Send the call to whoever should answer it

Generating the call is the easy half

Agencies spend enormous effort on the part of the funnel that ends when the phone starts ringing, and comparatively none on what happens in the four seconds after. It is an odd asymmetry, because the second part is cheaper to fix and frequently worth more.

The default arrangement on most tracked numbers is a single forwarding destination that never changes: the client's main line, all day, every day, from every source. That is fine when the client is a one-person business with a mobile in their pocket. It stops being fine the moment there is an office with opening hours, or a second location, or an out-of-hours service, or a sales team that should get the expensive paid-search enquiries while general reception handles everything else.

What happens instead, in practice, is that the client sets up a voicemail and considers the problem solved. It is not solved. A caller with a burst pipe who reaches a voicemail does not leave a message; they ring the next result. Somewhere between half and three-quarters of callers who hit an answerphone at a local services business never call back and never leave a message, and none of that shows up in a report unless somebody is looking for it.

Routing rules are the cheap fix for an expensive problem. They do not require the client to hire anybody, change their phone system, or install anything. They sit on the tracked number — which the agency already controls — and decide where it points based on conditions the agency can see and the client agrees with.

And because the number is tracked, the effect is measurable. Add an out-of-hours destination in March and the missed-call count for evenings should collapse in April. If it does not, the rule is wrong or the destination is not answering, and either way you will know within a month rather than never.

How a rule is evaluated

Deliberately simple. A routing system nobody understands is a routing system that eventually sends a caller somewhere embarrassing.

1. 1

   The call arrives on a tracked number

   Which already knows the client it belongs to, and — because dynamic number insertion assigned it to a session — which traffic source the caller came through.

2. 2

   The current time is read on the client's clock

   Not the server's. A rule that says "weekdays, nine to five" means nine to five where the business is, which is the only interpretation the client will ever expect and a surprisingly common thing for tracking platforms to get wrong.

3. 3

   Rules are checked in priority order

   Each rule states its days, its hours and optionally its source. The first one that matches wins. Ordering is explicit rather than inferred, so a narrow rule can sit above a broad one and behave predictably.

4. 4

   The matching rule supplies the destination

   A different number: the sales line, a mobile, a second branch, an answering service, an overflow queue. Whatever the client actually wants that call to reach at that moment.

5. 5

   If nothing matches, the number's default applies

   There is always a fallback, and it is the destination the number would have used with no rules at all. A misconfigured rule set degrades to the current behaviour rather than to silence.

Rules worth setting on almost every account

These five cover most of what local businesses actually need. Start here and add complexity only when a measured problem demands it.

The timezone problem, and why it is not a detail

Routing rules are made of hours, and an hour only means something relative to a clock. Which clock a tracking platform uses for "nine to five" is one of those questions that sounds pedantic until it is wrong, at which point it is the only thing anybody wants to talk about.

The failure is quiet and specific. An agency in London sets business hours for a client in Denver. The platform evaluates the rule on server time. The client's nine-to-five rule now takes effect from two in the morning to ten in the morning, local. Calls through the middle of the client's working day route to the out-of-hours answering service, and calls at two a.m. ring an empty office. Nobody notices for weeks, because the answering service does take messages and the messages do arrive, just from an expensive service that was meant to handle a handful of evening calls.

The same class of bug affects daylight saving. A rule written in winter that is evaluated against a fixed offset drifts by an hour in spring, so the last hour of the working day starts routing to the night destination. Storing a proper timezone identifier rather than an offset is what makes that not happen, and it is why the client's timezone is a first-class setting on the account rather than a display preference.

This matters more than it used to because agencies increasingly run clients in several regions at once. One rule engine, evaluating every client on that client's own clock, is the only arrangement that scales past the point where everybody happens to be in the same country as the server.

Routing by where the caller came from

Time-based routing is common. Source-based routing is rarer, more interesting, and only possible when the number already knows which campaign the visitor arrived through — which is exactly what dynamic number insertion provides. The number was assigned to the session before the visitor ever touched their phone, so by the time the call arrives its origin is a known fact rather than something to be inferred.

It is worth being precise about what is being matched, because a lot of platforms get this subtly wrong. The condition is the traffic source of the visit that produced the call — the campaign, the medium, the referrer. It is not the caller's country, not their area code, and not the number they happened to dial from. Matching on caller geography is a different feature entirely, and one that answers a question almost no local services agency is actually asking.

The reason source is the interesting axis is that it is the closest thing available to intent before anybody speaks. Somebody who clicked a paid search ad for emergency drain clearance and rang within ninety seconds is in a different state of mind from somebody who found the site through a blog post about seasonal maintenance. You cannot know what either wants, but you know where they came from, and where they came from predicts enough to be worth routing on.

Combining the two axes is where it becomes genuinely useful. "Paid traffic, weekdays, nine to five, to the sales line" and "paid traffic, any other time, to the answering service that has the out-of-hours script" are two rules that between them make sure the calls you paid for are never the ones that go unanswered — which, on most accounts, is where the money is.

What source routing lets you do

Each of these is a real pattern agencies use, and none of them is possible with a static forwarding number.

The things that stop routing going wrong

Routing is one of the few settings in a marketing platform that can lose a client real money in real time. It is built accordingly.

• Preview before you publish. Ask the system where a call would go on a Saturday at 10:40 from paid social, and it tells you which rule matches and which destination wins — without a real caller finding out for you.
• There is always a fallback. A call that matches no rule uses the number's default destination. A rule set that is wrong degrades to today's behaviour, never to a dead line.
• Priority is explicit. You set the order rules are checked in. Nothing is inferred from specificity, so a rule does not silently change behaviour because another rule was edited.
• Overnight windows work. A rule from 18:00 to 08:00 spans midnight correctly rather than matching nothing, which is the single most common bug in hand-rolled hour matching.
• Rules are scoped to the client. A rule belongs to one tracked number belonging to one client, enforced on the server. No cross-client rule can exist even by accident.
• Changes are visible. Who changed a rule and when is recorded, because "the calls started going to the wrong place on Tuesday" is a question that needs an answer.

Setting rules a client will actually agree to

The technical part of routing takes about ten minutes. The part that takes a fortnight is getting the client to agree where calls should go, because routing touches their staff, their money and occasionally their pride. A few things make that conversation shorter.

Start with evidence, not proposals. Walking into a meeting with "we think you should route evening calls to an answering service" invites a debate about whether evening calls exist. Walking in with "you received 41 calls after six last month and answered four of them" invites a debate about what to do, which is the debate you wanted. The missed-call data almost always precedes the routing change, and should.

Then propose the smallest rule that addresses the largest gap. One rule, one window, one destination. An elaborate seven-rule scheme presented in month one will be rejected on general principle; a single out-of-hours rule that visibly recovers a dozen enquiries buys you the credibility to propose the rest. Agencies routinely over-engineer this and then wonder why the client went quiet.

Be honest about the cost side. An answering service is a real monthly expense and a duty mobile is a real imposition on somebody's evening. If the recovered enquiries do not obviously outweigh that, say so — the calculation from the missed-call arithmetic usually makes it clear either way, and a client who has seen you argue against a change that would not pay for itself will believe you the next time you argue for one.

Finally, agree how it will be reviewed. A routing rule set in March and never looked at again is a liability, because businesses change their hours, staff leave, and answering services get cancelled without anybody telling the agency. Put the answer rate by hour on the monthly report and the rules will get revisited when they need to be rather than when something breaks visibly.

What routing cannot do

It cannot make somebody answer. A rule directs a call to a destination; whether a human picks up at that destination is outside the software entirely. This is why routing and call intelligence belong together — the routing decides where calls go, and the missed-call reporting tells you whether the decision worked.

It is not a full PBX. There is no auto-attendant tree, no hold music, no queue with position announcements, no extension dialling. If the client needs a phone system they should buy a phone system; this decides where a tracked marketing number points and stops there, deliberately, because the alternative is a half-built telephony product competing with real ones.

It does not route on things it cannot know. Caller intent, which department the person wants, whether they are an existing customer — none of that is available before the call connects. Source is the closest available proxy for intent, which is why source routing is the interesting axis rather than an attempt at something cleverer.

And it will not fix a business that has nobody to answer the phone. Routing to an answering service that takes a message the client never actions is worse than a voicemail, because it costs money and creates the impression of coverage. Where the honest answer is that the client needs another person, that is the recommendation, and the missed-call data is what makes the case.

How it fits with everything else

Routing sits on top of call tracking, because a rule needs a tracked number to attach to and a source to match on. It sits beside call intelligence, which measures whether the routing is achieving anything: answer rate by hour, missed calls by day, abandoned calls by source.

That pairing is the whole workflow. Measure where calls are being lost, write a rule to catch them, and then measure again to confirm it worked. Without the measurement, routing is a set of assumptions about when a business is busy. With it, each rule is a response to an observed gap and has a number attached to whether it closed.

Nothing about routing changes attribution. A call routed to the sales line is attributed to the source that produced it exactly as it would have been otherwise, so you can route freely without worrying that the reporting will move underneath you.

Downstream, a routed call behaves like any other tracked call: it lands on a lead record with its source attached, it can be recorded and transcribed, and it is pushed into the client's CRM with the destination it reached noted. That last detail is quietly useful when a client asks why a particular enquiry went to the answering service — the record says which rule matched and where the call was sent, which turns a suspicion into a checkable fact.

Common questions