The Effort Score and What It Predicts That Satisfaction Does Not

Why this matters

Of the three feedback numbers a shop usually ends up with, satisfaction, loyalty and effort, the effort score is the one that reliably tells you whether the customer will call you next time, and this card tests it head to head against satisfaction. Satisfaction rates how the story ended. Effort rates what it cost the customer to get there. Those come apart constantly, and they come apart hardest on exactly the jobs a shop is proudest of: the one that went wrong, got fixed generously, and earned a top rating from a customer who had already decided never to do that again. This card makes that case with a shop's own numbers rather than asserting it, then names the specific things high effort points at.

What each question is actually rating

Satisfaction asks for a verdict after the fact. The composition is the mean rating across responses received in the window, and the rating is dominated by the outcome: did I end up with a working system and did somebody treat me decently about it. A generous resolution moves it a long way, because the outcome is what the customer is grading.

Effort asks a process question: how easy it was for the customer to get their problem handled, which is the inverse of how much work they personally had to do, aggregated across respondents the same way. A generous resolution does not move it, because the chasing already happened. You cannot refund somebody's four phone calls.

That is the whole difference and it is why they disagree. One is a judgement about you. The other is a measurement of your process, taken by the only person in a position to take it.

Both are computed over the people who answered, which is a constraint neither of them escapes. See related: The Satisfaction Score and Who Actually Answers a Survey.

The direction of the scale is part of the number

Write the direction on the form and never change it. A 1 to 5 scale asking how easy it was runs with 5 as the good end. A 1 to 5 scale asking how much effort the customer had to put in runs with 5 as the bad end. Both wordings are in common use and they are exact opposites, so a shop that reworded its question between quarters has inverted its own series without a single number looking wrong. This card and the effort-score card in the references both put 5 at very easy on a 1 to 5 form; if your own form runs the other way, or runs 1 to 7 as that card's default does, restate your bands on this 1 to 5 easy-is-high scale before you compare anything here to your own figures.

The tell is cheap. If the effort mean jumps by more than your noise floor in the same period anybody touched the survey wording, suspect the wording before you suspect the work, and go read the two versions side by side. Everything below uses 1 to 5 where 5 is very easy, matching the effort-score card's short form, so a HIGHER number is better.

One job where the two answers point opposite ways

An equipment replacement, booked for a Monday.

• Day 1. The tech arrives and the replacement unit on the truck is the wrong one. The job is rescheduled.
• Day 4. A second tech puts the unit in. That evening the customer finds a leak at a connection.
• Days 5 and 6. The customer calls twice. Neither call is returned. They email. On day 6 they call again and reach somebody.
• Day 9. A return visit fixes the leak properly.
• Day 10. The owner calls personally and takes a third off the bill.

The customer rated satisfaction 5 of 5 and wrote "they stood behind it." They rated effort 2 of 5, which on this form is hard.

Both ratings are honest and both are correct. The shop did stand behind it, and the customer did have to work for that. Look at what the job record carries alongside those two numbers: four contacts the customer had to start, three scheduled dates on one job, and four different people from the shop who spoke to them. The satisfaction rating sees none of that, because none of it changed how the story ended.

Eleven months later the annual service came due and the customer did not book it. That is one job and one customer, so it predicts nothing on its own; it is here because it shows the two questions pointing opposite ways on the same work. Sizing the pattern needs the aggregate, which is next.

What a trailing year says

The same shop, twelve months, 132 respondents who answered both questions, with "bought again" defined as any paid job for that customer within 12 months of the surveyed one. Across all 132, 69 bought again, a repeat rate of 52.3 percent of respondents.

Split those same 132 respondents by effort:

Now split the same 132 respondents, on the same repeat definition, by satisfaction:

Three of those bands are thin. A survey figure needs about 30 respondents behind it before it is worth quoting, and the 26-respondent effort band and both 22-respondent bands are under that floor, so their individual rates are a direction and not a size. Do not quote 27.3 percent at anybody yet; roll a second year first.

What you can quote is the same comparison at a two-band cut, where every band clears the floor:

• Effort easy (4-5) against effort not-easy (1-3). 51 of 84, 60.7 percent of that band, against 18 of 48, 37.5 percent of that band. A separation of 23.2 points.
• Satisfaction top-box (5) against everything below (1-4). 42 of 79, 53.2 percent of that band, against 27 of 53, 50.9 percent of that band. A separation of 2.3 points.

Same respondents, same repeat definition, same two-band resolution: effort separated the returners from the non-returners about ten times as far as satisfaction did. And inside the satisfaction split, the two largest bands both clear the floor and land on top of each other, the 4s at 54.8 percent of their band against the 5s at 53.2 percent. That is a 1.6 point gap on a 31-respondent band where one customer is worth 3.2 points, so read it as no separation rather than as a reversal. A top-box satisfaction rating bought this shop nothing in return visits it would not have had anyway.

One shop's year is not an industry finding, and you should run this split on your own records before believing it about yourself. It is an easy split to run: you already have the ratings and you already have the repeat purchases.

Why the process reading is the one that predicts

The mechanism is not mysterious once the two compositions are in front of you.

Satisfaction is a memory of an outcome, and memories of outcomes decay and get revised. It is also the rating a customer gives partly out of politeness toward a specific tech they liked. Neither of those survives to the moment eight months later when something breaks at half past seven in the evening and the customer has to decide who to call.

Effort is a memory of friction, and friction is what the customer is actually forecasting when they make that decision. The question in their head is not "were they good people" but "how much of my evening is this going to cost me." A customer who chased four times has a direct answer to that, and the generous discount did not change it.

There is a second, quieter reason the effort number works better as a management signal: it points at something you can act on. A low satisfaction rating tells you a customer was unhappy and leaves you to guess why. An effort rating down at the hard end on a job with three scheduled dates on it tells you the scheduling is the problem, and the fix is a process change rather than an apology.

What high effort points at, and the tell for each

Effort is a symptom. Six causes produce nearly all of it in a field-service shop, and each leaves a different fingerprint in your own records. The point of the last column is that no two tells point at the same cause, though the first and the last are separated only by timing: a paperwork chase is a chase, and what distinguishes it is that the contacts start after the work finished rather than during it.

The fixes are ordinary and they are worth stating as numbers rather than as intentions:

• Chasing. One named person owns each open job, and every customer message gets a human reply the same working day, next working day at the outside. See related: Service Request Conversion and How Long You Take to Answer.
• Repeating themselves. Whoever picks the job up reads the notes first. Where the shop is small enough, one person owns the job start to finish, which removes the cause instead of managing it.
• Channel switching. Answer on the channel they used. A customer who booked by text and gets an email has been asked to do your filing.
• Reschedules. If an arrival window is going to slip by more than 15 minutes, the customer hears it from you before the window closes, not after. Tune the 15 if your windows are wider, but commit to a number, because "we call if we are running late" is not a rule anybody can follow.
• Second visits for one fault. This one is not a communication problem and no amount of calling fixes it. Effort is just where it surfaces first, usually a quarter before the callback rate moves.
• Paperwork. Invoice within one working day of the work being complete, and check it before it goes.

Watch the effort mean and the contact-count field together. If the mean holds while the average number of customer-started contacts per job climbs, the friction is growing and the survey has not caught up yet, because the customers feeling it are the ones least likely to answer.

References

• See related: The Satisfaction Score and Who Actually Answers a Survey
• See related: The Loyalty Score and Why a Middling Answer Counts Against You
• See related: Service Request Conversion and How Long You Take to Answer
• See related: The Satisfaction Score Held While the Repeat Rate Fell
• See related: Customer Effort Score (CES) + Friction Reduction