What if stopping all your marketing changed nothing? That is where to start.

The biggest fundamental change I made when I took over was test everything and start with a blank slate. I sat all 120 marketers down and said I had a hunch that if we stopped doing all of our marketing nothing would change. The same pipeline. The same sales. Some people were stunned or insulted. For most it was a freeing thought. We wiped the slate clean and gave ourselves permission to experiment.

Nothing was off the table, but we would measure everything and be ruthless about cost per MQL and cost per op. That moved the mindset toward data-driven accountability, but more importantly it unlocked creativity, because anyone could come to me with a webinar idea, a podcast, a new series, and the answer was always let's try it. If it beats the cost-per-op target, we will do it forever and pour money in. If it does not, it is not a failure, it is a learning.

We started with demand because the data on demand is the cleanest. Cost per MQL and cost per op for every channel, every programme, every campaign, every webinar, every event. We knew the bookings target at the start of the year, the pipeline needed to hit it, the conversion rates, and the budget to play with. From that we derived a target cost per op, and every activity was measured against it.

The culture that came out of that was: anyone can propose anything, you just have to beat the target. Where things did not work, we learned and moved on.

Webinars. They worked during COVID because people were home and bored. We kept doing them out of marketing lethargy, the assumption that everyone needs to be doing them. When we looked at the effort against the audience we were genuinely reaching, the numbers were terrible. The ROI model often missed it because the webinars were notionally free, in-house. Nothing is free.

So we asked the inverse question: who already has the audience we want, and how do we use theirs instead of begging people to show up to ours? We were trying to grow our CMP business and reach content marketers, and the Content Marketing Institute is both a huge event and a publication. Instead of running our own webinar to thirty people, we used their audience to reach three hundred or five hundred.

We had a lot of smaller events that were not working. We said: fewer events, but go much bigger. When Optimizely shows up at an event, we show up with a big booth, a big stage, a big presence. Everyone leaves knowing who we are.

We measure the pipeline, but the brand impact is genuine and should not be underestimated. The lines between brand and demand are blurry. About half our pipeline is inbound. People fill in a contact form, but rarely because they searched a lower-funnel pain term in isolation. They had a longer relationship with the brand, then filled in the form.

Optimizely is a testing company, so we have testing in the DNA. People think A/B testing applies mostly to a website button or a landing page. You can test everything about content, beyond just headlines: tone, language, themes, length, format. We test before we create.

We do painted door tests, where the headline goes up without the underlying piece. If enough people click, we know it is worth investing in. It is annoying for the ten or twenty people who clicked through to nothing, but it saves tens of thousands of dollars on content that would not have earned the spend. The content that does get made is proven by science to work, and then we put real effort behind those big rock pieces.

When somebody arrives at the site, depending on their industry, persona, or what content they have consumed before, we personalise the path. Not one-to-one individualisation. The effort is not worth the ROI. Personalisation at the persona or industry level. And we extend the same persona, theme, and language everywhere, beyond digital alone. The out-of-home campaign, the booth at an event, the content marketing all show up consistently. The consistency is part of how the brand compounds.

Content has its intrinsic engagement metrics, and I weight engagement over page views or uniques. But for the business itself, I goaled the entire marketing organisation on the goals of the business, which supersedes any marketing KPI. Our number one goal is sales. Our number two goal is all-company pipeline.

It used to be marketing-generated pipeline. The realisation was that Optimizely generates pipe through five paths: marketing, sales, customer success, SDRs, and partners. Marketing touches every one of them. We were spending too much time on attribution and credit, and we scrapped all of it. Marketing is accountable for all pipeline regardless of the origin path.

The whole thing was a test. We designed everything in-house, bought direct through AdQuick, and used a tech platform to put creative onto New York City taxicabs. We could change it every day if we needed to. Most companies would consider that channel too expensive, too slow, or too committing. The testing mindset took us in cheaper.

We are our own number one user, and probably a pain to our product team because we have a lot of feedback and high expectations. When I wanted personalisation, I told them my use case as a marketer, not as someone who works at Optimizely. I said I need to personalise based on customer data sitting in Salesforce. They told me it was hard, the connector did not exist. I told them it was a massive pain point and we had to solve it.

For any marketing team at a B2B software company, especially in martech, using your own product aggressively should not be a privilege. It should be an expectation.

We use generative AI to draft campaign ideas, taglines, briefs, blog posts, decks, and imagery, always as first drafts. We would never publish AI-generated content, but the SDR outbound emails are researched and written by AI. We are already doing a great deal of it. It is the tip of the iceberg.

AI is not going to replace people or jobs, but a person who does not use AI may be replaced by a person who knows how to use AI. As marketers, we should always be pushing the envelope.