Almost every organization has a customer journey map. Far fewer use it to make decisions. In most companies it gets referenced, pulled out when needed, and put back in the drawer when the meeting ends.
The usual response is to blame the map. Rebuild the dashboard, break the stages down further, attach more data. But the actual cause sits somewhere else: the definition of "customer" is vague.
Certain phrases repeat in meetings about markets and customers.
They all mean the same thing. The team is judging against an average customer rather than a specific one.
The problem is that the average customer does not exist. No amount of extracting shared behaviors produces a coherent profile. A journey built on that vague definition stays abstract, and because it feels unreal to the people working on it, they stop paying attention. Everyone ends up watching only the metric for their own stage.
A journey that actually works should answer questions like these:
If those go unanswered, the journey is a plausible-looking picture — a compliance artifact you keep because everyone else has one.
So the baseline has to be rebuilt. The useful category is neither "valuable customers" nor "VIP customers" but favorable customers — defined by attitude rather than spend or demographics.
Favorable customers show a specific profile:
The word "customer" is doing work here. A customer, at root, is a guest who might come back. Coming back is what makes someone a customer. Someone who visited many times through yesterday may stop today, and that possibility exists everywhere.
Favorable customers matter not because of short-term revenue but because their churn probability is markedly lower. They raise the floor under traffic and revenue and keep it from dropping further. This is the territory now often called fan business. On designing fandom as an asset, see Your Channel Gets Searched Before Your Resume, which describes the same structure from the individual side.
"But can we actually identify these customers in our data?"
Most organizations stall here. Not for lack of data — there is usually plenty. The problem is shape.
A typical service's data splits like this:
None of it is connected as one person's flow. The customer exists not as a persona but as a scatter of events. You cannot tell who generated the data or why a trend looks the way it does.
The organization then splits along the same seams. Marketing owns the "acquisition volume" event, product owns the "session time" event, operations owns the "complaint count" event. Each team can judge whether its own metric is a good or bad signal, and nobody can judge which customers are favorable.
The fix starts with a shift in framing: stop dividing customers by amount or attribute, and divide them by behavioral flow.
Not every service follows this, but favorable customers generally show shared patterns inside a product:
Crucially, these are observable signals in data you already hold — not intuition. They only require stitching event-level data back together at the person level.
Segmentation comes next, and it starts from purpose of use and in-product behavior, not payment amount and tenure.
Define which traits and dispositions mark your most favorable customers, then connect that group to how satisfied they are with the service. Then compare two things: how efficiently you spend to keep transacting with them, and whether they in turn are solving their problem at a reasonable cost. Both sides have to hold for the relationship to continue.
One question sits underneath all of it: who is this service supposed to keep?
Decisions made without answering it always default to an uncertain average. Average-based decisions satisfy no one in particular, and because it is unclear who they were for, nobody can be accountable for them. They look like the safest choice and are actually the least responsible one. At worst, they accelerate the departure of the favorable customers you still have.
Use these to test whether your journey functions as a decision tool:
When all five have answers, the journey becomes evidence rather than decoration. For a related look at reading performance metrics from the brand's side, see Reading Olive Young's Retail Media From the Brand Side.
Usually it's the baseline, not the map. A journey drawn without a clear customer definition implicitly assumes an average customer, and since that person doesn't exist, the journey stays abstract. Teams then default to watching only the metric for their own stage.
VIPs are defined by spend or attributes; favorable customers are defined by attitude and behavior. They try to understand the product's value, use it repeatedly and explore on their own, don't leave at first friction, and have room to go deeper. Their low churn probability is what raises the revenue floor.
Segment by behavioral flow rather than amount or attribute: self-directed feature exploration after signup, sustained repeat use of core features, and reporting friction instead of churning on it. This only works if acquisition, usage, payment and complaint data are stitched together at the person level.
To apply what you just read to your own site, start with a free audit of where things are now.
A strategist replies within 24 hours on business days.
Jensen Huang reads reports from every level, and founders everywhere took it as permission. But the variable is not frequency — it is whether your gate controls execution or collects information.
Assessments fail for two reasons: purposeless design and unexecuted results. Here is how to design an organizational diagnosis and actually use what it produces.
Cohort retention analysis exposes what growth rate hides. Here are the three curve shapes, benchmarks by category, and what investors actually read off the chart.