AI's real value is not generating content. It is giving time back to marketers.

Product marketing means different things in different companies, but in tech it sits in the centre of a Venn diagram between sales, field marketing, and product. The remit is unifying the messaging behind the products, articulating that to customers and prospects, and ensuring sales teams understand the message so they can have the right conversation with the customer.

At Salesforce there is a huge number of sales people, so the discipline is making sure all of those people are supported in the right way: marketing programmes, field events, ABM, and the other tactics that generate demand. Product marketing is in the middle of all of that, making sure the pieces line up.

No AE left behind. The challenge is making sure every account executive gets covered, programmatically and at scale, with the resources they need to help their customers with their business challenges.

Generative AI is, for marketers, both the number one opportunity and the number one challenge. Every business leader, whether CEO, CIO, or CMO, has a concern about it. The opportunity is obvious. The challenge is that companies sit on significant data they cannot put into a large language model without confidentiality risk. Most of that data is in slides, Slack channels, and the CRM.

The framework is real. Retrieve the right data. Put it into a prompt. Pass that prompt to a large language model where the data is not retained. Audit the output, including for toxicity. None of that should require turning every marketer into a prompt engineer, just as we did not turn every marketer into a SQL expert.

Our customers want to use unified customer data to build audience segments. Build me an audience segment of people who bought my product in the last 14 days but have not activated their credit card. Anyone in the room can describe that audience in a sentence. Building it in the standard toolchain, even for a skilled operator, takes time. With our data cloud you can type the prompt in. It does the work in about 15 seconds.

It is not doing anything a skilled operator could not do. It is doing it dramatically faster, and the operator still has the edit button. They can iterate the prompt, change the Boolean logic, or do it the old-fashioned way.

If you have a thousand audience segments but only two creative messages, you only have two segments. So the iteration on the message matters as much as the iteration on the segment.

The personalisation work has to use all of the data we have on a customer, beyond the marketing data. If a customer has an open service ticket, or is in the middle of an insurance claim, maybe I should not be trying to upsell them on their home insurance. They have got enough on. Marketers have had to make do with marketing data for too long. The single view of the customer has been lusted after for years. It is now here. But it is not the end. It is the means to activate the right message at the right time.

A real example from two weeks ago: I was looking at home insurance. This morning, two weeks later, I got an email quote from an insurer. Yes, it was very personalised. No, it did not come at the right time.

Stripped back: if there is someone in a company thinking what are the technology companies that solve this kind of problem? we want Salesforce on the list, ideally the first name in. Unaided brand awareness is important for any company, no matter the size, when relevance is held.

Brand campaigns also articulate how you come at a problem. We talk about trust as the guiding principle for generative AI. We have a brand spokesman, Matthew McConaughey, doing creative work that does not go into the technical detail of how the technology operates, but does articulate what we think is important and what we think is important to customers, with a dose of humour.

The ability to put yourself in someone else's shoes is the most important quality, whether that is a potential customer's or a salesperson's. Then the ability to absorb information and enjoy doing it. Generative AI is a good illustration. The role requires you to absorb where it fits in the world quickly. Then the ability to spit it back out in really simple, accessible language.

Product marketers in tech are the worst people for writing five-syllable words in fifty-word sentences. It sounds clever. It is also unintelligible to the reader. The reader, whoever they are, does not have time. They want to understand quickly. Do not make it hard for them.

Be hungry to learn. Lean into the opportunities. The first thing I worked on was producing CD-ROMs. Most people listening to this will not have a computer that takes one. A few years later I was working in search engine optimisation, which in 2002 nobody knew what that was. Both turned out to matter.

Today's exciting thing may, in twenty years, be the same kind of curio I am now telling you CD-ROMs were. So do not get lost in the new-fangled thing. Keep sight of what you, as a marketer, are trying to do: help your company articulate its value to a customer in a way that is relevant and useful to them.