DID Number Provider: What a DID Is, How It Works, and How to Pick One

1. What a DID number is (and why “DID” survived)

DID stands for Direct Inward Dialing. The term comes from 1960s analog telephony, when large organisations started buying blocks of phone numbers so that outside callers could reach a specific extension inside the building without going through a switchboard operator. A “DID number” was any one of those externally-dialable numbers that mapped directly to an internal endpoint.

In modern SIP and cloud telecom the analog hardware is long gone, but the concept is intact. A DID is still a phone number you rent from a carrier, and it still routes inbound calls directly to whatever endpoint you tell it to. The endpoint is now a SIP trunk, an AI voice agent, an IVR, a webhook, or a Microsoft Teams user, rather than a physical phone on a copper line. Everywhere you see the term “virtual number,” “VoIP number,” or “business phone number” in marketing, the underlying object is a DID.

Every DID has three attributes that matter operationally:

• An E.164 number: the externally-dialable address, written with a leading + and the country code, e.g. +442071838750. This is what people punch into their phones.
• A routing target: where calls are delivered when someone dials that number. A SIP trunk URI, a registered SIP endpoint, a BYOC integration with Vapi or Retell, an HTTP webhook for an AI agent, or a forwarded PSTN number.
• A set of capabilities: voice (always), inbound SMS (sometimes), outbound SMS (less often), fax over T.38 (rarely), caller-ID name registration. These are determined by the number type and the country.

Two clarifications that catch people. First, a DID has nothing to do with whether it “goes through the internet” or “is a real phone number.” It is always a real, regulator-registered phone number; the delivery to your infrastructure happens over IP, but the caller has no idea. Second, the number itself does not encode any capability. +1 415 555 0100 tells you nothing about whether it can receive SMS. That depends on which carrier owns it and what they have provisioned, not the digits.

2. What a DID provider actually does

The phone-number supply chain has three layers, and where your provider sits in that stack determines almost everything about pricing, coverage, and how flexible they can be with routing.

Tier 1 carriers operate the physical telecom infrastructure that connects to the public switched telephone network in a given country. Verizon, AT&T, BT, Deutsche Telekom, Orange, NTT. They own number ranges directly from the national regulator and they exchange traffic with other Tier 1s.

Wholesale aggregators sit on top of multiple Tier 1s. They contract with several carriers per country to assemble large multi-country DID inventories and sell them on as a single SKU. Inteliquent (now Sinch), Bandwidth, iBasis, and the former Voxbone (now part of Bandwidth) are classic examples. They abstract away the carrier-by-carrier integration work.

Retail DID providers are who you actually buy from. They source numbers either from Tier 1s directly or via aggregators, layer on a portal and an API, handle regulatory documents on your behalf where they can, and bill you a unified monthly statement. DIDHub is in this layer, as are Twilio, Telnyx, Bandwidth (which spans aggregator and retail), DIDWW, and dozens of regional specialists.

Some companies span multiple layers. Bandwidth is its own Tier 1 in the US and goes through partners internationally. Twilio leans aggregator-plus-retail. The reason this matters: a pure retail reseller has thinner margins, less control over compliance turnaround, and may go through three or four hops to source a number in a regulated market. A provider that owns inventory in more markets directly typically has shorter activation times, cleaner porting, and more honest answers when something breaks.

A retail DID provider does, concretely:

• Procures DID inventory in the countries it advertises, either via direct Tier 1 contracts or wholesale partners.
• Handles know-your-customer (KYC) and regulatory document submission per country, since regulators almost always require proof of identity and a local address before activating a number.
• Operates the SIP infrastructure that bridges the public phone network to your IP endpoints: edge SBCs in multiple geographies, routing engine, channel control, call-recording, real-time billing.
• Provides the tooling you actually interact with: portal, REST API, webhooks for inbound events, billing dashboard, support.
• Handles porting in (taking over existing numbers from another carrier) and porting out (releasing numbers when you leave). The quality of porting support is where providers most differentiate.

3. How a call to a DID works, step by step

The clearest way to understand a DID is to follow a single call from the caller’s handset to your servers. Take an inbound call from a UK mobile user to a London geographic DID, +442071838750, that you have rented from a provider and routed to your AI voice agent over BYOC.

1. Caller dials. The mobile handset signals the dialed digits to its serving carrier, in this case Vodafone UK. Vodafone’s switch recognises +44 20 as the London geographic area code and routes the call through the UK PSTN towards the carrier that owns the specific number.
2. PSTN delivery to the owning carrier. Number ownership is registered with the national regulator (Ofcom in the UK). The owning carrier is either your DID provider directly (if they are a Tier 1) or the Tier 1 partner your provider sources +44 20 numbers from. The PSTN delivers the call to that carrier’s switch over SS7 / TDM, or increasingly over a SIP interconnect.
3. Routing lookup at the carrier. The carrier’s switch consults its routing table for that exact number and finds: “deliver this to DIDHub’s edge SBC.” It hands the call off as a SIP INVITE.
4. SIP INVITE arrives at your provider’s SBC. The INVITE’s Request-URI carries the dialed number (+442071838750). The From header carries the caller’s number, subject to caller-ID privacy rules. Headers carry signaling metadata; SDP carries the proposed media parameters.
5. Your provider’s routing engine resolves the destination. It looks up which customer owns +442071838750 and what their delivery method is. Options include: registered SIP trunk (PBX has registered to the provider), IP-ACL trunk (provider sends INVITEs to a pinned customer IP), BYOC delivery to a SaaS platform (Vapi, Retell, Microsoft Teams, Zoom Phone), or HTTP webhook for fully programmable agents.
6. Provider forwards the INVITE. For SIP-based delivery, the provider sends a fresh INVITE to your endpoint with the original caller information preserved in headers. For HTTP-based delivery, the provider POSTs a JSON event to your webhook and waits for your response with a TwiML-style instruction or a SIP URI to bridge to.
7. Your endpoint accepts. Your PBX, agent, or app responds with 200 OK and SDP for the media. Once both sides have agreed on codec, ports, and SRTP keys, media flows. Audio (RTP) travels between the caller’s carrier and your endpoint, brokered by SBCs along the way to handle NAT traversal, codec transcoding, and encryption.

Total round-trip latency for this whole flow is usually 200 to 800 milliseconds, dominated by geographic distance and any transcoding. The caller hears ringback within that window.

Two structural points worth holding in mind. The PSTN-to-SIP handoff happens at your provider’s SBC: everything upstream of that point is the traditional phone network with its own signaling and economics; everything downstream is your IP infrastructure with its own. And the routing target you set in your portal or via the provider’s API is what step 5 consults. Change the target, and the very next inbound call goes to the new destination. There is no “rebuild” or downtime; the routing engine looks up the current target at call time.

If you want to go deeper on the SIP side of the flow, see our guide on how SIP engines decide where to send an INVITE and the SIP trunk glossary entry.

4. Multiple concurrent calls on one DID

The single biggest misconception about DIDs is that one number can only handle one call at a time. That is not how it works, and the confusion costs people money in over-provisioning and downtime when callers hit a phantom busy signal.

A DID is an address. How many simultaneous calls it can accept is governed by the channel capacity of the trunk it routes into, not the number itself. Three concepts to keep separate:

Practical example. You run a US contact centre with a 50-channel SIP trunk and 200 geographic DIDs, one per state for local-presence dialing. At peak, you can have 50 concurrent inbound calls in any combination across those 200 numbers: it could be 50 calls to one popular DID, or one call to each of 50 different DIDs, or anything in between. Call 51 hits the trunk capacity wall. The provider returns a busy signal or routes to your overflow destination, depending on how you have configured failover.

This is why DIDs and channels are almost always sold as separate line items on your invoice. They scale independently. You might have 10,000 DIDs (cheap, ~$1 each per month) and 50 channels (the actual call capacity). Or you might have 1 DID and 500 channels (a single hotline that handles huge concurrent inbound, like a TV-show vote line).

Some providers offer elastic channels with per-minute billing: no fixed channel cap, you simply pay for actual minutes used. Twilio popularised this model. It is more expensive per-minute but eliminates capacity planning entirely. Useful when traffic is spiky and unpredictable.

Rough capacity-planning rule. For a normal inbound-call business, you need roughly one channel per 60 to 100 inbound minutes during your busiest hour. A business doing 1,000 minutes a day across an 8-hour window peaks at around 3 to 5 channels. AI voice agents are different: every interaction is a real-time call, so plan one channel per concurrent conversation you expect, plus 20% headroom.

5. Types of DIDs in one screen

Five types you will actually encounter, with the facts that drive the decision. For the deep dive on each, see DID number types explained.

Two type decisions trip almost everyone up. First, geographic landline DIDs usually cannot send SMS reliably, even where the carrier technically supports it. If your use case is texting, default to a mobile DID. Second, toll-free is not free for you: the caller pays nothing, but you pay per inbound minute. A high-volume support line on a US toll-free can run $200-500/month in inbound minutes alone, on top of the DID rental.

6. Where DIDs are used

The same product (a phone number you rent and route) ends up powering radically different workloads. The most common buckets you will see in the wild:

• Contact-centre-as-a-service (CCaaS): fleets of geographic DIDs per region for local presence, with inbound routed to ACD queues and outbound caller-ID rotation across many numbers to avoid spam-labeling. See provisioning numbers at scale.
• AI voice agents: one DID per agent or per customer, delivered via BYOC into Vapi, Retell, ElevenLabs, OpenAI Realtime, or a custom WebRTC stack. The DID is what the caller sees; the AI is what answers.
• SMB business phone lines: a single DID per location, routed to a softphone, a desk phone, a hosted PBX, or directly to Microsoft Teams or Zoom Phone via BYOC.
• IVR and auto-attendant: one DID per language, brand, or business unit, delivered to a hosted IVR (Asterisk, FreeSWITCH, or a SaaS like CXone).
• A2P SMS: a single DID (typically mobile or US toll-free) used to send 2FA codes, alerts, marketing, or transactional notifications. US traffic requires 10DLC registration; see A2P 10DLC registration.
• Local presence for global brands: a US SaaS buys DIDs in UK, Germany, Australia, Japan. Customers in each country dial a local number; calls route back to the US support team. Reduces caller cost and dramatically improves answer rates.
• Redundancy and failover: a primary DID routes to your main PBX; a secondary DID routes to a backup softphone or voicemail-to-email when the primary trunk is unreachable. See SIP trunk failover.

7. Pricing models you will see

Three models dominate, often combined:

1. Per-DID monthly plus per-minute usage. The most common. You pay a flat monthly rental per number ($0.50 to $3 for common geographic types, $5 to $50 for regulated international or toll-free) plus a per-minute rate for inbound and outbound traffic. Inbound on a geographic DID is often free or near-zero; toll-free inbound is typically $0.02 to $0.06 per minute in the US.
2. Channel-based. You pay per concurrent channel as well as per DID. Cheaper at scale because you size the channel pool to peak rather than paying per-minute, but you do pay for capacity you might not use.
3. Pooled minutes. A bundle of N inbound minutes per month is included with each DID rental; overage at a per-minute rate. Common for SMB-friendly plans.

Two pricing signals to watch for. International rates vary by an order of magnitude: a UK geographic DID at $1.50/mo can sit next to a Pakistan DID at $30/mo on the same rate card. Always pull the actual per-country sheet. And setup fees should be zero in 2026: any provider charging a per-DID setup fee is overpriced or is sourcing through too many wholesale hops.

8. Compliance, KYC, and activation timelines

Most countries legally require proof of identity and a local address before a phone number can take its first call. Strictness varies wildly:

• Easy (instant or same-day activation): US, Canada, most Caribbean. A 911 address suffices in the US/Canada.
• Address required (1 to 3 days): UK, most of Western Europe, Australia, Singapore. Standard KYC bundle: proof of identity plus a recent utility bill or lease for an in-country address.
• Strict regulatory markets (1 to 2 weeks, often rejected on minor mismatches): Germany, France, Italy, India, Brazil. Document quality matters; addresses must match exactly.
• Mobile DIDs restricted: many countries do not sell mobile-range DIDs to non-carriers at all. Where they do, the KYC is typically stricter than the country’s landline tier.

For the full country-by-country breakdown and the documents involved, see International DID compliance & KYC bundles.

9. Porting in and porting out

You will almost always need to move existing numbers from one provider to another at some point. The process is called porting, and it is a regulated handshake between the losing carrier and the gaining carrier.

Typical timelines:

• US local geographic: 2 to 5 business days
• US toll-free: 4 to 7 business days
• UK geographic: 7 to 10 business days
• International: 2 to 6 weeks, varying wildly by country and losing carrier

The single biggest cause of port rejection is a data mismatch on the Letter of Authority (LOA): account number, service-address details, or PIN that does not exactly match the losing carrier’s record. Pull the customer service record from the losing carrier first and copy details verbatim. For the full walkthrough, see number porting explained.

One practical signal when evaluating providers: how easy do they make porting out? Some providers make porting in seamless and porting out a multi-week ordeal designed to make you stay. A provider confident in their product will publish a one-page porting-out process and execute it in days, not weeks.

10. How to choose a DID provider (10-point checklist)

The evaluation checklist that catches the differences that actually matter:

1. Coverage by country and type. Not the marketing “150+ countries” headline. Check the actual per-country availability page for the specific number types you need. Geographic only? Mobile required? Toll-free?
2. Channel scalability. Can you scale to your peak channel count without negotiating a custom contract? Is there a hard cap?
3. Pricing transparency. Public rate card per country, or “contact sales for pricing”? The first is honest; the second usually means “we will price-discriminate based on what you tell us.”
4. API quality. Search, order, route, release, and report programmatically. Look for a published OpenAPI specification, webhook events for inbound calls and SMS, and idempotency support on the order endpoint.
5. Compliance support. Do they handle document submission and follow-up for strict markets on your behalf, or do they hand you the regulator’s form and disappear?
6. Porting support, both ways. Ask explicitly about port-out turnaround. A provider that hedges here is signaling lock-in.
7. Routing flexibility. Register-based SIP, IP-ACL, BYOC to SaaS platforms, HTTP webhooks. Per-number routing or trunk-level only?
8. SLA and status page. Historical uptime, MTTR for incidents, transparent status communication. A public, detailed status page is a strong signal.
9. Support response. 24/7 NOC for severity-1 issues, or business-hours email tickets? AI-only first-line support is a red flag.
10. Lock-in risk. Long minimum-term contracts, proprietary call-control APIs that require rebuilding to leave, exclusive number ranges. The right answer is “monthly cancellation, standard SIP delivery, standard porting.”

Two flags that should rule a provider out for most buyers: multi-year minimum-term contracts (you should be able to cancel monthly), and the absence of a publicly-listed API specification (signals an internally-tooled provider where everything goes through their portal, which limits automation and scales badly).

11. Where DIDHub fits

DIDHub is a DID number provider focused on three things. First, AI voice infrastructure: one-click BYOC to Vapi, Retell, ElevenLabs, OpenAI Realtime, Bland, Cartesia, Deepgram, LiveKit, Pipecat, Synthflow, plus every standard PBX (Asterisk, FreeSWITCH, FreePBX, 3CX, Kamailio, OpenSIPS) and the BYOC SaaS platforms (Microsoft Teams Direct Routing, Zoom Phone). Second, wholesale-friendly pricing: no setup fees, monthly cancellation, public per-country rate card. Third, a programmable API that scales from a single number to ten thousand, with OpenAPI specification, idempotent bulk ordering, webhook events, and porting both ways.

130+ countries, channels billed separately from DIDs, instant activation on most markets, full porting support. Geographic DIDs from $1.99/mo. Get a number, browse the rate card, or talk to sales for pooled-channel arrangements.

12. FAQ

What is a DID number?

A DID number is an externally-dialable phone number that routes inbound calls directly to an endpoint of your choosing: a SIP trunk, a PBX extension, an AI voice agent, an IVR, or a webhook. In modern usage it is synonymous with “virtual number,” “VoIP number,” or “business phone number.” You rent DIDs from a DID provider and configure where calls go via their portal or API.

What does DID stand for?

DID stands for Direct Inward Dialing. The term originated in 1960s analog telephony, when organisations bought blocks of numbers so outside callers could reach internal extensions without going through a switchboard operator. The hardware is gone; the term and the concept survived.

How does a DID number work?

When someone dials your DID, their carrier routes the call through the public phone network to the carrier that owns the number range. That carrier hands the call off as a SIP INVITE to your DID provider’s SBC. Your provider looks up where you have routed that specific number and forwards the call to your endpoint, whether that is a SIP trunk, an AI agent, or a BYOC SaaS like Microsoft Teams. Total round-trip is usually 200 to 800 milliseconds.

Can multiple calls come in on the same DID number?

Yes. A DID is an address, not a call-handling capacity. Multiple simultaneous calls to one DID is normal and expected. Capacity is governed by the channel count on the trunk the DID routes into. A 50-channel trunk with 1 DID can carry 50 concurrent calls to that one number; a 50-channel trunk with 200 DIDs can carry 50 concurrent calls split across any combination of those numbers.

What is the difference between a DID and a SIP trunk?

A DID is a phone number. A SIP trunk is the IP delivery pipe between your provider and your infrastructure. One trunk can carry many DIDs. DIDs are addresses; the trunk is the connection. You buy them as separate line items.

How much does a DID number cost?

Common geographic DIDs cost $0.50 to $3 per month. Toll-free DIDs cost $1 to $5 per month plus per-minute inbound charges ($0.02 to $0.06 in the US). International DIDs vary by an order of magnitude: a UK geographic DID can be $1.50/mo while a Pakistan DID is $30/mo. Always check the actual per-country rate card.

Can I send SMS from a DID number?

It depends on the type. Mobile DIDs almost always support SMS. Geographic landline DIDs in most countries do not support SMS reliably. Toll-free DIDs in the US support SMS only after a verification process. If your use case is SMS, default to a mobile DID and check the specific country’s rules.

Can I port my existing phone number to a new DID provider?

Yes, almost always. Porting is a regulated handshake between your losing carrier and your gaining carrier. Timelines: 2 to 5 business days for US local, 4 to 7 for US toll-free, 7 to 10 for UK geographic, 2 to 6 weeks for most international. The single biggest cause of port rejection is mismatched information on the Letter of Authority. Pull your customer service record first.

How long does DID activation take?

For new (non-ported) numbers: instant to same-day in the US, Canada, and most Caribbean countries; 1 to 3 days in the UK and most of Western Europe; 1 to 2 weeks in strict regulatory markets like Germany, France, Italy, India, and Brazil. The variation is driven by KYC document review at the regulator level, not by the provider.

What is the difference between a DID provider and a reseller?

A DID provider sources numbers either directly from Tier 1 carriers (who own the actual telecom infrastructure) or via wholesale aggregators. A reseller is a thinner layer on top of an existing provider: they take your order, pass it to their upstream, and mark up the price. Resellers typically have less control over compliance turnaround, fewer routing options, and longer support escalation chains when something breaks. The distinction is sometimes hard to see from the outside; the test is whether they own customer relationships with regulators in the countries they sell.

13. Bottom line

A DID is just a phone number you rent and route programmatically. A DID provider sources the numbers, handles the regulatory compliance, and runs the infrastructure that bridges the public phone network to your IP endpoints. Good providers are transparent about coverage, pricing, channel capacity, and porting; bad ones bury those details behind a sales call. Channels and DIDs scale independently, and you almost always need fewer channels than you think. Pick on coverage, pricing transparency, API quality, and porting support, in that order.

More from the blog