The Verdict Nobody Had Voted On
I spent three weeks with a speaker I was coaching toward a TED stage, and almost none of the work had anything to do with the speech.
She was tense, sharp, and angry in a way that had no target in the room. Her position was settled, and she put it in one flat sentence: "I can't do this." Read that sentence again. It sounds like a description of a task. By the time she said it out loud, it had hardened into a verdict about a person.
I did not argue with the verdict. I asked what would change if she set the outcome down for an hour. Something in her let go. Her shoulders dropped, her voice slowed, and inside that quieter place she found a choice she had not been able to see while she was braced: keep carrying the weight, or put it down and work. What came back underneath was the thing I had always known about her, a standard she had set for herself years earlier about responsibility and about the people who would be watching.
Here is the part that matters for a business. The evidence she had been using to decide whether she was capable was arriving from outside her: the size of the stage, the imagined reaction of an audience that was not in the room, the judgement of strangers who had never met her. The verdict that carried her back into her own skin came from the standard she already held.
Every founder I work with runs two versions of the same question, and they are built from different machinery.
How do I know whether this is working?
How do I know whether I am any good?
The first is a measurement question, and its answer lives in numbers, customers and cash. The second is a reference question, and its answer lives in whoever you have appointed to issue the verdict. Founders mix the two constantly, which is how a flat month turns into a crisis of identity while a genuine warning sign gets waved away as noise.
The Filter the Field Named Reference
The second question has a name in the NLP tradition. It is a filter called reference, and it sorts every piece of evidence you meet according to one condition: where does the standard of proof sit?
Run an internal reference and the standard sits with you. You know a piece of work is good before anyone confirms it. You can walk out of a meeting, decide the proposal is right, and hold that position through a week of polite disagreement.
Run an external reference and the standard sits outside you. A piece of work becomes good when the market, the board, a mentor or a competitor says so. You can hold the same opinion as the internal-reference founder and still feel unsure until somebody with authority confirms it.
Richard Bandler and John Grinder modelled this distinction from the language and behaviour of the people they studied through the mid 1970s, and Leslie Cameron-Bandler developed it into a working set of filters in They Lived Happily Ever After in 1978.
Bandler has said in interviews since that the substance of the field is an attitude of curiosity about subjective experience rather than a bag of scripts. That attitude is exactly why the filter earns its place ahead of any technique you might learn. A script tells you what to do. The filter tells you who gets to decide whether it worked.
Reference itself is a sorting device built from observation. It is not a scale that was standardised on a sample and normed, and I will not present it as one.
You can hear the filter in a single sentence. "I'll know it worked when the numbers come in" is external. "I'll know it worked because I could feel it land" is internal. Neither sentence tells you whether the person is right. Both tell you who has been handed the gavel.
Before you decide the internal version is the strong one, hold one thing in view. Neither setting is a virtue. Each one protects you from a specific failure, and each one walks you into a different one.
The Same Pattern Shows Up in the Research on Control
Psychology has spent half a century studying a construct that sits close to this filter, and the research is worth knowing even though the two ideas are not the same thing.
In 1966 Julian Rotter published his work on locus of control: the generalised expectancy a person carries about whether outcomes follow from their own behaviour or from forces outside them. The internal version expects outcomes to follow from what you do. The external version expects them to follow from luck, powerful others, or a system you cannot move. Rotter's monograph became one of the most cited papers in the field, and the construct has been measured in thousands of studies since.
Two meta-analyses give the idea weight in the domain you care about. Timothy Judge and Joyce Bono, writing in the Journal of Applied Psychology in 2001, pooled the evidence on four core self-evaluations and found locus of control, alongside self-esteem, generalised self-efficacy and emotional stability, reliably related to both job satisfaction and job performance. Andreas Rauch and Michael Frese, in the European Journal of Work and Organizational Psychology in 2007, analysed studies of business owners and found an internal locus of control related to both business creation and business success.
Self-determination theory adds the mechanism. Edward Deci and Richard Ryan set out in Psychological Inquiry in 2000 that autonomy, the experience of your actions being your own, behaves as a basic psychological need, and that behaviour which is self-endorsed holds up better over time than behaviour driven by external pressure.
Keep the mapping honest. Locus of control is an expectancy about cause and effect. Reference is a filter about where the standard of proof sits. The two overlap in the territory they cover and they come from different traditions. Take the research as support for the mechanism running underneath the filter, and take the filter itself from the modelling tradition that produced it.
Where Each Setting Breaks a Founder
Here is how each setting fails, in language I hear inside real businesses.
External reference at the wheel. The founder cannot move until the market confirms the move. The proposal waits for one more opinion. The pivot is announced after the competitor announces theirs. The mood of the week tracks the last comment somebody made. You will hear the tell in the diary: decisions queued behind other people's verdicts, and a business that keeps asking permission from a room it is supposed to be leading.
Solomon Asch showed how strong that pull runs. In his 1951 conformity experiments, a substantial share of people who could plainly see which comparison line matched the reference were willing to state the wrong answer because a unanimous group had stated the same answer first. Hand the verdict to a room, and the room will hand you a number that has nothing to do with what you can see with your own eyes.
Internal reference with no calibration. This is the quieter failure and the more expensive one, because it wears the costume of confidence. Arnold Cooper, Carolyn Woo and William Dunkelberg surveyed 2,994 entrepreneurs for the Journal of Business Venturing in 1988. Eighty-one per cent placed their own odds of success at seventy per cent or better, and a third put their own odds near certainty, while the objective survival rate for new ventures sat far lower. That gap is what an uncalibrated internal standard produces: a decision-maker who never receives the correction.
The correction has a known shape. Anders Ericsson, Ralf Krampe and Clemens Tesch-Römer set out the deliberate practice framework in Psychological Review in 1993, and immediate feedback from a coach or teacher sits inside the definition. Practice without feedback is repetition, and repetition will not repair a standard that has drifted. An internal reference with no outside contact becomes a closed system that answers to itself.
Put the two failure modes side by side and the design becomes obvious. The standard belongs inside. The evidence belongs outside. Let the internal filter decide what counts as a good decision, let the outside world report what actually happened, and keep those two jobs separate. When they disagree, you update on the evidence and you keep the gavel where it belongs.
Five Markers That Tell You Which Filter Is Running
Reference hides well, because it presents itself as character rather than as a setting. These five markers bring it into view.
- What you check first after a launch. Numbers first with your own read second runs on internal reference plus an evidence habit. The read first, with numbers collected afterwards as confirmation, runs on a standard that needs a witness.
- Who you tell first about a win. If a result becomes real to you only after you have reported it to the person whose judgement carries weight, the verdict is sitting outside you.
- What a bad month changes. External reference revises your view of yourself. Internal reference revises your view of the plan. Watch which one moved.
- The sentence you use under criticism. "They're right, I'm not cut out for this" is an identity verdict delivered by a stranger. "That is useful, here is what I will change" treats the same words as data.
- How long a decision waits. Count the decisions in the last quarter that waited for a person outside the business to bless them. That count is the reading.
You can also hear the filter in the words people choose. External reference talks in report: "they said", "the market wants", "the board will never". Internal reference talks in perception: "I know", "I could feel it going wrong", "my read is".
One caution before you run the markers on yourself. The filter shifts with context. A founder who runs an internal standard on product can run a purely external one on pricing, and the same person will sound completely different in each conversation.
The Protocol: Six Steps on One Live Decision
Take one live decision, the one you have been circling this week, and run it through the six steps in order.
- Name the decision and the last verdict you accepted. Write the decision in one sentence, then write the last occasion you let somebody else answer it for you. Naming the pattern takes two minutes and it settles most of the argument about whether the pattern exists.
- Declare your bar before the number lands. Before you look at any data or ask any opinion, write what success looks like, what it will look like specifically, and what evidence would change your mind. This single step does more work than the other five, because a standard written in advance cannot be reverse engineered from the reaction of the room.
- Split the page into two columns. One column holds your read. The other holds the evidence you actually have. Fill both without editing either. Where the two columns agree, move. Where they disagree, you have located the real decision and you now know which kind of decision it is.
- Ask for feedback about the performance rather than the person. "What do you think of my plan" invites a verdict. "What did you see me do, and what happened next" invites data. The second question returns the correction that keeps an internal standard honest.
- Set a deadline and a single owner. Internal reference decays inside a committee. Write the decision, the date it will be made, and the name of the person who will make it. At the deadline you decide with the evidence in hand, and the evidence informs the decision without issuing it.
- Score three past decisions once a quarter. Pick three calls you were confident about and compare them with the outcome. The record keeps the internal standard calibrated. Handing the verdict away keeps it unstable. Scoring yourself keeps it honest.
Two of these steps do the heavy lifting. Step two protects the standard from being rewritten after the fact, and step four keeps it in contact with reality. Run them together and the filter stops swinging between permission-seeking and wishful thinking.