The job was booked by phone, done in a van, and invoiced two weeks later

A plumbing group running fourteen vans has a marketing dashboard showing website conversions, form fills and cost per lead. Meanwhile, four out of five of its jobs start with somebody ringing a number they found on a search result, and the value of each job is not known until an engineer has been under a floor for an hour. The dashboard is measuring a real thing. It is not measuring the business.

In field services the marketing report and the money are in two different systems, and almost nobody has ever put them in the same room.

This article is about doing that: the call and job match, why invoiced value matters, and the three findings that come up in nearly every multi-crew business that runs the analysis for the first time.

The two systems that matter

Field service measurement is a join between the call tracking platform, which knows where an enquiry came from, and the job management system, which knows what was completed and invoiced. Both exist in most businesses of this size and they have never been connected.

The job management system is the revenue source of truth: it holds the completed job, the customer, the address, the invoice value and the date. That is the equivalent of a CRM's closed-won record and it is usually better maintained, because invoicing depends on it.

The call platform holds the source. Where dynamic number insertion is in use, it can attribute a call to a session and therefore to a campaign; where it is not, it will at least distinguish the numbers used on different marketing assets. Either is enough to start.

The join is the phone number, which is why the normalisation article matters more in this sector than in any other. A job management system storing numbers as typed by a receptionist and a call platform storing full international caller ID will match nothing at all until both are normalised.

Invoiced value, not quoted value

Attribute on the invoiced amount rather than the quote or the booking, because field service jobs routinely change value once work begins. Quoting-based attribution understates exactly the channels that generate substantial work.

A boiler replacement quoted at two thousand two hundred that becomes three thousand eight hundred once the pipework is exposed is an ordinary event in this industry. So is a job that shrinks. Attribution built on the initial figure carries an error that is not random: larger, more complex jobs move more, and those are the ones that matter most.

This has a practical consequence for the export. The invoiced value often lives in the accounting system rather than the job system, and connecting them may require a third file. It is worth the effort, and where it genuinely cannot be done, using completed job value rather than booked value captures most of the difference.

The first finding: missed calls

The most common discovery in this sector is a large number of enquiries that were never answered or never returned, and their revenue value is usually greater than any plausible improvement in media efficiency.

The pattern is consistent: calls outside office hours, calls during the morning dispatch rush, calls arriving while the one person on the phones is already on a call. Each one is a customer with an urgent problem who will ring the next result within about ninety seconds.

Quantifying it is straightforward once call records are matched to jobs. Take unanswered calls, look at how many of those numbers ever became a job, and value the gap at your average job value. The number produced by that calculation is routinely large enough to fund a full-time person, which is usually the correct remedy.

It also distorts channel reporting until it is fixed. Channels generating out-of-hours enquiries — which in this sector are frequently the highest-intent emergency channels — will appear to convert poorly, and the natural response of reducing spend on them makes the business smaller.

The second finding: emergency and planned are different businesses

Emergency work and planned work have different customers, different decision processes and different winning channels. Reporting them together produces an average that describes neither and hides the more profitable one.

A single blended report will conclude that search is the dominant channel, because emergency volume is high and search wins it. That conclusion then drives budget away from the reputation and referral activity that produces the larger planned jobs, which is precisely backwards for a business trying to grow value rather than volume.

Splitting them requires a job type field on the export, which the job management system almost always holds because dispatch depends on it. It is one column and it changes the conclusion.

The third finding: repeat revenue is where the value is

A customer acquired once may generate jobs for a decade. First-job attribution values that customer at their first invoice, which understates the channels that acquire loyal customers and overstates those producing one-off work.

This is the most consequential adjustment available in field services and the least commonly made. A four hundred pound call-out from a channel whose customers return annually for servicing is worth several times a four hundred pound call-out from a channel whose customers never return, and the standard report values them identically.

The data supports the better view without any new collection. Because the job system holds the customer, subsequent jobs can be tied to the original acquisition source directly. Reporting cohort revenue at twelve and twenty-four months by acquisition channel is entirely feasible from history you already have.

CloseRev matches on normalised phone numbers with libphonenumber and reports on whatever dimensions your export carries, so job type, crew, branch and repeat status all come from columns your job management system already holds.

Directories, aggregators and the lead resale problem

Lead aggregators sell the same enquiry to several contractors at once, so the enquiry you paid for was simultaneously sold to three competitors. Matched revenue is the only way to see what that costs you.

The economics are unforgiving and rarely modelled. If a lead is sold to four contractors and one wins, your expected conversion is a quarter of what a direct enquiry would produce, before any consideration of quality. Aggregators price on the lead, not on the outcome, so the risk of that arithmetic sits entirely with you.

The matched view frequently shows aggregator leads converting at a fraction of the rate of direct enquiries and producing lower average invoice values, because a customer comparing four quotes is a price-comparing customer by construction. That is not an argument for never using them; it is an argument for pricing them correctly against alternatives.

It also changes what speed is worth. On a resold lead, being first to call is close to the entire game, and the value of answering within sixty seconds rather than twenty minutes is far higher than on a direct enquiry where you are the only contractor being considered.

Crew-level variation

Conversion from booked job to completed invoice, and the value of that invoice, vary substantially between crews. Attribution by crew turns a marketing report into an operational one, and the operational finding is usually larger.

The pattern mirrors what multi-site businesses see. Two engineers attending comparable jobs will produce noticeably different average invoice values and different rates of follow-on work, and the difference is generally about how the problem is explained to the customer rather than about technical skill.

This matters for marketing measurement because crew allocation is not random. If your best engineer is dispatched to the jobs from one channel and a less experienced one to another, the channel comparison is partly measuring dispatch policy. Knowing that prevents a confident and wrong conclusion about media.

Handled well, it is also one of the more useful findings to come out of this work, because the gap between the best and average crew is usually closable through training and scripting at very low cost relative to any marketing investment.

Seasonality, and how to compare like with like

Field service demand is strongly seasonal and weather-driven, so month-on-month comparison is close to meaningless. Compare against the same period last year, and hold the channel mix constant when you do.

A heating business comparing November against October is measuring the weather. One comparing this November against last November is measuring something closer to performance, provided the two Novembers were not radically different, which they sometimes are. Keeping a note of the conditions alongside the numbers costs nothing and prevents a great deal of misattribution.

The more subtle version is that seasonality changes the channel mix on its own. Emergency search volume spikes in cold snaps and planned work collapses, so a seasonal shift in the reported channel ranking may be entirely demand-side. Splitting emergency from planned, as described above, largely resolves this.

For businesses with strong seasonality, rolling twelve-month figures alongside the current period are worth the small extra effort. They smooth the weather out and make genuine trends visible, which monthly reporting in this sector generally does not.

Capacity, and why it breaks the numbers

When crews are fully booked, additional marketing cannot produce additional revenue, so measured performance falls for reasons unrelated to the marketing. Report capacity utilisation alongside conversion or the report will be misread every busy season.

The seasonal version is familiar to everyone in this industry. In the first cold week of winter a heating business is turning work away, and its marketing appears to collapse: enquiries up, conversion down, cost per job worse. Nothing about the marketing changed; the constraint moved.

Once capacity is visible in the report, the strategic question changes usefully. In a capacity-constrained period the objective is not more enquiries but better ones — higher value, tighter geography, less travel between jobs — and that is a different and entirely measurable marketing goal.

Geography, which is a marketing variable

Travel time between jobs is a direct cost, so two jobs of equal value in different places are not equally profitable. Attribution that ignores geography will favour channels that generate widely scattered work.

This is specific to field businesses and it is genuinely important at scale. A channel producing tightly clustered work in a few postcodes allows more jobs per crew per day than one producing the same revenue spread across a county, and the difference goes straight to margin.

Since the job system holds the address, adding a geographic dimension is available without new data collection. Even a simple grouping by postcode district, compared across acquisition channels, will show whether one source is systematically producing more expensive work to serve.

Where to start

Export three months of completed jobs with the customer phone number, invoice value, completion date and job type, and three months of call records with source and number. Normalise both, match, and look at the missed-call figure first.

Three months is enough in this sector because cycles are short, and starting small keeps the phone number formatting problem manageable while you diagnose it. Expect the first attempt to match poorly and expect the reason to be formatting rather than anything conceptual.

Look at missed calls before looking at channels. It is the finding most likely to be large, most likely to be fixable this month, and most likely to be distorting every other number in the report. It is also the one that will get the attention of an operations director, which matters when the next stage of this work needs somebody outside marketing to change how a rota is written.

One practical warning about the first run. Field service data has a characteristic problem that other sectors do not: the same customer appears under several phone numbers, because a landline was used for the original booking and a mobile for the follow-up, or because a letting agent called on behalf of a tenant. Expect a portion of repeat work to look like new acquisition until you match on address as well as number.

Address matching is worth adding once the phone match is working, and it is more tractable here than in most sectors because the job system holds a structured address that the engineer had to navigate to. Matching on postcode plus house number will connect most of the records that a phone number alone misses, and it is what makes the repeat-revenue view described above genuinely reliable.

In a field service business, the highest-return marketing analysis available usually ends with a recommendation to hire one more person to answer the telephone.