How Multi-Site Estates Actually Form
It is worth naming the pattern, because recognising it removes the temptation to treat the current state as somebody's mistake.
| Stage | What happens | What it leaves behind |
|---|---|---|
| Site one | A phone system is chosen properly, with some thought. | A sensible baseline nobody writes down. |
| Site two | Opens under time pressure. Somebody local arranges a service because the fit-out is next week. | A second provider, a second bill, a second admin login nobody at head office has. |
| Sites three to six | Each repeats site two, with whatever was quickest locally. | Different handsets, different voicemail conventions, different greetings, four different definitions of "missed call". |
| An acquisition | A site arrives with its own system, its own numbers and a contract with eighteen months to run. | A platform nobody chose, numbers possibly registered to the former owner, and an exit date nobody has diarised. |
| A franchisee change | A site changes hands. The new operator sets up their own service. | A live advertised number still registered to the previous operator — the single most common serious problem in franchise estates. |
| Head office notices | Usually when a customer complains about being unable to reach a branch, or when someone tries to build a network report. | The realisation that there is no single source of truth, and no single place to change anything. |
Nothing in that sequence was irrational. Every decision was the fastest correct decision available at the time, made by someone with a fit-out deadline. Which is why the fix is not a process document telling sites to consult head office. It is an architecture where the fast local decision and the coherent network decision are the same decision.
What Fragmentation Costs
Four costs, in rough order of size. The first is the one on the invoice and usually the smallest.
Duplicated spend
Minimums, line rentals, per-site maintenance contracts and per-site support fees, multiplied by the number of locations. Real, visible, and typically the least of it.
Unmeasured missed calls
The big one. With no network reporting, missed calls at each site are invisible individually and enormous collectively. A handful per site per day across eleven sites is thousands of lost contacts a year, and nobody is accountable because nobody can see it.
No load sharing
Every site is staffed for its own peak, so every site is over-staffed at 11am and under-staffed at 9am. Sites in different time zones cannot cover each other at all, which is capacity you already pay for and cannot use.
Inconsistent experience
A customer who deals with two of your locations encounters two different businesses. Different greetings, different hold behaviour, different chances of reaching anyone after five. Brand consistency ends at the phone line.
A quick way to size the second cost
For each site, take last month's inbound calls that rang out, went to voicemail with no message, or were abandoned in a queue. Add them up across the estate. Multiply by whatever proportion you believe were genuine enquiries, and then by your average transaction value. Most organisations doing this for the first time find the answer is larger than the entire annual telephony spend — which reframes the project from cost reduction to revenue recovery. Our note on the metrics that matter sets out how to count these consistently.
What One Platform Across Sites Looks Like
The shape is simple: one platform, one administrative view, and configuration scoped per site. Nothing is on any site except handsets, apps and a router.
| Element | How it works across sites |
|---|---|
| Sites as objects | Each location exists in the platform with its own numbers, users, devices, hours, greetings, queues and holidays. Change one without touching the others. |
| One dial plan | Extension ranges per site, and any extension can ring any other extension anywhere at no cost. A four-digit dial between Perth and Brisbane is an internal call. |
| Local identity, central capability | Each site presents its own local number outbound and answers its own local number inbound, while sharing queues, routing, recording and reporting with the whole network. |
| Delegated administration | A site manager can change their own hours, greetings and staff. They cannot change another site, and cannot change what the network has fixed. |
| Portable users | A person who covers three locations is one user with one identity. They log in wherever they are; calls follow them. No second handset, no second licence. |
| One set of reports | Per site, per region, and consolidated, from a single source with a single definition of every metric. |
This is a straightforward consequence of the architecture rather than a premium feature: once the exchange is a shared platform rather than a box in a comms room, a site is a configuration boundary instead of a piece of hardware. The background is in what a PBX actually is and the five things called one.
The Inbound Number Decision
This is the first real design decision and it is worth thinking about properly, because it shapes both the customer experience and your reporting.
| Strategy | Best for | Watch out for |
|---|---|---|
| Local number per site, published | Businesses where customers deal with a specific location — clinics, dental practices, real estate offices, trade depots, restaurants. | Each number needs its own answering path and its own after-hours plan. A published number that rings out is worse than no published number. |
| One network number, routed | Retail chains, service businesses where any site can help, and anything advertised nationally. Route by caller location, time of day or queue depth. | Callers who want a specific branch will resent being routed elsewhere. Give them a path to a named site. |
| Both — the usual answer | Most estates. Local numbers for people who know which branch they want; a 1300 or 1800 for advertising and for overflow. | Reporting must attribute correctly, or the network number swallows all your attribution and you cannot tell which site generated the enquiry. |
| Local numbers for sites you do not have | Genuine remote service areas you actually cover. | Do not present a local number for a place you do not serve and cannot answer for. It is a promise, and callers treat it as one. |
The number ownership problem, which is specific to franchise networks
In a franchise estate, published local numbers are frequently registered to the franchisee, the previous franchisee, or a company the network has no relationship with. When a site changes hands, the number can leave with the departing operator — taking with it every directory listing, every printed reference, every review page and every inbound call from a customer who has had that number in their phone for six years. Whoever holds the rights of use decides the fate of the number, and that is a matter of record, not of who pays the bill. Settle it deliberately: it is entirely reasonable for the network to hold rights of use over numbers used for network-branded marketing, and for the franchisee to hold their own. What is not reasonable is nobody knowing. The mechanics are in who owns your 1300 number.
Overflow Between Sites
This is the single largest operational gain and the reason most multi-site consolidations pay for themselves. Four patterns, in increasing order of sophistication.
| Pattern | Rule | Where it earns its keep |
|---|---|---|
| Sequential fallback | Unanswered at site A after N seconds, try site B, then a central team. | The simplest possible improvement, and it eliminates most ring-outs immediately. Start here. |
| Time-zone coverage | Calls to eastern-state sites before 8:30am go west; calls to western sites after 3:00pm eastern go east. | Australia's three-hour spread is an asset almost nobody uses. It extends effective opening hours without a single extra roster line. |
| Capacity-based | Route to the site with the shortest queue or the most available staff, not the nearest. | Evens out load across an estate where peaks land at different times. Requires shared queues and honest presence. |
| Skill-based across sites | One qualified person in the network can serve every site — a specialist, an interpreter, a licensed adviser. | Scarce skills become network-wide instead of accidental to one location. |
The overflow rule that needs a policy, not a setting: what a receiving site is allowed to do with another site's call. Take a booking on their behalf? Quote? Complete a sale and credit it where? Franchise networks in particular must answer this before enabling overflow, because the technical change is five minutes and the commercial argument afterwards is not. Decide the revenue attribution first, then switch it on.
Reporting That Rolls Up
Fragmented estates cannot compare sites, because each site's system counts differently. One platform means one definition, which is worth more than any individual metric.
| What to report | Per site | Network |
|---|---|---|
| Answered within target | The core site measure. Set one target for the network and let sites see their own position against it. | Weighted across the estate, and the distribution — not just the average, which hides two failing sites behind nine good ones. |
| Unanswered and abandoned | Ring-outs, voicemails with no message, queue abandons. The revenue leak. | Total, and by hour of day, which is what tells you whether the problem is staffing or routing. |
| Overflow volume and direction | Calls sent away, and calls received from other sites. | The map of which sites subsidise which. Essential in a franchise, uncomfortable, and better known than not. |
| After-hours volume | Calls arriving outside opening hours and what happened to them. | Usually the biggest surprise in the first report, and the cheapest thing to fix. |
| Repeat callers | The same number calling twice in a day is a first-contact failure. | Network-wide repeat rate, which is a better proxy for customer experience than any satisfaction survey you will get returned. |
Publish the comparison, carefully
Site-by-site league tables change behaviour fast, and not always in the direction intended — the fastest way to improve an answer-rate ranking is to answer and transfer immediately. So pair any answered-within-target measure with a repeat-caller or resolution measure, and be explicit that both are being watched. Otherwise you will get exactly what you measured.
The Franchise Governance Question
Everything above applies to any multi-site business. Franchise networks have an additional problem, and it is not technical: the sites are separate businesses, owned by people who took commercial risk, and telephony sits directly on the boundary between the brand and the operator.
| The tension | The franchisor's legitimate interest | The franchisee's legitimate interest |
|---|---|---|
| Caller experience | A customer ringing any location should meet the brand they were advertised. Inconsistency damages every site. | Their local market has its own rhythms, and a script written centrally can sound absurd in their suburb. |
| Numbers | Network marketing drives calls; those numbers should not walk out with a departing operator. | They built a local customer base and their local number is part of the goodwill they paid for. |
| Data | Network-level insight requires the data. Recordings help with brand-standard compliance. | Their customer list and call recordings are their business records, and their liability. |
| Cost | Buying once for the network is cheaper for everybody. | A mandated supplier at a mandated price is a cost they cannot negotiate, and they will notice if it is above market. |
The mistake that stalls these projects
Announcing a network telephony standard as an IT rollout. It is not an IT rollout; it is a change to the operating agreement in substance, whatever it is called procedurally. Networks that get this right present it as a commercial proposition — here is the price, here is what it replaces, here is what you gain, here is what is fixed and here is what remains yours — and they present the reporting to the franchisees first rather than to the board. A franchisee who can see their own missed-call number becomes the project's advocate. A franchisee who learns their calls are being measured before they see the measurement becomes its opponent.
What to Standardise, What to Leave Alone
A defensible split. The principle: standardise what a customer experiences and what must be comparable; leave what a local operator must be able to change on a Tuesday.
Standardise centrally
The platform. The greeting structure and how the brand is identified. The maximum time a caller waits before something happens. Metric definitions. Rights of use over network marketing numbers. The compliance baseline — recording notification, retention minimums, emergency address accuracy. Reporting visibility.
Leave with the site
Opening hours and holidays. Staffing and rosters. Who answers what. Voice of the greeting, within the structure. Local promotions and messages on hold. Day-to-day routing changes. Their own local number, and their own customer relationships.
Negotiate explicitly
Overflow participation and revenue attribution. Who may listen to whose recordings. Whether network staff can see site-level detail or only aggregates. What happens to numbers and data when a site changes hands.
Write it down once
Two pages, in the franchise documentation, in plain words. The absence of this document is what turns every individual telephony question into a negotiation, permanently.
Compliance Across Entity Boundaries
A single platform spanning separate legal entities raises questions that a single-business estate does not. Worth settling in writing rather than discovering.
| Question | Why it matters across sites |
|---|---|
| Who is the recording party? | Recording law is state-based and turns on the parties to the conversation. When a franchisee's call is recorded on a franchisor-administered platform, be clear who holds the recording and on what basis. Our summary is in the call recording guide. |
| Whose customers are they? | Call records and recordings are personal information. If the franchisor can see a franchisee's customer detail, that is a disclosure between entities and needs a basis and a boundary. |
| Who answers an access request? | When a customer asks for their information, the entity that collected it is on the hook. Decide in advance which one that is and how the other assists. |
| Who holds retention responsibility? | Sector rules differ — health, aged care, NDIS and finance all carry their own retention expectations, and they attach to the operating entity, not to the platform. |
| Automated decisions | From 10 December 2026, using personal information in automated decisions capable of affecting a person's rights or interests carries a privacy-policy disclosure obligation. If routing or qualification is centralised, both entities need to know whose policy says what. See the December 2026 obligation. |
None of these are reasons not to consolidate. They are reasons to write two paragraphs before you do, and they are far easier to settle at design time than after a customer complaint has arrived at both entities at once.
Emergency Calls, Per Site
This deserves its own section in a multi-site article, because it is the failure with the worst consequences and the easiest cause.
On a cloud platform, a user's number is not tied to a physical place — that is the point of it. So the address associated with each number is a record somebody must maintain, and in a multi-site estate that record is wrong more often than in a single office, for obvious reasons: sites open, close, relocate, and staff move between them.
| Do this | Why |
|---|---|
| Hold a service address per site, not per organisation. | An emergency call from the Bendigo branch should not present a Melbourne head office address. |
| Review addresses whenever a site opens, closes or moves, as a checklist item on the fit-out. | Relocations are where this breaks. Nobody thinks of the phone record when the lease changes. |
| Brief staff who work across sites and on mobile apps. | A person calling 000 from a laptop must state their actual location out loud. Assume they will not know this unless told. |
| Include it in the site induction. | The one telephony fact every staff member at every location should know. |
The full picture, including how camp-on and call-back work, is in Triple Zero from a cloud phone and what to tell staff about the camp-on delay.
Migrating an Estate Without a Big Bang
Never cut an entire network over at once. There is no upside and the downside is every site at the same time.
| Step | What to do | Why in this order |
|---|---|---|
| 1. Inventory | Every site: numbers, provider, contract end date, who holds rights of use, what equipment is on site, and who administers it. | This single spreadsheet is 80% of the project. It also surfaces the numbers registered to the wrong entity, which is the finding you want early. |
| 2. Design the standard | One site template: greeting structure, hours pattern, queue behaviour, after-hours path, reporting definitions. | Designing once and applying eleven times is the entire economic argument. Designing per site recreates the mess in a new platform. |
| 3. Pilot the awkward site | Not the biggest and not the easiest — the one with an unusual requirement. A clinic with an on-call roster, a depot with a paging system, a site with a legacy analogue device. | The template survives contact with the exception, or it changes now rather than at site nine. |
| 4. Sequence by contract expiry | Convert as each site's contract ends, not by size or importance. | Avoids paying twice. A three-year rolling programme with no overlap costs far less than a coordinated cutover. |
| 5. Enable overflow last | Get sites onto the platform first, individually working. Turn on inter-site routing only once the attribution policy is agreed. | Overflow before agreement produces a commercial dispute that stops the whole programme. |
| 6. Then publish reporting | Show sites their own numbers before showing anyone the comparison. | Reporting introduced as a tool is adopted. Reporting introduced as surveillance is resisted, and the difference is entirely in the sequence. |
Numbers are the part that needs the most lead time — porting runs to an industry framework with standard windows, and a site with numbers held by a third party may need a signed authority from an entity that no longer trades. Start that early. Our porting guide and provider-change guide cover the mechanics.
Three Shapes of Multi-Site Business
| What it looks like | What matters most | |
|---|---|---|
| Branches of one business Clinics, law firms, agencies, depots | One entity, several locations, staff sometimes shared. | Local numbers that are always answered, overflow to a central team after hours, and one report. The governance question barely arises — design it and switch it on. |
| A franchise network Food, services, retail, home services | Separate businesses under one brand, with an operating agreement between them. | The standardise-versus-leave-alone split, number rights of use at handover, and overflow attribution. The technology is the easy part; do the two-page document first. |
| A group by acquisition Roll-ups, practice groups, trade groups | Sites acquired with their own systems, contracts, numbers and habits. | Contract expiry sequencing and number provenance. Every acquisition brings numbers whose registration nobody has checked — check them at due diligence, not at migration. |
How We Set These Up
We run one platform on our own network, and a site in it is a configuration boundary rather than a piece of equipment. That has a few practical consequences worth stating plainly.
Each location gets its own local numbers, hours, greetings, queues and holidays, administered by whoever you nominate, with the network-level settings you decide to fix left unchangeable at site level. Extensions dial each other across the whole estate at no cost, four digits, regardless of state. Overflow between sites is a routing rule you can change yourself, by time, by queue depth or by availability. Reporting comes per site, per region and consolidated, from one source with one definition — so a comparison between two sites is a real comparison. Numbers we allocate are registered with rights of use to the entity you nominate, which is the mechanism that stops an advertised number leaving with a departing operator. And we will migrate site by site as contracts fall due, rather than asking you to coordinate a network-wide cutover that has no upside.
The part we will push you on is the two-page governance document, because we have watched enough of these to know that the estates which go badly are almost never the ones with a technical problem.
The summary
Multi-site telephony is an accumulation, not a design — which is why the costs are invisible and the largest one is unmeasured missed calls rather than duplicated spend. One platform with site-scoped configuration gives each location its own local identity while sharing routing, overflow, recording and reporting. Design five things: the inbound number strategy, overflow rules, what the caller hears, who administers what, and how per-site emergency addresses are maintained. In a franchise, do the governance first: standardise the caller experience, metric definitions, compliance baseline and rights of use over network marketing numbers; leave hours, staffing, greetings and local routing to the operator; and negotiate overflow attribution explicitly before enabling it. Migrate one site at a time in contract-expiry order, pilot the awkward site rather than the easy one, and show sites their own numbers before you show anyone a league table.
Related reading: the metrics that matter, who owns your 1300 number, changing provider without losing calls, what a PBX is and the five things called one, and how phone redundancy works.