Lost Reasons Are Only as Good as the Field Being Filled

Why this matters

Every shop that tracks lost reasons eventually gets the same answer back: price is number one. It shapes pricing decisions, discount policy, and how an owner talks to their techs about quoting. Two things are wrong with acting on it. The field is blank on most of the losses, so the ranking is decided by a minority that selected itself. And "price" is a single label sitting over five different problems that need five different fixes. The ranking is real. What it ranks is not what the owner thinks it ranks.

What the ranking is a ranking of

• The numerator is a count of leads closed as lost in the window carrying a recorded reason, grouped by which reason was picked.
• The denominator, when a share is shown, is the leads that got a reason entered. Not the leads you lost.
• Every lost lead with a blank reason is outside the whole calculation. It is not a row, it is not in the total, it does not dilute anything.
• The time anchor is the close date, so a lead worked for six weeks and given up on in March is a March loss, whatever month it arrived in.

A reason field is opt-in by construction, and a rate computed only over the rows somebody chose to fill improves as you record less. That claim is derived in full on the SLA compliance card in the references, and none of it behaves differently here, so it is not re-argued below.

What is specific to a lost reason is which rows go missing. Somebody fills the field when a tidy answer is already at hand: the customer said a number out loud, named a competitor, or asked for a date next spring. Now picture the lead nobody called back for nine days. There is no tidy answer, there was no conversation, and the person who would enter the reason is the person who did not make the call. That lead gets closed silently, and it is exactly the loss the shop caused. So the blanks do not lean neutral. They lean toward the losses that are the shop's own doing, which is the half of the list an owner most needs and the half a thin field reliably hides.

Coverage changes the order, not just the level

This is where a ranked field parts company with every other opt-in figure. A compliance rate computed on a third of the work reports a wrong number, and better coverage moves the number. A ranking computed on a third of the losses reports a wrong winner, and better coverage moves the rows past each other. You do not get last quarter's answer corrected. You get a different answer, and the old one was never a rough version of it.

Same shop, two consecutive quarters, with the closed list set out further down and a required field introduced between them.

Quarter one, on the 52 filled rows:

Price is 46.2 percent of the filled rows and 17.1 percent of the quarter's 140 losses. The blank bucket, 88 leads, is 3.7 times the size of the price bucket and the largest group in the quarter by a wide margin.

Quarter two, on the 120 filled rows:

The row that leads the second table is the row that came last in the first. Could not reach them, 4 leads and the bottom of five, becomes never reached them, 34 leads and the top of seven, in a quarter where total losses fell from 140 to 132. Price led the first table by 25 points over second place and does not lead the second at all. No amount of care reading the first table would have produced that, because the rows that moved were not in the first table's denominator to be read.

The row in fifth belongs with the one at the top. Never reached them at 34 and too slow to quote at 13 both record something the shop failed to do rather than something the customer decided, and together they are 47 of the 120 filled, 39.2 percent. In quarter one both causes sat inside the 88 blanks while the owner looked at pricing.

Do not read the two price figures as a trend. Price rows are 36 of 120, 30.0 percent, against 24 of 52, 46.2 percent. That is not a fall in price objections. The earlier figure was computed over a field filled on a third of the losses and the later one over a field filled on nine tenths, so the only thing that provably changed between them is coverage, and quarter one's 24 is a floor on an unknown rather than a measurement. Treat quarter two as the first real baseline and compare forward from it.

"Price" is a bucket, not a reason

Even at perfect coverage, one label collapses at least five different problems. All five produce a customer saying some version of "it was too expensive", and each needs a different fix.

• Price, scope agreed. They understood exactly what they were buying and would not pay it. The real objection. Fix is pricing, packaging, or accepting you are not that customer's shop.
• Price, scope questioned. They compared your number against a different scope, usually a thinner one. Fix is the estimate document, not the number on it.
• Trust. The number was fine and they did not believe it would hold, or did not believe you would show up. Fix is proof: references, a written warranty position, what happens if the job goes long.
• Timing of the number. The number arrived after they had already decided. The price was never evaluated. Fix is turnaround, and it is usually the cheapest of the five.
• Funding. They wanted the work and could not fund it now. Fix is staging the job or offering a payment route.

Four questions separate them, and they take under a minute on a call. "Did the other number cover the same work as ours?" splits scope-questioned from scope-agreed. "If our number had been lower, would you have booked that week?" splits a genuine price objection from trust. "Had you already decided by the time our number reached you?" finds the timing losses, which otherwise wear the price label permanently. "Would staging the work or spreading the payments have changed the answer?" separates funding from a scope-agreed refusal, because the customer who could not fund it says yes to that immediately and the customer who would not pay it says no.

The list that makes the field worth filling

Free text is why price swallows everything. It is the shortest true-enough thing to type at the end of a call somebody did not want to be having, and it is never wrong enough to be challenged. Replace it with a closed list picked in one go, where every option changes a different decision and no option is a place to hide. Nine options, and the last of them is an honest one:

Three rules make a list like that work, and they are the whole of the design.

Every option names something you would do differently. An option that changes no decision absorbs losses quietly, which is what free text was already doing.

The honest-unknown row exists so the others stay clean. With nowhere to put "we do not know", the unknowns get distributed across whichever rows look plausible, and price is always the most plausible. That row running high is a coverage finding, not a sales finding.

Keep the list at about this length. A longer list gets picked less accurately, not more, because the person closing the lead stops reading it and lands on whichever option they used last time. If you want a tenth, retire one first.

Two of the nine took no picks at all in the quarter two table: could not fund it now, and did not believe we would deliver. A zero is a reading. An option sitting at zero for a full year is either genuinely absent from your market, in which case retire it, or it is the one nobody wants to be the person who typed. Both are worth knowing.

Getting it filled without a policing campaign

Three things, and only the first of them is about coverage. The other two protect accuracy, which is the half that gets lost once coverage arrives.

Make the reason part of closing, not a separate task. Whoever closes the lead picks the reason at that moment, and a lead cannot reach closed without one. The sibling card on overdue follow-ups sets the other half of the same rule: an open lead either carries a next step or gets closed. Closing then costs one pick, which is the price of getting the count out of a stalled-lead pile.

Never attach the field to somebody's review. This is the one that quietly destroys accuracy while coverage looks excellent. The moment a reason code can be read as a scorecard, price becomes the safe answer, because price is nobody's fault. You will get 95 percent coverage and a table that means less than the 37 percent one did. Say out loud that it is a diagnostic, and never cite it in a performance conversation.

Audit by calling, not by reading. Coverage you can see. Accuracy you cannot, and the only way to check it is to speak to people the field has already described.

The five calls a ranked count can never replace

A count tells you which option got picked. It cannot tell you whether the option was right, and it never explains a cause, because nobody writes a sentence into a field with nine choices in it.

Call five lost leads a month, two to four weeks after the close, and use somebody who was not on the job. Two questions: "When you decided to go another way, what was the thing that decided it?" and "Was there a point where we could have changed your mind?" The second one is where the useful material is, because it dates the decision and often moves a price loss into the timing row.

What you are listening for is agreement between the file and the person. Over a quarter that is 15 calls. In quarter two above, 11 of the 15 matched the recorded reason, which clears a two-thirds gate at 10 of 15, so the ranking is usable. Below that gate, stop reading the table: your problem is accuracy, not coverage, and more coverage on an inaccurate field just produces a more confident wrong answer.

References

• See related: SLA Compliance Only Counts Jobs That Carry a Deadline, which owns the opt-in denominator claim in full
• See related: Lead Source Performance: Rank by Conversion, Not Volume, for reading the rate these losses sit behind
• See related: Overdue Follow-Ups Cannot Count a Lead With No Date Set, for the open-lead half of the closing rule
• See related: The Quote Follow-Up Checklist, for the process that produces fewer never-reached losses