Tuesday, August 25, 2026  ·  7 Min Read

WELCOME BACK!

In my previous issue I called out a failed test costs time and money that we can’t afford with bootstrapped and lean teams. On the flip side, here’s what happened when I removed the pressure of failing.

Last issue, I talked about the idea of CEOs doubting if marketers truly add to revenue or move deals in your sales pipeline and that chasing attribution is putting your hope in something they realistically do not care about.

What actually wins is the evidence you bring to the table, rather than evidence you traced. Which means testing things should be fundamentally ingrained in your go-to-market teams. Especially in marketing.

Almost nobody does, and it’s not only slowly killing momentum but also creativity.

What six tests in eight weeks actually taught me.

We ran an eight week sprint that pushed our limits. The rule was simple and slightly uncomfortable: bring creative ideas to test, rather than polish the things we already knew how to do.

That is way harder than it sounds. Trust me! Day to day in marketing you live inside a fairly narrow loop. A loop that should get questioned but we don’t. Write the copy, do the design, read the stats, plan the integrated plan. All of it is necessary but also all of it is well inside our comfort zone. The sprint existed to force us out of that loop, and to let anyone put any strange idea on the table as long as it laddered to revenue.

We ran about six tests across various regions and here are three of them to give you an idea.

The one that failed. On Linkedin ads, we tried a static image creative instead of a carousel, built to work like a glimpse into the PDF rather than a summary of it. The theory was that a partial view creates more pull than a complete one. The result we obtained after much optimization was that it did not work. Cost per lead went from $105 to $125 and we got no conversions at all.

The one that surprised me. Talking head video! Instagram and TikTok have been showing that straight to camera content holds retention, and I wanted to know whether that carried into B2B or whether it was a consumer only effect. We tried this on both organic and paid media. This was by far the most annoying test to run, because it depended on several people actually recording themselves and sending back something usable. We ultimately worked with what we got which led to an increase in click through rates by 10%.

The one that expired. Reddit, to build a presence and to see whether we could show up in AI answers. At the time ChatGPT was leaning heavily on Reddit, so it was worth a slot. We ran a quick search check at the end of the eight weeks and our brand name came up. A small result on a small KPI, because search is a long game and anything else would have been a lie. Take this with a grain of salt because ChatGPT no longer gives Reddit the same callouts like it used to. In a recent article on Promptwatch, called out an 86.4% drop in Reddit citations on ChatGPT search, primarily from mid Aug 2026. Gaming LLMs isn’t going to give your brand the search position anymore. More focus needs to be put into deep topical authority, strategic backlinks & Digital PR and demonstrating First-Hand Expertise.

Now the part I did not expect. The pushback did not come from leadership. It came from my junior team members, who said this was too much on top of their existing workload. Fair enough that they said it during the busiest season of the year. Why I did not expect this response though was because I expected all my team members to want to test and learn at the same level as I did.

Founders deploying AI agents across sales, marketing, and customer service are closing more with smaller teams. 65% more sales leads. Less headcount.

Download the free Practical Guide to Agentic GTM for Startups and start building the stack your competitors dream of.

Get the Free Playbook

The objection was not resistance but it was an accurate description of how the team is built.

I want to sit with that pushback, because my first instinct was to read it as my team was simply being difficult and reluctant. But I now think that was wrong.

They were describing our operating model correctly.

Look at how a marketing team is actually resourced. Every hour is pre committed to producing something finished; the campaign launches, the copy goes out, the report gets sent. Nowhere in anyone's week is there a line that says find out whether we are wrong about something.

So learning is not something you find time for. It is something you take time for, out of work that already has a name, an owner and a deadline.

That is why testing culture dies inside teams that genuinely believe in it. It’s not about conviction but purely about capacity.

If everything had worked, the tests were too safe.

The static image test was a clean failure. The cost per lead went up and zero conversions which meant our theory was wrong.

I would run it again. Just not the same test, obviously.

But a sprint that returns six wins has not told you anything, because you only ran things you already expected to work. The failure is the part that proves the exercise was real, and it costs much less than running a quarter long campaign. A couple of weeks and a creative concept vs a wrong assumption we would otherwise have carried into a full campaign.

Let’s take a moment on the one that won. The talking head test was the most operationally irritating thing on the list, the one that needed people to film themselves and send it back. All while managing different personalities and timezones. That is usually the exact reason an idea dies at the planning stage. Too much hassle! We ran it anyway and it moved click through by 10%.

The annoying test and the failed test both earned their spot on the list. If it wasn’t wrapped in a sprint concept, I may have said no, let’s postpone it.

The mechanic that made six tests survivable in eight weeks.

The thing that made this work, and I would not run it again without it, is that no test was measured against revenue. Yes, we all knew that was the main goal and all tests did have that mindset but we told ourselves, let’s try concepts with smaller KPIs and if we can move those KPIs by a little bit, let’s double down on it.

Each test carried a KPI drawn either from a tactic we had run before and could compare against, or from the nearest measurable thing if it was genuinely new. That sounds like a downgrade but it is the exact opposite.

Measure a two week test against pipeline and you guarantee an inconclusive result, because pipeline moves too slowly and too many other things touch it. You end the sprint with six tests and six shrugs.

The Reddit test is the clearest example. The honest KPI was not leads and it was not pipeline. It was whether our brand name appeared at all in an AI answer for our category by the end of the eight weeks. That is a small question.

The counterargument I would raise if I were reading this: smaller KPIs are exactly how marketing ended up with a decade of metrics nobody in finance respects. That is fair. The difference is what you do with the answer. A vanity metric is one you report. A test KPI is one you act on and then throw away.