Touchpoint

For a long time, what product teams called "the product" meant the inside of the screen. Button placement, copy, and flow were what got improved, and anything outside that belonged to another team.

Customer experience, though, gets built outside the screen too. The text message that arrives when a payment fails, the sentence a support agent writes back, a two-year-old blog post sitting in search results, the line item printed on an invoice — to the customer all of that is the same product. None of those places appear in a product team's review deck, so nobody fixes them.

Pagedeck ran into this directly. Mia rewrote the signup screen three times trying to lift conversion and the number barely moved. Digging into the records of people who dropped out showed that half of them hadn't left the screen at all: they received the invite email, opened it, and disappeared. The wording in that email had been written two years earlier and had no owner. Nobody had chosen it badly; nobody had chosen it at all.

A touchpoint is the unit that keeps places like that from going missing. Once you see the product as a set of places a customer touches rather than as a screen, the board of things you can fix widens from the inside of one page to the whole journey.

It works as a counting unit. Sorting good contact from bad contact isn't the first step. Counting all of it without gaps is.

What you include is the whole point of the unit. Screens and emails, yes, but also text messages, invoices, app store reviews and the replies underneath them, search results, and the sentences a support agent sends. Places where a person is involved and places that fire automatically get counted the same way, because from the customer's side both of them came from the company.

The counting goes in three passes.

1. Lay them out in time order, from the moment someone first hears about the product to the moment they leave. Sequencing them exposes where the same thing gets said twice and where a necessary explanation arrives too late.
2. Write down an owner for each one, plus who wrote the words in it. This is the pass that pays for the exercise. A full list usually runs to somewhere in the twenties, and around half of the owner cells come back blank.
3. Write down what the customer expects at each one. Somebody reading a failed-payment message wants to know what to do next, not to receive an apology. Writing the expectation down is what turns the cell into a spec instead of a label.

Touchpoints don't carry equal weight. The few where an impression hardens get marked separately as a moment of truth, and the places where customers stall or turn back get marked as a pain point. Marked cells become the list of things to fix. The rest stay on the list as counted and known, which is already more than they were before.

Four places.

1. The skeleton of the journey map — the map's horizontal axis is the touchpoints laid out in time order. Skip filling that row and draw the emotion curve first, and the curve rises and falls on nothing.
2. Finding the unowned ones — a touchpoint with a blank owner turns out to be an organizational gap rather than a product one. The usual state is that engineering's original string still goes out in the text message and every agent writes support replies their own way.
3. Matching the voice — when the voice differs across touchpoints, the set doesn't read as coming from one company. Speak casually on screen and send a stiff notice by text and the customer can't tell which one is the real product. Consistency here is cheaper than any of the other three, since it costs editing rather than building.
4. Gathering the material — what fills the cells comes from three sources. User research means meeting people to watch and ask, voice of the customer (VoC) collects the complaints that already arrived on their own, and a persona — several people's stories consolidated into one figure — takes the lead role on the map.

When the list is complete and it still isn't clear why one place performs badly, the next move is a service blueprint. It draws internal steps as layers underneath the row the customer sees, and that's where a four-day refund turns out to be caused by one person checking a queue once a day.

"I've rewritten the signup screen three times and the number won't move. Let's write out every touchpoint."

Proposing to check whether the problem sits outside the screen.

"Who writes this text message? If it has no owner, a two-year-old sentence is still going out."

Pointing at a blank owner cell, and saying to assign one before editing anything.

"The failed-payment notice carries more weight than the rest. Let's treat that wording separately."

Arguing for an order of work rather than fixing every touchpoint with equal force.

Pagedeck's first count came to 24. Eleven of them had no owner, and six contained wording nobody on the team had read recently. Once Noa collected the copy in one place, it turned out the same situation was being called three different things in three touchpoints. Sofia confirmed that four of the most frequent support questions all came from touchpoints where an explanation was missing entirely. The team picked the four places where impressions get decided, worked on those, and left the rest on the list with owners assigned and nothing else changed.

The most common mistake is using the word as a synonym for product screens. Screens are what a product team can change, so the scope narrows on its own. But customers don't file the screen, the text message, and the support reply in separate memories. Narrow the scope to screens and a dozen of those 24 places never make the list at all, and the drop-off happening there stays an unexplained number in a report.

The other mistake is treating all of them as things to improve. Once the list exists, every blank cell looks like a defect. But the 24 don't weigh the same. The first five minutes after signup and the arrival of the first invoice decide most of the impression; the other twenty only need to avoid contradicting it. Try to fix everything and four months later you've polished twenty-four places a little and the four decisive ones are exactly as they were.