Twelve dimensions, twenty combinations: how to actually score a client relationship

Most retention scoring is a satisfaction survey wearing a lab coat. You ask the client how things are going, they say fine because saying anything else is awkward, you log a number, and the number tells you nothing you didn't already fear.

The problem isn't the math. It's the input. You're asking the wrong person. The client is the last to admit a relationship is cooling, often because they haven't admitted it to themselves yet. The person who actually knows is you. You just haven't had a structured way to write down what you already sense.

That's what a real relationship score is for. Not to grade the client. To force you to be honest about what you're seeing.

Why a single number lies

A relationship is a shape, not a score. Two clients can both sit at the same overall health and be in completely different situations.

One trusts you completely but is quietly shopping competitors. The other is loyal as a dog but micromanages every deliverable because they don't trust you to make a call. Same number. Opposite problems. Opposite moves. If your scoring collapses both into "68, watch closely," it has told you nothing about what to actually do on Tuesday morning.

So you score the dimensions underneath the number, and you read them together.

The twelve dimensions

Think of them in two groups: the ones that decide whether the relationship survives, and the ones that tell you how to handle it.

The heavy ones, the dimensions that carry the most weight because they predict churn most directly:

• Trust. Do they trust you to do your job? When trust is low, you get micromanagement, second-guessing, approval bottlenecks. High trust is delegation.
• Loyalty. Are they looking at other options? Low loyalty is a client actively comparing you to alternatives, even if they're perfectly pleasant about it.
• Expectations. Are their expectations realistic? Misaligned expectations are a ticking clock; every week you don't address the gap, it widens.
• Grace. When something goes wrong, how do they react? Zero-tolerance clients turn one bad week into a relationship crisis. Clients with grace give you room to recover.

If you only score four things, score those four.

The supporting ones add texture, and several of them are U-shaped, meaning both extremes are the risk, not just the low end:

• Budget Commitment. Will money be the reason they leave?
• Relationship Depth. Is there a real relationship beyond the work? Moderate is healthiest; zero is fragile, but total enmeshment is its own kind of unstable.
• Replaceability. How embedded are you? More embedded is safer.
• Communication Tone. Warm and direct, or cold and clipped?
• Decision-Making Authority. Can your contact actually say yes, or are they a relay to someone who can?
• Communication Frequency. Here's the U-shape: radio silence is a risk, and so is constant frantic contact. The healthy zone is a rhythm.
• Stress Response. Another U-shape. The client who goes silent under pressure and the client who escalates loudly both create risk, in opposite directions.
• Reporting Need. "Don't bother me" and "send me everything" are both harder to sustain than a steady middle.

Read the combinations, not the columns

The insight isn't in any single dimension. It's in the two or three lowest ones, read together.

Low Trust plus low Grace is a different animal than low Loyalty plus high Trust. The first is a client who doesn't trust you and won't forgive a mistake, which means you operate with no margin for error and should be over-communicating every step. The second is a client who likes your work fine but isn't committed, which means the move is to deepen the relationship and make leaving feel like a loss, not to tighten your process.

There are roughly twenty of these combinations worth naming, and once you start looking for them, you stop being surprised by departures. The "looks good" client whose tone has cooled and whose engagement has dropped isn't a mystery. It's a pattern with a name, and a play.

Always read against baseline

The last rule, and the one people skip: a score only means something compared to where it was.

A client steady at 70 for six months is stable. A client at 70 who was at 85 a quarter ago is in freefall, and 70 is just the number you happened to catch on the way down. The absolute value is almost meaningless. The delta is everything.

This is why scoring once and filing it away is worthless. The relationship moves. Your read has to move with it, which is the entire reason to keep the profile current instead of treating it like a form you filled out at onboarding.

What the score is actually for

It is not a verdict. Low Loyalty doesn't doom a relationship, and high Trust doesn't make one safe. The profile tells you where to look and how to communicate. It does not tell you who's leaving.

What it does is end the guessing. Instead of a vague unease about an account, you have a specific read: here's the dimension that slipped, here's what that pattern usually means, here's the conversation to have this week. The number gets you to look. The shape tells you what to do.

Retayned scores all twelve dimensions for every client and reads the combinations automatically, so the pattern surfaces before you'd have noticed it on your own. How those dimensions combine into a single Retention Score is the engine; what you do with it is the job.