Quality Control

The role

The formal title is Sensory Panel Lead. The factory calls her the Nose, which she has never liked and has never corrected, and which is anatomically wrong.

Ingrid Solheim tastes between three hundred and four hundred and fifty product samples a day across eleven production lines and reports deviations against a reference profile.

She has done this for nineteen years.

What she is actually detecting

Not "good" and "bad." That is the thing people cannot get past.

The reference profile is a specification. Her job is to say whether a sample matches it, and if not, in what direction and by how much, on a scale that has been calibrated against her four times a year since 2006.

Her personal opinion of the biscuit is irrelevant and she has never once been asked for it.

She does not like the biscuit. She has not liked it since about 2011. Nobody at the factory knows this and it has never affected a single result, because it is not what she is measuring, and being able to hold those two things apart is the entire skill.

The record

Nineteen years, four calibration panels a year, seventy-six panels.

She has never failed one.

She has never been overruled by instrumentation. This has come up eleven times — a sample she flagged that the gas chromatograph passed — and on all eleven occasions a subsequent investigation found a genuine deviation that the instrument had not been configured to look for.

The most celebrated was 2014, when she rejected an entire day's output of the caramel line on the grounds of "something like wet cardboard, very faint." Two hundred thousand units. The plant manager, who was new, overruled her.

The complaint volume in the following fortnight was the highest in the plant's history.

The cause was a batch of packaging film from a new supplier with a residual solvent at four parts per billion.

She had tasted four parts per billion, through a biscuit, and been told she was wrong by a man with a spreadsheet, and had put it in the log anyway, in the box marked NOTES, which is the only reason anybody found it.

The plant manager left in 2015. Nobody at the factory has ever discussed this with Ingrid directly. It is simply that since 2015 nobody has overruled her.

The problem

Ingrid is fifty-nine.

Sensory acuity declines. This is not in question and she has never pretended otherwise; it is in the literature and it is in her own calibration data, which she can read as well as anybody.

Her scores are still inside tolerance. They have been drifting for four years in the way that everyone's do, and they will cross the line at some point in the next five or six years, and on that day she will be, professionally, finished.

She has told the company this. She told them in 2021, unprompted, in writing.

The company's response was to promote her.

The panel

She now runs a panel of nine, which she recruited and has trained, and which is the actual point of the last four years.

Selection is brutal. She screened four hundred and eleven internal applicants to find nine, and the tests are not about having a good palate — most people have an adequate palate — they are about consistency, and about the willingness to write down a thing that nobody else has noticed and that will make you look stupid.

That last one is the whole job and it cannot be trained. She can identify it in about ninety minutes and she has never been able to explain how.

The nine

They are calibrated against her, which she has raised as a methodological problem four times, because it means her drift becomes their reference.

The company has not acted on it. She has raised it again in writing this year, more sharply.

Her best is a woman of twenty-six called Danijela who came off the packing line and who has, in Ingrid's assessment, better acuity than Ingrid had at that age.

Danijela disagreed with her in September.

A sample off line 4. Ingrid passed it. Danijela flagged it — faint, unspecified, "something not right at the end."

The instrumentation passed it. The line ran.

Ingrid spent that evening in the sensory booth with eleven retained samples and could not find it.

She could not find it on the Tuesday either.

On the Wednesday she signed the report as PASS, with a note recording the dissent, because that is the procedure and because she could not detect what Danijela had detected.

Then she went to the plant manager and asked for a review of the line 4 packaging film.

He said the instrumentation had passed it.

Ingrid said: "I know."

He looked at her.

"Do it anyway," she said. "I'll put it in writing."