Table of Contents

• 1. Product: the conversation is now part of the product
• 2. Price: AI made comparison instant
• 3. Place: there's a new channel and you're not measuring it

  • The visible place: chat as the storefront
  • The invisible place: being inside the AI's answer
• 4. Promotion: from shouting at a thousand to talking to one
• The other 3 Ps: people, processes, and presence
• How to start without burning cash
• The three mistakes I see the most
• To close

1. Product: the conversation is now part of the product

The product used to end at the box. Today, if you sell something digital or a service, the experience of asking, quoting, buying, and claiming is as much a part of the product as the product itself.

A badly built chatbot isn't "a channel that failed". It's your product feeling cheap.

What changes in practice:

• Onboarding went conversational. The bot that explains how to use what you just bought reduces returns and support tickets. That's product, not support.
• The bot tells you what your product is missing. Every question the agent can't answer is a gap in your offer or your communication. Review unresolved conversations once a week: it's the cheapest market research you'll ever get.
• Response speed is a feature. Answering in 5 seconds at 11 p.m. goes head to head with having a better price.

An uncomfortable question to ask yourself: if your competitor answers instantly and you answer tomorrow at 9 a.m., are you really selling the same product?

2. Price: AI made comparison instant

When someone asks an AI model "compare these three options for me", your pricing strategy stands naked in three seconds. You can no longer hide the price behind a form and hope the customer gets tired of searching.

Three concrete moves:

Publish the price or publish the range. If your price isn't anywhere, AIs can't recommend you and the customer assumes it's expensive. An honest range ("from $X depending on number of users") beats silence.

Justify the price in text, not in design. A language model doesn't read your pretty infographic. It reads text. If the argument for why you're worth more lives in an image, as far as the AI is concerned it doesn't exist.

Let the bot quote, but with rules. A conversational agent can put together a quote in the chat. It can also give away your margin if you don't set limits. Define in the prompt what discounts it can offer, up to where, and when it must hand the conversation to a human. Guardrails aren't bureaucracy: they're your commercial policy in writing.

3. Place: there's a new channel and you're not measuring it

This is the big change. Place stopped being just "where they buy" and now includes where they ask.

You now have two new places:

The visible place: chat as the storefront

WhatsApp, Instagram, the widget on your website. The conversation became the counter. This is especially true in Latin America: a lot of people don't want a shopping cart, they want to message someone. A well-configured agent handles, qualifies, charges, and books while nobody is awake.

The invisible place: being inside the AI's answer

This is the one almost nobody is working on. When someone asks ChatGPT, Gemini, or Claude for a recommendation in your category, you either show up in the answer or you don't exist. This is called GEO or AEO (generative or answer engine optimization), and in practice it means:

• Content that answers concrete questions, with the answer explicit in the text and not implied.
• Structured data (schema) on your site: product, price, FAQ, organization.
• Mentions in sources the models consult: directories, comparisons, forums, industry media.
• Data consistency. If your name, your category, and your price say different things in five places, the model would rather not cite you.

Classic SEO puts you in a list of ten blue links. This puts you inside the only answer the user is going to read.

4. Promotion: from shouting at a thousand to talking to one

Mass promotion still works for capturing attention. But conversion no longer happens on the ad: it happens in the conversation that follows.

The real change is that you can now have thousands of one-to-one conversations without hiring thousands of people. That brings power, and it brings the temptation to ruin it:

• Make the ad land in the chat. A "click to WhatsApp" ad that opens a conversation with context converts far better than one that sends people to a cold landing page.
• Segment with what the bot already knows. If someone asked about the annual plan and didn't buy, that's a different message from the one you send someone who never asked anything.
• Respect the opt-in as if your account depended on it, because it does. Conversational channels punish spam by blocking your number, not by lowering your reach. The cost of getting this wrong is losing the entire channel.
• Don't pretend to be human. Have the bot say it's an assistant. People accept it without a problem if it solves their issue; what they don't forgive is finding out the "personal" conversation was a script.

The other 3 Ps: people, processes, and presence

When services marketing expanded the model to 7 Ps, it added people, processes, and physical presence. A chatbot touches all three at once:

• People: the bot is one more member of the team facing the customer, and it inherits your tone. If your brand is warm and the bot talks like an instruction manual, you broke the brand.
• Processes: everything the bot automates has to exist first as a clear process. Automating chaos only gives you faster chaos.
• Presence: the chat is your storefront today. Its design, its speed, and its clarity are your facade.

How to start without burning cash

If you're thinking about building this, here's the order that works:

1. Collect the 30 questions people ask you the most. Pull them from your actual WhatsApp, not from your imagination. That's 80% of the work.
2. Define the bot's single goal. Booking, quoting, answering questions, or selling. One. A bot that does everything does nothing well.
3. Write the prompt as if you were training a new employee: personality, business context, tone, goal, limits, and who it escalates to.
4. Give it a human escalation plan. The bot must know how to say "let me get someone for you" and do it well. That's the difference between an assistant and a trap.
5. Measure what matters: conversations resolved without a human, time to first response, conversations that end in a sale. Don't look at message count: that's not a metric, it's noise.

The three mistakes I see the most

• Building the bot before the sales process is clear. The bot amplifies what you already have, good or bad.
• Optimizing only for Google and ignoring that a growing share of people no longer search — they ask.
• Treating the conversation as a mass-blast channel. It burns fast and it doesn't recover.

To close

The 4 Ps didn't get old. What got old is assuming the customer will walk up to your shop window, read your catalog, and fill out your form. Today the customer asks an AI, and then messages another one.

If your product can't be explained in a conversation, if your price can't be found, if your place doesn't include the chat, and if your promotion doesn't end in a message, the mix is incomplete. And an incomplete mix doesn't show up in the strategy: it shows up in the sale that never came.

---

If you want to build this in your business with a method instead of trial and error, at leotransforma.co I work through these topics step by step with entrepreneurs who are building seriously.