A workable audit answers five questions: What is the marketing function trying to produce more of (pipeline, content, qualified leads, retention)? What is the team's current AI fluency? What systems are already wired together? Who is the audience, and how comfortable are they with AI-generated material? What is the constraint that is keeping the team from moving faster today?

The output is a ranked opportunity map. Three to five moves, in order. Each move has an effort estimate, an expected impact, and a one-line rationale. The first move is typically the lowest-friction lift with the highest payback — usually a workflow inside the team's existing stack that AI compresses from days to hours. The last move is usually the most ambitious — an automation or a content system that requires real architectural work and a longer time horizon.

A useful audit also returns a portable prompt. The prompt is structured so it can be pasted into Claude, ChatGPT, or any other assistant the team already uses, and continue producing decisions inside the team's working context. The audit ends; the prompt persists.

The free version of this audit, built by Eduardo de la Espriella for B2B marketing leaders, lives at /audit/. The paid version with a working session and a prompt library is the AI Marketing Sprint.

The practice splits the difference. It leads with the operational pain that the technical product removes. Then it uses the technology as proof rather than as headline. The structure is consistent.

First, name the operational problem in the language the operator uses. "Customers abandoning checkout on mobile." "Engineers paged about an outage they cannot trace." "Sales and finance arguing about the same quarter's numbers." Second, name the consequence in the language the buyer's CFO uses. "Cart abandonment costing X percent of revenue." "SLA penalties last quarter." "Decisions delayed a week each time we re-run the report." Third, name the technical mechanism that removes the consequence. "Adaptive payment routing." "End-to-end distributed tracing." "A single governed source of truth, column by column."

Stripe is the textbook example. The homepage reads Financial infrastructure to grow your revenue — not we process card payments. The CFO buyer reads grow-your-revenue; the CTO reads the change-log on adaptive 3DS routing and idempotency keys; both close the tab convinced the product is for them. The same pattern shows up at Datadog (See your full stack in one place), Linear (The system for modern software development), and Vercel (Develop. Preview. Ship.) — each leading with the operator's day, not the underlying stack.

The result is positioning a non-technical buyer can repeat back to their board, with the technical credibility their CTO will not push back on. The buyer reads the headline; the technical decision-maker reads the proof; both buy.

Companies that get this right share a habit: they describe operations more carefully than they describe their stack. Companies that get this wrong describe their stack more carefully than the operations the stack changes.

This is the heart of the practice at eddieespriella.com.

The slower brand: what happens when you stop shouting

There is another way to be seen. It is quieter. It takes longer. It lasts.

Written by Eduardo de la Espriella

Growth marketing has meant, for the last ten years, a particular kind of volume. Hooks. Drops. Seven-slide carousels. Outrage bait. The schedule is so packed that a week off the feed feels like a month off the calendar. The idea is that attention adds up. So more attention adds up faster.

It does not, really. Most of the brands I come back to are doing something else. They publish less. They repeat themselves. They do not chase the algorithm. They set a rhythm and the audience learns it. They show up on Tuesday because they show up on Tuesday.

A slower brand is not quiet to the point of invisibility. That is a different problem. It just picks its moments, and trusts that the right reader does not need a megaphone. Just a signal, in the same place, every time.

A hospitality brand I worked with at Story Jar had been posting three times a day across five channels. Hooks, drops, the whole playbook. Nobody on the team could say what any of it was doing, but stopping felt like surrender. We stopped. We moved to a weekly post, and a short monthly note to people who had already stayed with them. For the first few months, nothing moved. Then repeat stays began to creep up. The posts that had been meant to "drive awareness" had mostly been driving nobody. The awareness they needed was already in somebody's inbox.

Part of the case for slowing down is honest: fast marketing burns out the people making it. Constant posting forces shallow thinking. You publish before you have finished thinking, because the calendar is in charge. A slower brand has time to be careful. Three things a quarter that actually matter, instead of thirty that do not.

The other part is about positioning. If you speak once a month, every sentence is a promise. If you speak every day, the promises blur. Posts written by the same team six weeks apart start to contradict each other, because nobody had time to remember what was said before. Brands that publish constantly tend, over time, to mean less. Not because any single piece is bad. The shape just gets fuzzy.

Whether it is payroll software, a boutique hotel group, or a neighbourhood bakery, no one buys something serious because they saw a funny LinkedIn post. They buy because six months ago they ran into a problem, and on a Tuesday afternoon they remembered someone had written something clear about it. They go back. They search. They find you.

That is the only attention that matters. The kind that arrives late. The kind you cannot really measure on a weekly dashboard. The kind you earn by being there, year after year, saying a small number of true things more than once.

Slower brands also age better. The writing ages slower because it was not built to win this week's argument. The visual system ages slower because it was not designed to win one campaign. The voice ages slower because it was never trying to sound like whatever was working on the feed in March.

There is a simple test for whether a brand is fast or slow. Pull up what you published a year ago. Read it out loud. If most of it makes you cringe and you would never write it now, you are fast. If most of it still feels true, you are slow. Neither is a virtue on its own. But only one of them turns into something that outlasts a team.

The hardest part of working this way is the quiet stretch. The first six months feel like nothing. Traffic is flat. Engagement is flat. The founder asks if marketing is broken. It is not. It is just that what you are building, a reputation for saying specific things clearly, does not move on a weekly dashboard. It moves on a two-year dashboard, and most companies do not have one of those.

This site is the same argument applied to itself. Three essays a quarter. A short letter once a month. No daily feed. No streak to protect. If it is working, you will know in a year.

Stop shouting. See what is left. If the answer is nothing, then the shouting was the brand. If there is a product, a point of view, a reason someone should pick you, that is what you were looking for. Build around that, slowly.

A short defense of the homepage that says one thing

The hero headline is a promise. A second CTA usually means you have not picked your reader yet.

Written by Eduardo de la Espriella

Most homepages try to say everything. Product features. Testimonials. Investor logos. Awards. A carousel. A chatbot. A cookie banner. A newsletter popup. A "watch the video" hero. A "schedule a demo" button. Another "schedule a demo" button, in case you missed the first.

The result is a page that says nothing. When everything on a page is given the same weight, nothing on it stands out.

The homepage that says one thing is harder to build. It needs a sentence the team actually agrees on, short enough to hold in one breath. Most teams do not have that sentence yet. The homepage is where that gap shows up.

Rule of thumb: if the hero headline does not make a promise, the rest of the page is decoration. A promise is specific. It names a buyer, an outcome, and often a number. "Payroll errors down ninety percent, processed in an afternoon." "Direct bookings up, without ads." "One fee, zero surprises." Those are promises. "Platform for the connected world" is a placeholder where a promise should go.

Second rule: if there are two CTAs above the fold, one of them is a hedge. "Watch demo" next to "Book a call" means you did not know which customer you were talking to, so you invited both. The reader picks up on that faster than you do.

The work here is subtraction. You have to take off the page most of what the team wants on it. You tell product the integrations grid can wait for page two. You tell sales the case study PDF does not belong in the hero. You tell the founder the logo wall is fine, but later. None of these conversations are fun. All of them are needed.

I've been through this clean-up with a law firm that had eighteen service lines on the home page, a webshop that opened with twelve categories and a chatbot, and a restaurant group fronting nine hero carousels. In each case, three conversations later, the page ended with one promise and one invitation. The rest went to sub-pages. Bounce rate did not tank. People stayed longer.

What you are left with is a page a stranger can read in fifteen seconds and walk away understanding one thing. That is the goal. Not engagement. Not bounce rate. Understanding.

One last test. Read the homepage out loud to someone who does not work at the company. Ask them, in their own words, what the product is and who it is for. If they can answer in one sentence, you have a homepage. If they can't, you have a brochure. And a brochure is a thing people close.

Related essay: Positioning is a subtraction problem →

Positioning is a subtraction problem

What you are really working out is everything you are not.

Written by Eduardo de la Espriella

Most positioning sessions I have sat through start by adding. Here is a whiteboard. What are we? Smart. Fast. AI powered. Enterprise grade. Trusted. Global. Innovative. By the time the markers run out the board says nothing, because it says everything. Someone takes a photo and the photo is never opened again.

Positioning is the opposite move. It is what you cross off the list.

The exercise I keep coming back to is simple, slow, and gives you ugly output. Write down every adjective you would honestly use about your product. Write it long. Do not edit. Then write down the three closest competitors. Not your fantasy competitors. Not Excel. The three companies that actually show up on a buyer's shortlist with you. Now, for each adjective on your list, ask a single question: would any of those three claim this word, too? If yes, cross it out.

What is left is your real position. Usually it is one or two words. Often the team is disappointed. "We are not going to put that on a billboard." Good. If the remaining word makes the team a little sad, it is probably the true one.

The exercise works because it forces the distinction to come from the market, not from the room. You can tell yourself you are "innovative" all day. So can everyone else. Subtraction takes out the words that are really about how the team wants to feel, and leaves the ones that are about how the product actually behaves.

The hardest part, in practice, is naming the competitors honestly. Teams like to pretend they compete with no one ("we are a category of one") or with everyone ("we compete with spreadsheets"). Neither is true. There are three to five companies a buyer is looking at next to you. Those are the ones on the whiteboard, with their adjective columns next to yours.

A worked example. A law firm I helped at Story Jar had eighteen different service lines: corporate, tax, IP, family, immigration, real estate, every flavor. Every page on their site opened with the same three words: trusted, experienced, responsive. So did every law firm's site on the first page of Google. Literally the same three adjectives. We did the subtraction. Everything the nearby firms could also say, we crossed out. What we were left with was not glamorous. It was the one thing none of the other firms were doing: they had published, on a fixed fee, a small menu of packaged legal products. Five figures. Here is what you get. Here is when.

Ugly. Not what a managing partner wants to say out loud. But the only thing, in that market, that was actually theirs. The homepage was rebuilt around those four packaged products. The rest of the service lines went to sub-pages. Six months later their inbound was up, and for the first time the conversations coming in were from people who already knew what they wanted to buy.

A second one, different industry. A heritage retailer I helped at Fausto Salazar had an adjective column stuffed with words the whole category used: trusted, family owned, local, curated, friendly. Every nearby competitor could say the same. What only they could say was that they had sold the same household categories, continuously, since before most of their customers' grandparents were born. Awkward to put on a billboard. But a real boundary, and a useful one for the way the site, the email program, and the physical store started to speak.

The same logic applies in deep tech. I ran this exercise with the marketing team at Outsight against our closest competitors, and the word that survived was unglamorous, technical, and only ours. Same shape of finding, different vertical. Subtraction is industry agnostic.

A few follow on notes.

Subtraction does not tell you how to talk about the remaining word. It tells you what is yours. Turning a technical sounding remainder into a sentence a buyer will actually nod at is a separate job. It usually goes better when someone who was not in the room the whole time helps you write it.

Subtraction also does not tell you the remaining word is valuable. It only tells you it is yours. Sometimes you finish and the one word left is something the market does not care about. That is an even more useful finding. It means you are not different where you thought you were, and the next step is product, not marketing.

Positioning by subtraction is slower than positioning by brainstorm. It gives you smaller, uglier outputs. But those outputs survive contact with the buyer, because they have already survived contact with the competition. And that is the only test that counts.

Related note: B2B technical positioning →