How to Fail at NLP Modeling (and How to Plan to Succeed!)
In today's post, I share how and why an NLP Modeling project could be set up to fail, and what to do instead!
A few simple distinctions separate Failing from Succeeding at NLP Modeling.
Years ago I was asked to do some minimalist NLP-based Sales training for an organization that held monthly sales meetings. I was asked to provide one four-hour training session, once a month, for a group of sales professionals (times two, for two different groups at their company).
If you know anything about soft skills training to medium to large groups, you know that four hours is next to nothing. Its enough to get people thinking, and its enough to inspire them, and its enough to get them listening a little differently. One-on-one coaching can be profoundly life-changing even just in minutes, let alone four hours, but in group training contexts (where most of the attendees do not have a background in NLP), its not really enough for immersion training, so just four hours per month are unlikely to be sufficient to change anyone's habits in any quick and heavily measurable way. You might get a little mileage out of that, but probably not a lot.
I told the client all of the above, and the response was "Unfortunately, that's what we've got. Lets do the best we can."
While having a private discussion with the managers, I happened to mention the idea of NLP Modeling, which they didn't really know about. NLP Modeling, in business contexts, can be used to model their highest performers, so as to then train the rest of their team in whatever their highest performers were doing both consciously AND unconsciously. The results usually include reduced training time, as well as significantly improved sales or performance, customer service quality, etc. If you apply this to enough employees, this could result in many millions of dollars of saved losses, and/or additional earnings.
For the most part, they had come to NLP from the intention of using NLP to be more persuasive salespeople. When I discussed NLP Modeling with them, they got very excited. But they still weren't willing to step sufficiently outside of their map of "do what's possible in a couple of hours."
So when they started thinking about how they could have me spend just an hour or two modeling their top sales person (in front of all the other sales people, no less), the modeling project they had in mind was doomed from the start.
So what are some of the Top Factors for Success with NLP Modeling? (In any domain, not just Sales/Business).
- Full Access to the Exemplar for NLP-styled interviewing. And not just for an hour or two. You need access until the model is built. The duration will depend on many things.
- Optimistic, Interested involvement from the Exemplar (VERY helpful; not always necessary, but VERY helpful). Without this, the cost/duration of a modeling project can massively expand if you want to achieve accurate and effective results.
- Continued Access to the Exemplar after a model has been encoded, for feedback and refinement purposes while testing, validating, verifying the model.
- Time to encode and design a training process
- Time to train the model to other people
- Time to measure and discuss the results (which sometimes leads to another round of refinement).
What happens to an NLP Modelling project when you don't have all of the above Six Factors covered?
- When you're asked to "do the best you can in a couple of hours:" The results you get are an utterly incomplete cognitive model, which isn't likely to help anyone -- and in fact, it could be detrimental to others. Best case, you get the same checklist of 'gee this is what I *think* the high performer is doing' that one of the managers at the company would produce after months of the manager knowing them. Worst case, what you acquire will slow all of the other employees down further, and annoy the high performer to the point of hurting their results. You need to allot sufficient time for the NLP Trainer to *acquire as sufficiently complete a model as possible of the desired behavior.*
- If you don't have cooperative optimistic involvement from the Exemplar, then your model isn't likely to be accurate. You can 'trick' an exemplar into leaking other than conscious information but this is far too time-consuming and arduous a process to apply in the process of actual extensive behavioral modeling. Also, a modeling exemplar could intentionally throw you off the trail with red herrings. Which are not impossible to see through, but certainly an unwanted challenge. Best case, it doubles your modeling time. Worst case, an accurate model is never acquired, and/or an incorrect model is taught to others and that degrades performance even further.
- You might have a resistant exemplar. They may not be used to 'being under the microscope', and that could be a more important factor if the modeling is done in front of the rest of their team. They may have concerns about their image or how they're perceived, as the top performer. If they answer a lot of questions with "I don't know," that may make them feel self-conscious.
- If getting access to a modeling exemplar is difficult after the initial modeling has been completed, it makes it hard to test and refine the model. The quality of the model is often judged when it's not yet at its most refined level. Modeling is almost never completed just after a first pass at the exemplar. Measuring of results ought to be delayed until after a refinement process has been completed, during which, the exemplar needs to be available to confirm/deny accuracy of (the first version of) the model in action. Best case -- the 1st cut of the model gets minor incremental improvement in results or training time for some of those who'll learn the new model (instead of massive improvement for the vast majority of trainees). Worst case, myopic managers incorrectly judge the project prematurely based on results that aren't expected to ramp way up until after a model has been refined, and will NEVER realize the massive benefits of high-performance NLP/behavioral modeling.
- If the Modeler doesn't have sufficient time to encode the model properly and design the right training process, then the trainees will never learn the high performer's model well. And don't leave this to people not trained in NLP-based training unless the Modeler has trained the trainers. They don't have the experience to know how to 'install' a high-performer's model that includes more than behavioral checklists -- such as emotional state transitions, and cognitive representations such as visualizations or auditory experiences.
- If a myopic manager tries to fit the NLP Trainer's training time into some arbitrary window (i.e., "You'll have 30 minutes with each group...") then you may as well just go home and wish them 'good luck.' The training time takes what it takes. Give the NLP Modeler ample time to train trainees the first time through, until high performance results are achieved, and see how long it takes. Then, look for opportunities to streamline. Pre-determining a very short maximum training time for something you don't fully understand is like killing the project before it even gets off the ground. Any experienced NLP Modeler would tell you that in advance.
- There needs to be analysis and refinement time, and usually will be. Oddly, this is the part that most managers will agree to, because by the time a first training round has occurred, chances are, you'll have already seen some impressive results. Approving further consulting time, once those results have been quantified, is usually seen as the smart investment they originally hoped it would be, instead of an initial worrysome cost.
Why would an Expert/Exemplar / High Performer not cooperate?
Pretend you're the top salesperson in your team. You earn the most. You get the highest bonuses and rewards. And significantly, your team's income and incentivization structure includes competition. A limited pie divided up among competititors. So you have a position to protect.
So you're actually incented to keep outperforming your co-workers at lower sales targets. It helps you to stay at the top. It doesn't help the company, or your managers, but it helps you. So here we have a classic conflict of interest.
Then your manager comes to you and says, "something you're doing, or not doing, outperforms everyone else. We're bringing in an NLP Modeler who's going to capture what you do, or don't do, and is going to teach it to everyone else, so we can get everyone performing at your level."
To do this, they're going to require you to participate in NLP modeling sessions, using time you'd probably actually have spent... selling. Doing your job. And earning money. So not only are they not paying attention to the conflict of interest, they actually want to hinder you from reaching your potential this month (and maybe next), in order to cooperate. And if you don't, of course, they'll say "you're not a team player."
Now, they can try and demand your cooperation, they can try and strong-arm you, but if they're not smart enough to recognize the outrageous conflict of interest they're creating for you, then they truly deserve the pathetic results they'll get when you tell them "of course I'm a team player..." then give the NLP modeler disinformation, and act surprised when the modeling project doesn't work as intended.
When I meet managers and business owners who attempt to structure things like this, I always aim to educate them about the pitfalls of ignoring major conflicts of interest, and incentivizing your exemplars. You really should properly reward them for their full and active cooperation. Don't skimp. It's the difference between failure and success.
Can you model from books, audio or video?
Remember what modeling is. It's the process of capturing as much information as possible from an exemplar about their cognitive strategies, their somatic (physical skills), their emotional state(s), their conscious and unconscious knowledge, and how all that ties together. If at all possible, you're going to want and need direct access to a living, breathing, cooperative expert!
So, yes you can still model some things from books, audio, or video... but the results will never be as in depth, or as accurate, as they would if you had lots of one-on-one time with an exemplar. Video is better than audio, audio is better than books, and modeling from books and other written records alone is, simply put, unlikely to be taken too seriously in the NLP Community.
The most successful modeling projects that I know about were done in contexts where:
- The modeling exemplar did not have any dis-incentive to cooperate; on the contrary, the modeling exemplar was proud of their expertise, and was flattered that so many people wanted to share it more readily with others. They were committed to contributing to the project.
- The people paying for the project were different than the managers, and those who commissioned the modeling process had a long-range large-scale financial perspective. If the person asking you to do the modeling is the person that signs your checks, and they're trying to "nickle-and-dime" the project, the best you can do is say "no" and go home. You may as well avoid associating yourself with a project doomed to fail from the start.
- There is direct access to the primary exemplar(s) until a model can be acquired, codified, and then tested and refined several times.
Here's how NLP was used to improve the US Army Pistol Marksman Training
Some of you may be aware that NLP Modeling was used to improve the US Army Pistol Marksman training. According to NLP Master Trainer Eric Robbie: the primary NLP Modelers for that project were LTC Robert Klaus, Wyatt Woodsmall, Richard Graves, Paul Tyler, John Alexander, and Dave Wilson. A then-young Anthony Robbins was brought along by Wyatt Woodsmall to help train the successful marksman learning patterns to other army personnel, after the primary modelers had elicited it (Robbins was not a key player as some have heard or assumed through the rumor mill, but he was there to assist).
Results? The qualifications back then were Marksman, Sharpshooter and then Expert (from basic to advanced, in that order). A Marksman has to get 30 hits on target from 45 rounds fired. One group of soldiers received the nlp-based training, a second group of soldiers (the ‘control group') received standard army pistol training. The control group took 27 hours to get 73% of their soldiers to Marksman (only 10% made Expert). The NLP-based group took 12 hours to get 100% of the soldiers to Marksman level (and 25% made Expert). (Thank you Eric, for keeping the facts about this project alive!)
Success Factors? The modelers modeled successful exemplars, and were able to distill down the key elements leading to more rapid success, and train it sufficiently well to a group of new trainees, and a proper opportunity to measure the results was allowed. The exemplars were not dis-incented to cooperate; their income was not based upon doing an activity in some other location; their time was not being co-opted at their own loss. And they were getting attention for their high performance in a way that did not threaten their performance levels. Simply put, this modeling project involved win-win, not win-lose.
What sorts of modeling have I done? Many of my courses over the years were the result of my own modeling projects.
As a Knowledge Engineer, prior to even learning NLP, I modeled (and then built software to replicate the decision making for):
- Check Guarantee experts -- for the purposes of building an expert system for approving/declining bank checks at retail points of sale.
- Credit Report Incoming Data Correction experts -- for the purposes of an internal project at a major credit bureau (with massive implications on the consumer credit system, that I cannot discuss the details about). One outcome was that it took a year for the credit bureau to realize how utterly inconsistent the knowledge base was across their experts (they couldn't agree on anything), and how little the upper managers knew about these inconsistencies. Our project cleaned all that up, and gave the company one consistent set of rules for moving forward.
- Chase Bank Silver & Gold Customer Service Reps -- for the purpose of building intelligent software to maximize client satisfaction and further financial product buy-in.
After learning NLP and opening my training business, I continued to model exemplars met during my global travels, whenever I met with "high performers."
- Charisma -- for the purposes of teaching others (became a 3-CD-set called "Charisma Fuel")
- Humor -- for the purposes of teaching others (became my "Becoming Outrageously Funny" workshop)
- My own work with Beliefs -- for the purposes of teaching others (became my "Knowledge Engineering" home-study and live courses, and half of "Belief Craft").
- Great Interviewing (both roles) -- for the purposes of teaching others (became my "Own the Interview" course)
- Great Public Speakers -- for the purposes of teaching others (became my "Speaking Ingeniously" course)
- Great Salespeople -- for the purposes of teaching others (became my "NLP Sales Wizardry" course).
- (and more that I just don't write about!).
If you have a need to acquire or re-acquire some amazing expertise from one of your employees, or experts, or associates -- especially before they retire or jump ship to go somewhere else -- consider hiring a skilled NLP Modeler today, and cut your losses before they could happen. It's the fastest way to replicate someone's expertise, and share it with others, successfully.
author: Jonathan Altfeld