Working the model

How to build this up from nothing, keep it honest, and get value from it early.

Start in the middle, not at the top

The instinct is to begin with the mission and work down. It is the slowest possible order, because the mission is the hardest thing to write and the least useful in isolation.

Start where the knowledge already is:

  1. One persona you know well, and the situation you meet them in.
  2. What they are reaching for — the goal, in their emotional language.
  3. One expectation — the measurable version, in that situation.
  4. What measures it. If nothing does, you have learned something on day one.

Then widen. A second context for the same persona usually produces a second expectation immediately, and that is the moment the persona-versus-stakeholder split stops being abstract.

The mission and the beliefs get easier once several goals are written down, because by then you can see what you have been assuming.

Let Paroksh draft it

Paroksh is faster at this than you are, and wrong in ways only you can see. That combination is the point.

Describe the business out loud — who you serve, what you promise, how you would know it worked — and read what comes back. Two things will happen:

  • Most of the structure will be roughly right, and would have taken you an afternoon to type.
  • Some of it will be confidently wrong in a way that is obvious to you and invisible to them, because they have never met your customers.

The second is the valuable half. The parts they get wrong are exactly the parts that carry your actual knowledge, and seeing which ones those are is the fastest orientation available.

Everything arrives as a draft. Correct it, then activate what you agree with. Until you do, it is a proposal — which is what keeps the model holding what you actually believe rather than what was plausible.

Correct early and bluntly. A wrong belief left standing propagates into the goals and strategies that root in it, and unpicking it later is much more work than saying "no, that's not why they leave" in the first ten minutes.

Keep it honest

Grade conviction truthfully. The temptation is to mark beliefs VALIDATED because it feels like progress. The value is entirely the other way: a strategy resting on three ASSUMED beliefs is a strategy with three named places to look when it stops working.

Let things fail their gates. A story that will not commit, a goal with no expectation, an expectation with no metric — each is the model reporting something true. Routing around them is how a graph of intent becomes a graph of fiction, and the failure is silent because everything still looks filled in.

Archive rather than abandon. A strategy left active while everyone has stopped working on it makes every roll-up wrong and keeps the initiatives beneath it looking justified.

Revisit on a cadence. Beliefs decay, contexts change, personas get reclassified. A quarterly pass over convictions and lifecycles is enough, and it is where most of the value shows up after the first month.

Two questions to ask regularly

These are cheap to ask and awkward to answer, which is a good sign.

Is there a belief that one of our strategies and one of our customers' goals both root in?

That coincidence is alignment as a fact rather than an assertion. Its absence is not a crisis — it is a conversation you would otherwise have had to be tactless to start.

What have we delivered and never validated?

Most backlogs hold far more of it than anyone believes. Here it is a queryable state rather than an archaeology project.

Common mistakes

MistakeWhy it hurtsFix
Goals written as KPIsyou have recorded the proxy and lost the thing it proxieswrite the emotional version; the KPI is the expectation
One persona per customer segmentsegments are a pricing decision, not an identityfewer personas, more operating contexts
Expectations nothing measurescannot be validated, everadd a metric, or record that it is deliberately unmeasured
Restating links in prosethe copy drifts and the drift is silentlink; author only the capability
Everything VALIDATEDthe conviction axis stops carrying informationgrade honestly; ASSUMED is the useful state
Treating this as a trackercommitments quietly become statusesexecution state lives on the work items, not here

Where it pays off

The model earns its keep at three moments, and none of them is the day you write it:

  • Starting a piece of work — the five-hop walk from story to goal tells you who wants this, in what situation, and what would count as it working. Before you build.
  • Being asked why — the answer is a link, not a memory.
  • When something stops working — you can walk from the disappointing metric back to the expectation, the goal, and the belief the whole thing rested on. Usually the belief is the part that was wrong, and usually it was marked ASSUMED all along.

That last one is the real return. Most organisations discover a bad assumption by running into it. This one writes it down first, in a place you will look.