Almost every CRM strategy I look at is measured on what it sold, which is reasonable enough until you notice that the measurement is quietly shaping the work. If revenue per send is the only number anybody reports, then every email has to earn its place by asking for something, and a programme where every email asks for something is a programme nobody looks forward to.

Which brings me to my fridge freezer.

I bought an American fridge freezer from AO.com, the first time I had ever given them my details, and what happened over the following fortnight was the best piece of CRM I have been on the receiving end of in a long time. Not one of those messages tried to sell me anything else. Not one of them was a campaign. And by the end of it I had an association with that brand so positive that I am, slightly ridiculously, looking forward to whatever they send me next.

So the argument of this post is simple, and it runs against how most programmes are built and judged. Email is not there to sell. Email is there to create experiences that build positive associations, and to increase the chance of somebody taking an action, whether that action happens inside the email or somewhere else entirely.

The shift

A campaign asks what we want somebody to do this week.

An experience asks what this person is going through right now, what they are worried about, and what they need from us to feel looked after.

Only one of those two builds something you can sell into later.

The argument

Action inside the email is the smallest thing you are trying to cause

Nearly everything AO sent me was designed to make me do something that had nothing to do with clicking. Be in on Tuesday. Clear a path through the kitchen. Check the appliance before the driver leaves. Keep the invoice somewhere you can find it. Know who to ring if the thing breaks in eight months.

None of that shows up in a click-through rate, all of it decided whether the delivery went well, and whether the delivery went well decided what I think of the company. Which is worth sitting with if your programme is currently being graded on in-email engagement, because the most valuable things your emails cause may be completely invisible to your reporting.

Three things an experience-led programme is producing, in rough order of how little anybody measures them:

• Action outside the email. Turning up, preparing, checking, keeping, using, renewing. The practical behaviour that makes the thing you sold work properly.

• Reduced anxiety. Somebody who knows what happens next does not need to contact you, does not stew, and does not arrive at the delivery already annoyed.

• Association. What your name comes to mean. Built slowly, from the accumulated feeling of every message, and almost impossible to repair once it has settled in the wrong direction.

The example

What happened after I bought a fridge freezer

Worth saying up front that I had done the research, found the appliance I wanted at a good price, and bought it. A completely ordinary transaction, and the sort of purchase where most retailers send you a receipt and go quiet until they can start marketing at you.

The order confirmation that was not a receipt

The first email was a thank you for the order, and it did not simply restate what I had bought and what it cost. It communicated what was going to happen next and what to expect, which is a completely different job.

The structure of it did a lot of work. We are getting your fridge freezer ready now. To make sure everything runs smoothly, we need you to do the following things. We are here to help, and if this does not look like what you expected or ordered, do this.

Underneath that, the delivery detail: what was coming, the date, the address. On the morning of the delivery they would text me between certain hours with the time slot. And a button to add it to my calendar so I would not forget, which is a tiny piece of design that quietly does somebody's admin for them.

The dynamic block that knew what I had bought

AO sell every kind of appliance, so the email carried a dynamic block keyed to the product. Because I had ordered an American fridge freezer, it told me the delivery team would unpack it on arrival and allow me time to inspect it, and asked me to make sure I was happy with the product before accepting the delivery, because claims for damage could not be accepted afterwards.

Read that again from their side. It is a warning, phrased as a courtesy, that protects them commercially and protects me practically at the same time. And it only makes sense because the content changed based on what I bought, which is personalisation doing an actual job rather than sticking my first name on something.

The bit that made me laugh, and then made me comply

Then a section headed with the important stuff, opening with a line I loved: we will do everything we can to make your delivery a success, but we need your help.

Look out for email updates. Check your order. Read through your delivery checklist. Followed by a proper delivery checklist, clearly laid out, then really important information, then help and advice, then a few reasons to be an AO member which was the only cross-promotion in the whole thing.

A massive transactional email, very long by any normal standard, and it worked because the structure and the design carried it. Length is not the problem people think it is when every section answers a question the reader was about to have.

Then the other channels started

An instant text message, straight after the order. Then, two days before delivery, a phone call, which is where it got interesting.

The call was mainly a check in. All okay, here is what to expect, let me confirm the details. They talked me through the process, including that the driver would ring if delivery was going to be early and that I would know the night before. And then they tried to sell me insurance, and it did not feel like an upsell at all, because it arrived inside a conversation that was already about making sure my purchase went well.

Worth pausing on that, because most businesses would kill for an upsell that lands as care rather than as a pitch. What made it work was everything that came before it in the same call. The sell was not the reason for the contact, it was a thing that happened during a contact that had its own reason to exist.

The day before, and the day itself

A text message the day before. An email too, still on for tomorrow, your order is on its way, see you soon, along with what was being delivered, when, what was going to happen, and how to change the delivery if I needed to. All of the important information from the first email, repeated at the moment it mattered rather than assumed to have been remembered.

On the day, a text message with the time slot, a phone call from the driver, and an email confirming it was on its way, what my slot was, what I needed to do, what was being delivered, and a reminder about the delivery checklist.

And then, the part most brands never build

After delivery, an email of important information about the order. The invoice to keep somewhere handy. The number for Beko, the manufacturer of the fridge freezer, so I would not have to go looking for it in a year. What to do if something goes wrong. How to register the warranty, and how to extend it.

Then the loveliest touch of the lot. Did we make you smile? How was your shopping experience? One to five, five being amazing, and a text message version too. All I had to do was tap five, which took under a second, which is presumably why they got an answer out of me at all.

Why it worked

Removing uncertainty, over and over

Strip the whole sequence back and one thing is happening at every single touchpoint. They are removing uncertainty, before I had a chance to feel it.

Buying a large appliance is a small anxious project. It costs a lot, it has to physically get into your house, something might be damaged, you might not be in, you might not know what to do if it goes wrong, and there is a two week gap between paying and knowing whether any of it worked out. Every message in that sequence closed one of those loops.

Which is why I felt informed at every step. Not marketed to, not nurtured, not engaged. Informed, and looked after, which are the two feelings that make somebody want to come back.

They had clearly done the thinking-feeling-doing-saying work

You cannot build a sequence like that out of a content calendar, because a calendar starts from what the business wants to say and when it has capacity to say it. Everything AO sent me started from the other end, meaning what a person buying a fridge is thinking, feeling, doing and saying at each stage.

Run that properly across before, during and after a delivery and the touchpoints fall out of it almost automatically:

• Before. Did it go through? Is it definitely the right model? When is it coming? Will they take the old one? Do I need to be in? What if I am not?

• During. Where are they? Will it fit through the door? Who unpacks it? What if it is damaged? Am I allowed to check before I sign?

• After. Did I keep the invoice? Who do I call, them or the manufacturer? What does the warranty cover? What if it stops working next winter?

Every one of those is an email or a text, and none of them is a campaign. Which is the whole method, and it is written up properly in the TFDS post if you want to run it against your own audience.

Email on its own would not have done it

The point I would most want an email person to take from this. Email carried the detail, because detail needs room and needs to be findable again later. SMS carried immediacy, because a time slot on the morning of delivery is useless sitting in an inbox. The phone call carried reassurance, because a human voice two days before a large delivery does something no automated message can.

Same information, three channels, each doing the job it is best suited to. Had all of it arrived by email it would still have been decent, and it would not have been anywhere near as good. Had it all arrived by text it would have been unusable. The orchestration is the product.

Every message earned the next one

By the time that feedback request arrived, I had received maybe eight or nine messages, and every one had been useful. So the ask cost me nothing, which is why I answered it, and it arrived at the exact moment the experience peaked rather than a fortnight later when the feeling had faded.

Compare that with the version most businesses run, where the first eight messages are promotional and the ninth asks for a review. Same request, completely different reception, and the difference is not the wording of the ask, it is everything that came before it.

What to steal

How to build this in your own programme

None of what follows requires a bigger team or a better platform, and most of it requires you to stop doing something rather than start doing something.

• Map the questions, not the calendar. Pick one journey, run TFDS across before, during and after, and let the touchpoints come out of what people are worried about rather than what you have to announce.

• Treat your transactional emails as the main event. They get opened at rates your campaigns will never see, they arrive when the customer is paying attention, and in most businesses they were written once by somebody in engineering and never looked at again.

• Repeat the important information rather than assuming it landed. AO told me the delivery details at least four times across the fortnight, and it never once felt repetitive, because each time it arrived at the moment it was becoming relevant again.

• Use dynamic content to answer a real question. The American fridge freezer block existed because that product has different delivery implications. Personalisation that changes what somebody needs to know is worth ten times personalisation that changes what somebody is called.

• Exclude people from marketing while they are inside an experience. Nothing in this whole story matters as much as the campaigns I did not receive. If you build the sequence and then let your promotional calendar run straight through the middle of it, you have built nothing.

• Put the ask where the goodwill is. Feedback, reviews, referrals and cross-sells belong at the point the experience peaks, and they should cost the person as close to nothing as you can make them. One tap, five options, done.

• Let a sell ride inside a contact that has its own reason to exist. The insurance conversation worked because it was not the reason for the call. An ask attached to something useful is a different object from an ask on its own.

• Give them what they will need later, before they need it. The manufacturer's number, the invoice, the warranty process. Cheap to send, memorable to receive, and the reason somebody thinks of you warmly in a year rather than at all.

The conclusion

AO did not sell me anything after the fridge. They spent a fortnight making sure the thing I had already bought went well, and the result is that I have a positive association with a company I had never used before, I feel informed, I know where to find everything I might need, and I am, weirdly, looking forward to hearing from them.

Whatever they send me first is going to land on all of that. It will be opened, it will be given the benefit of the doubt, and it will convert at a rate no subject line test could ever produce, because the work that makes it convert was done weeks earlier by emails that never asked me for a thing.

So the question worth taking back to your own programme is not what should we send this month. It is what is this person going through, what are they uncertain about, and what would make them feel looked after. Answer that consistently and the revenue arrives as a consequence, which is the only reliable way it ever arrives.