct". Our call center metrics guide lists the habits that stop other numbers from being gamed.
How do auto dialers and AI voice agents set dispositions?
A dialer logs the status for you and asks the agent for the outcome. That is the split from the two-level tree, and it is what to ask for when you compare tools. Our guide to what an auto dialer is has the checklist, including whether reports show the outcome per campaign.
An AI voice agent reads the whole conversation and chooses an outcome. That removes typing, but it adds a new risk: a model can pick the wrong code, with the same confidence as the right one. So audit its choices the way you audit a person. Pull a sample, listen, and compare. Test the agent on tricky calls before launch, as our guide to how to test a voice agent explains, such as a customer who says "do not call me again" in Hindi, and check that it picks the right code.
How Bananaflow fits
Bananaflow is an AI voice agent platform. Every call its agent takes ends with a disposition, a recording and a transcript, and call outcomes are recorded as dispositions, which are outcome codes. Each agent has its own list of codes, each with criteria you write. When the call ends, the AI picks one code from that list in a separate final step, and a cautious default code is stored first so no call is left without one. The end-call and transfer tools can also set a code, and you can map your codes to the codes in your own dialer or CRM. As described on our e-commerce page, a call returns a call disposition along with a transcript and a recording.
After each call, a webhook can send the result to your own system, and call events can export to BigQuery. The analytics Agent Runs list shows the disposition, the duration, the direction and the number, and filters by date, agent, campaign and channel. The CSV export has one row per call, including the outcome. Campaigns run from a CSV with retries, calling time slots and concurrency limits, and an A/B split shows outcomes for each agent version. The disposition also appears in the CSV export, the runs API filter and the success rates in reports. A QA step can score a finished call and return tags, sentiment and a summary (Call QA).
The disposition codes in this post are our suggestions for your floor. You write the list and the criteria for each agent.
What should you do next?
Start small. This week, choose your campaign and write 8 to 12 outcome codes from the tables above, with three reasons at most for each. Decide which codes the system will set and which a person will pick. Write the next action beside each code, and make the date and amount fields required where they matter. Then audit 20 calls against their codes in a month, and fix the first code that "Other" keeps hiding. A disposition list is only as good as the actions it triggers.
FAQ
Frequently asked questions
What is the meaning of call disposition?
A call disposition is the short code an agent or system records after a call to say how it ended, such as callback, not interested or promise to pay. It records the result, can trigger the next action (a retry, a callback, removal from a list), and feeds the numbers in your reports.
What is PTP in collections?
PTP is promise to pay. In collections it is the code for a call where the customer agreed to pay on a date. Record the date and the amount with it. Whether the promise was kept is set later, from payments, never on the call itself.
What is the difference between call status and call disposition?
Status is the network's answer: no answer, busy, switched off, not reachable, invalid number or voicemail. Disposition is what the conversation achieved, such as interested, promise to pay or callback. In our view the system should log status and the agent, or an AI agent, should choose the disposition, so the two never get mixed in one list.
How many disposition codes should a call center have?
Our own rule is 8 to 12 outcome codes per campaign, each with at most 3 to 5 reason sub-codes, so about 20 to 30 leaves in all. This is our advice, not a standard. The only published figure we found is 10 to 15 codes, from the MightyCall guide to call dispositions, and it is a vendor opinion. If more than about 5% of your calls land on Other, a code is missing.
What is RPC in a call center?
RPC is right party contact: the call reached the person you meant to speak to, not a relative, a colleague or a wrong number. It is the base for most collections numbers. Right party contact rate is the right party contacts divided by the calls dialed, as a percentage.
Should do not call be its own disposition code?
Yes. A do-not-call request needs a different action from every other outcome: stop calling and remove the number from your list the same day. If you log it as not interested, the number is dialled again next week. Keep a separate code for people who say they are on the DND list.
Can an AI voice agent set the disposition?
It can read the whole conversation and pick an outcome, but a model can still choose the wrong code. Check a sample of its choices against the recordings, the same way you would audit a person. Bananaflow returns a call disposition with every call it takes, picked from the code list you set for that agent, and you should review a sample of those too.



