Choosing a law firm answering service is a well documented problem. Firms compare providers, read the pitch, ask about coverage, and sign. What almost nobody does is go back a year later and check whether the thing they bought is doing what they bought it for.

That gap exists for a specific reason. An answering service reports on itself. Every month you receive a tidy account of calls handled and messages delivered, and every month it looks fine, because a report written by the party being measured will describe the calls it handled and not the ones it lost. If you have never audited yours, you do not have evidence that it works. You have evidence that it files reports.

This is the audit. It takes an afternoon, it needs no cooperation from the provider, and you should run it before you have any conversation about renewing or replacing. If you are still choosing a provider, the companion piece on what a legal answering service should do on a first call is the better place to start.

Start with the calls you never heard about

The calls that cost a firm money are invisible by construction. A message that arrives badly written is annoying but visible, so it gets fixed. A call that never became a message produces no artifact at all. Nobody complains about it, nobody reviews it, and it never appears in a monthly report, because the report is a list of things that happened.

So the audit does not start with the provider's numbers. It starts with a source the provider does not control: your own carrier's inbound call detail. Pull a full month. You want the timestamp of every inbound call, how long it lasted, whether it was answered or abandoned, and which number it rang on.

Then line that up against everything the service actually handed you in the same period: messages, files created in your case management system, recordings, follow up tasks. You are looking for rows in the carrier log with nothing on the other side. Every one of those is a person who called your firm and left no trace inside it.

Four failures, and why all of them look like good service

In practice the same four problems account for most of what an audit turns up. What they have in common is that each one produces something that reads like competence.

The message that reads well and decides nothing

This is the most common finding and the easiest to miss. The message is punctual, correctly spelled, polite, and complete as a message. It contains a name, a number, and a sentence about a car accident. It does not contain the date of loss, whether treatment was sought, whether the caller has already retained counsel, or who else was involved.

Read fifty of those and they look like a service performing well. Try to make a decision from any one of them and you cannot, so somebody at the firm calls the person back to ask the questions that should have been asked while they were already on the phone. Sometimes that second call connects. The pattern to look for is not a bad message. It is a run of good messages that all required a callback to become useful.

The call that arrived when everyone was busy

Every provider has a plan for simultaneous calls and very few volunteer what it is. The call rolls to voicemail, or to an overflow vendor, or into a queue the caller abandons. Look for clusters in the carrier log: several inbound calls inside the same few minutes, with fewer messages than calls. Then check whether those clusters fall at predictable times, because a Monday morning or a late Friday pattern is a capacity problem rather than bad luck.

The handoff that lands where nobody is looking

A service can qualify a caller properly and still lose the matter at the last step. The file arrives as an email to an address that nobody reads on a Saturday, or as a note in a system field that no view surfaces, or as a task assigned to whoever set the integration up two years ago. The information exists. It just sits somewhere until Monday. Our piece on the handoff from intake to attorney goes through where these breaks usually happen.

Test it the blunt way. Take three files the service delivered on a weekend and trace exactly who saw them, when, and what they did next. If the answer is that somebody found them on Monday morning, then your overnight coverage is producing a record rather than a response, and you are paying for speed you are not receiving.

The caller who was screened out by the wrong person

This is the failure worth the most and the one an audit almost never surfaces, because a caller who was told the firm could not help does not call again and does not appear anywhere as a problem. Somebody at the service decided the matter was not viable and closed it.

Ask for recordings of calls the service classified as not qualified. Nobody at your firm has listened to those. That is precisely why they are the sample worth hearing.

Listen to the calls, not to the summary of the calls

Ask for the raw recordings and listen to them yourself, or have the person who actually signs cases listen. A written summary is the provider's account of its own work, and it will be coherent whether the call went well or badly, because the same party wrote both.

Listen for a small number of things. Did the person answering know what kind of matter this was before the caller finished explaining. Did they ask when it happened, and did anything happen with that answer. Did they ask whether another firm was already involved. Did they tell the caller what happens next and when, or did they end on a variation of somebody will be in touch. And listen to the pace, because a caller who is being processed can hear it.

The test that actually tells you something: call your own firm

Everything above is documentary. This part is empirical and it is the single most useful hour in the audit.

Call your own firm from a number nobody at the service will recognise. Do not sound like a lawyer. Describe a plausible matter the way an ordinary person would, in the wrong order, with the important fact buried in the middle. Then stop talking and let the person on the other end do their job.

What you are measuring is not politeness. It is whether the questions came, whether anyone established when the incident happened, and what arrived at your firm afterwards, in what format, how fast. Then wait and see whether anyone follows up, and when.

Run it more than once, and deliberately at the hours you are worried about. Late evening, the middle of the night, and Sunday afternoon are frequently three different teams with three different levels of training, and the daytime team is usually the one you met during the sales process. Our note on after hours lead response covers why those windows behave differently.

What overnight actually means, and who is actually there

Round the clock is a claim about hours, not about who is working them. During the audit, establish plainly whether the people answering at 2am are employed by the company you contracted with, whether they are trained on your practice area or on a general script, and whether they can create a file in your system or only leave a message for somebody who can.

The reason to settle this on evidence rather than on assurance is that the difference is invisible from the inside. Both arrangements produce a message in your inbox in the morning. Only one of them produces a caller who felt attended to at the moment they called.

When the language of the call is the whole problem

If a meaningful share of your callers are more comfortable in a language other than English, check whether the coverage you were sold survives at night and at weekends, or whether it quietly becomes a callback promise. A caller describing an incident in a second language while upset leaves things out, and nothing downstream can recover a fact that was never spoken aloud. We wrote about this in more detail under bilingual legal intake.

Read your own contract against your own call log

Now put the two documents side by side, because a pricing model is an instruction about behaviour.

Per minute billing pays the provider for time on the phone. Per call billing pays them to finish the call. A per qualified lead arrangement pays them to judge viability, which means the definition of qualified is the most important sentence in the agreement, and it belongs there rather than in something a salesperson said. A flat retainer is neutral on call length and says nothing at all about quality.

None of these models is dishonest. The point of reading the contract next to the log is to find out whether the incentive you signed up for matches the behaviour you are seeing, because when it does match, the behaviour is not a service failure. It is the arrangement working as written, and no amount of quality conversation will change it.

What to ask the provider once you have evidence

Only now is the conversation worth having, because you are bringing findings rather than impressions. Reasonable questions at that point:

Here are timestamps from my carrier log with no corresponding message. What happened on those calls. What happens to a call that arrives when every one of your agents is busy. Who answers between midnight and six, and are they your employees. Can I hear ten recordings from my practice area, chosen by me, including calls you closed as not qualified. What exactly lands in my case management system, in which fields, and who is accountable when a caller is mishandled.

A provider who can answer those without a follow up meeting is probably running a serious operation. A provider who cannot tell you who answers overnight has told you something already.

When the answer is to change, and when it is not

Not everything an audit turns up is a reason to leave. A single mishandled call is a coaching matter, and a badly built integration is worth fixing rather than replacing, because a new provider will need the same integration built again.

What does not respond to coaching is structural. Calls that disappear between your carrier log and your files. Overnight coverage that turns out to belong to a different company than the one you pay. A pricing model that rewards the opposite of what your intake needs. Those are properties of the arrangement, and the only lever is the arrangement itself.

The reason to run this before renewal rather than after a bad month is that a bad month makes the decision for you and usually makes it badly. An audit run while nothing is on fire produces something better than a decision: it produces a record of what your firm actually receives when somebody calls it, which is the one thing almost no firm can currently describe.

If you want the same material from the buying side rather than the auditing side, start with what law firms should expect from a legal answering service, or with our guide to reducing missed calls.

Common questions