An insurer measures seven touchpoints on its home claim journey, and every one of them sits at or above target: satisfaction between 4.2 and 4.6 out of 5, first-contact resolution at 82%, an app rated 4.4 in the stores. About a third of the people who start a claim in that app end up phoning in anyway. The distance between those two facts is the practical difference between touchpoint-centric and journey-centric customer experience, and it has nothing to do with which of the two matters more.
A touchpoint-centric approach makes the individual interaction the thing you own, measure and improve. A journey-centric approach makes the whole end-to-end journey that thing, including the waits, handoffs and repeated steps between interactions. Both are real units of work. Only one of them can be the default that your dashboards, owners, budgets and review meetings are built around, and that default decides which problems your organization is capable of seeing.
Journey-led customer experience is the operating model built on the second answer. The case for it rests almost entirely on what the first one structurally cannot show you.
The difference in one line
A touchpoint-centric model treats one interaction as the unit of work: a call, a form, a page, a store visit. Whatever happens between interactions reaches your numbers only when it damages an interaction enough to move its score. A journey-centric model takes the whole journey as the unit, a claim or an onboarding or a renewal, and reads interaction scores as contributors to that outcome.
| Compared on | Touchpoint-centric | Journey-centric |
|---|---|---|
| Unit of work | One interaction or channel | One end-to-end journey |
| Who owns it | The team that runs the channel | A named owner across teams |
| What gets measured | Interaction quality and efficiency: CSAT, first-contact resolution, handle time, app rating | Outcome and effort across the sequence: completion, repeat contact, time to resolution, journey-level satisfaction |
| What counts as an improvement | This interaction got better | The customer got through with less effort |
| Where failure hides | Between the interactions | Inside an interaction nobody drills into |
| The review question | Why did this score drop? | Why do a third of these customers end up calling? |
Read that last row in both directions. A touchpoint model hides failure between interactions, and a journey model can hide it inside one, flattening a genuinely broken upload form into an acceptable average. That is a real cost of switching, and the reason interaction-level numbers stay in the picture.
Why touchpoint scores don't add up to a journey
Touchpoint measurement is not a smaller version of journey measurement. It measures a different thing, and the two do not aggregate: seven interaction scores, however carefully collected, cannot be summed into a statement about the claim. On the insurer's journey, all seven scores held or improved across two quarters while the share of app claimants who ended up phoning rose.
Interactions are not independent of each other
What happens at one interaction changes how the customer arrives at the next. Someone who has already entered the same policy number twice reaches the third form with less patience than a first-time visitor, and that form gets scored lower for a problem it did not create.
Lemon and Verhoef, writing in the Journal of Marketing in 2016, synthesize modeling work on carryover and spillover effects between touchpoints, where the measured effect of one interaction depends on the ones before it. Averaging conditional measurements produces a number with no clear referent. The same paper identifies four categories of touchpoint: brand-owned, partner-owned, customer-owned, and social or external. A touchpoint program can reliably instrument the first and part of the second, while a claim journey runs through all four.
Customers cut the journey in different places than you do
Your stage model is an internal artifact, following the shape of the organization, the systems, or the order in which work gets handed on. The customer's sense of where one part of the experience ends and another begins comes from noticeable changes in what is happening to them.
Reitsamer and Becker, in the Journal of Business Research in 2024, found that customers partition their own journeys using distinctive changes within, between and across touchpoints rather than the firm's stages. When the two boundaries disagree, one bad experience gets split across two measured touchpoints and reads as a small dip in each. Nothing breaches a threshold, so nothing gets escalated.
The seams carry the effort and usually carry no owner
The intervals between interactions are where a journey either holds together or stops making sense to the person in it, and they are the part of the experience most likely to have no metric and no name attached. Three of them sit between the insurer's app submission and its first adjuster call, and none appears on any dashboard, because each falls between two teams whose own numbers are fine.
- Silent wait. Time passes with no status and no expected date.
- Handoff. One team's work ends and another's begins, with nobody confirming the transfer to the customer.
- Re-entry of information. The customer supplies something the organization already holds.
- Forced channel switch. The current channel cannot finish the task, so the customer starts again elsewhere.
- Return after abandonment. The customer comes back to a half-finished attempt and finds no memory of it.
Finding these is cheap: walk one journey end to end with the people who run each part of it, and note every point where the customer is waiting, repeating themselves, or starting over. The difficulty sits downstream of detection, since repairing a seam takes a decision that spans two budgets, and a touchpoint-centric operating model has no venue for that decision.
Smaply maps the waits and handoffs between touchpoints alongside the interactions, so the seams are visible.
What changes when the journey becomes the unit
Going journey-centric gets described as a mindset shift, which makes it sound optional and impossible to check. In practice it shows up as four changes, each visible in a document, a meeting or a budget line.
The number at the top of the report changes
A journey-level measure spans the whole sequence: completion or resolution rate, repeat contact rate, total elapsed time, and one effort or satisfaction question asked after the journey ends rather than after each interaction. On a claim, "did this get resolved without you having to chase it" is a different question from "how was your call."
Widely cited research puts journey performance at roughly 35% more predictive of customer satisfaction and 32% more predictive of churn than performance on individual touchpoints, with a wider spread between best and worst performers on journeys than on interactions. If everyone in your category can run a decent call center and a decent app, interaction-level parity buys you nothing. Touchpoint metrics stay, moving from headline to diagnostic layer, which is how you find out where a journey number moved.
Somebody's name goes against a journey
A journey owner is a named person accountable for a journey outcome across teams they do not manage, with a standing forum, a budget line and access to the evidence. The role fails in a predictable way: an owner who has a target and no authority is a reporting line with extra meetings. The version that works holds authority over sequencing and instrumentation even where it holds none over headcount. For the insurer, one name went against the stretch between app submission and first adjuster contact, which had belonged to two teams at once and therefore to neither.
The work you can fund changes shape
Under a touchpoint model, fundable work is scoped to one interface or channel, because that is where the owner and the budget live. Status messaging during a five-day wait, carrying context across a channel switch, killing a data re-entry: none of these sits inside a single interface, so none has a natural home in a business case. They tend to be cheap fixes with no owner rather than expensive fixes nobody prioritized.
The review meeting asks a different question
"Why did app satisfaction drop half a point" and "why do a third of app claimants end up phoning" get answered by different organizations. The second needs three artifacts to exist: a journey definition with named endpoints, a measure that spans it, and evidence attached to the steps. Customer journey management is the practice of keeping those three current between reviews, and it is what stops a journey program decaying back into a list of channels with owners.
How to tell whether you're journey-centric or touchpoint-centric
Strategy decks are unreliable evidence here, since nearly every CX strategy document written in the last five years claims to be journey-led. The artifacts are more honest. Five checks you can run this week:
-
Read the row labels on your top CX reportIf they are channels, teams or interaction types, the report is touchpoint-centric whatever the cover slide says.
-
Ask who chairs the meeting where a cross-team experience problem gets decidedIf that person owns a channel, the seams have no advocate in the room.
-
Look for a name against a waitIf every owner in your CX documentation owns a screen, a queue or a channel, the intervals between them belong to nobody.
-
Find the last three funded CX improvementsCount how many changed something between two interactions rather than inside one. Zero out of three is common.
-
Take one recent complaint and try to place itIf it belongs to two teams, or to none, your unit of measurement is smaller than your customer's problem.
Failing three or four is the normal result rather than a sign of neglect. Channel-based org charts produce touchpoint-centric measurement without anyone choosing it, because reporting follows structure and structure follows channels.
Where touchpoint-level work is still the right call
Plenty of problems are genuinely contained inside one interaction, and treating those journey-first adds coordination cost for nothing. A defect with a single owner and an obvious fix should be fixed by that owner this week. Efficiency work on a high-volume, largely self-contained interaction is best scoped to that interaction, where the gain is arithmetic. Accessibility and regulatory obligations attach to specific interfaces and have to be met there regardless of how the journey performs around them.
The deciding question is whether the problem is contained inside one interaction or produced by the relationship between several. Get that wrong in either direction and you either convene a working group to fix a broken button, or ask one team to solve something that was never inside their control.
Making the shift without reorganizing the company
Nothing above requires a new org chart, and treating it as a restructuring program is the most reliable way to stall it. The smallest version that works:
- Pick three to five journeys with real volume and real cost. Define each by its first and last step, precisely enough that two colleagues would draw the same boundary.
- Instrument the seams before adding any new survey. The waits, handoffs and re-entries need timestamps and counts more than they need opinions.
- Put one journey-level number into a review that already happens, next to the touchpoint numbers rather than instead of them. A new row is a smaller ask than a new meeting.
- Name an owner per journey and give them the sequencing and instrumentation decisions, even if headcount stays where it is.
- Keep the journey definitions and their evidence current. A journey nobody maintains reverts to a list of touchpoints within two review cycles, which is the shape the rest of the organization is already in.
The insurer's repair was that unremarkable: named endpoints, status messaging during the wait that had been generating most of the calls, and the seven touchpoint scores kept exactly as they were, one level down.
Switching costs you things worth naming in advance. Clean attribution goes first, because when a journey number moves the answer to who moved it is a group, and the app team that spent a quarter shipping a better upload flow no longer gets its own line in the story. The fast local fix goes second, since sequencing now runs through someone weighing your interface against a wait two steps upstream. The tidy scorecard goes last, the one where every owner could point at a green number and be right. Channel leads feel that loss first, in the monthly review, where the number at the top of the page is now one that none of them owns by themselves.
Connect live journey metrics to the map they belong to, so the number and its evidence sit together.
Frequently asked questions
Is journey-centric CX just journey mapping with a new label?
No. Mapping produces an artifact, while a journey-centric model decides what gets owned, measured and funded. You can run a mature mapping practice and stay entirely touchpoint-centric: the maps get made, the insights get presented, and every improvement that follows is scoped to one channel, because that is where the budgets sit.
What happens to our NPS and CSAT programs?
They stay, and they change position. Interaction-level scores become the diagnostic layer underneath a journey-level measure, which is what tells you where a journey number moved. If you report one thing to a board, report the journey outcome, because that is the number a customer would recognize as their experience.
Can every touchpoint score well while the journey fails?
Yes, routinely, for two reasons at once. Interaction scores are conditional on the interactions before them, so they cannot be added into a journey-level statement, and the intervals between interactions are not measured by any of them.
Who owns a journey that crosses four departments?
One named person with a journey outcome target, a standing forum where the four departments show up, and authority over sequencing and instrumentation. They usually have no authority over headcount, and the model survives that. What it does not survive is an owner given the target alone, who then escalates every real decision to a steering group that meets monthly.
How many journeys should we manage this way?
Three to five to start, chosen by volume and cost rather than by which team volunteers. One proves nothing to a skeptic, since a single success is easy to attribute to attention rather than method, and twenty becomes a documentation project nobody feels accountable for.
