I started working in digital in 2012, specialized in online advertising, and moved through agency work, freelancing and building my own companies. In 2019, the work I was already doing on my own gave rise to the agency that now operates under the name SpartAds. Along this path I've billed over 5 million euros online through my own projects and directly supported more than 1,000 Portuguese companies through services and mentoring.
That experience changed how I evaluate business growth. A paid traffic campaign can't be judged in isolation from the landing page, the sales screening process, response times or the profitability of what's actually being delivered. It's entirely possible to lower the cost per click while at the same time worsening net profitability, if the sales team takes days to respond or if support costs eat into cash flow.
At SpartAds we work on acquisition and scale for service businesses and online stores. At SpartAds.ai and in the Laboratório da IA, I work with entrepreneurs and professionals who want to implement automation and technical architecture in their workflows. On Sextas Ímpares, the open broadcasts I keep running on YouTube, I share code demonstrations, internal dashboards and conversations about leadership. In this article I gather the decisions, metrics and boundaries I use to think about marketing, sales and operations.
The Offer: Clarity in the Promise and Definition of the Customer
The clarity of the value proposition determines the success of customer acquisition before a single euro goes into ads. The potential customer needs to recognize the solution to their pain point, understand the scope of what they're buying, and know what the next step is without ambiguity or empty promises.
In a service business, designing the offer means setting clear contractual boundaries: what's included, execution timelines, materials the client has to supply, and the customer profile you explicitly refuse to serve. An offer that's too open invites chaotic exploratory meetings that demand custom quotes from the first minute, and that wears the team down.
In e-commerce, evaluation comes down to catalog competitiveness, stock availability, shipping costs and checkout friction. The ad has to accurately reflect what the buyer finds in the store. When the ad promises one condition and the page doesn't confirm it immediately, you're paying to generate friction and rejection.
The customer isn't looking for exaggerated promises or elaborate language. Build pages that answer, without detours, the questions the sales team already hears every day: timelines, support format, base prices and cancellation terms. When the offer does this filtering upfront, sales conversations become more objective and less like endless consultations.
To move from the offer to campaign work, explore how I apply AI in digital marketing. The guide connects data, ad variations and message review with what the business can deliver.
Ask customers before preparing the next offer
To decide what to sell next, I start by listening to people who have already bought and still have an unresolved need. In my 2022 book, I describe research that changed what I planned to launch. I expected requests for more courses; the responses pointed towards closer support and led to a mentoring offer.
I expected topics such as e-commerce and affiliate marketing. Had I acted on that assumption alone, I would have produced more lessons without addressing the request that was emerging. Asking students helped me see that some already knew enough to get started but were struggling with decisions about their own projects.
Today you can use AI to organise open responses to a questionnaire and identify themes worth investigating. Keep each conclusion connected to the responses supporting it, and remove unnecessary personal data before processing them. If the model says “customers want ongoing support”, open the examples and check who asked for it, in what context and how often.
Then talk to some of those people. A request for help might mean a diagnostic session, a review of work or ongoing support, each involving a different commitment. That detail lets you design a clear offer and test whether people will buy it. Research gives you material for a decision; the purchase and delivery provide further learning.
This connection between listening, designing an offer and execution remains part of Laboratório da IA, where I review projects in two mentoring sessions each month. The example in the book concerns an earlier offer. The principle I carry forward is to listen before producing and ask again after delivering.
Building a Sales Funnel the Team Can Actually Work
An effective sales funnel organizes incoming contacts in a way that helps the team in the following steps, without creating unnecessary barriers. Long forms push away people who are already ready to decide, while forms with no qualification at all fill the processo comercial with contacts who have no budget or urgency.
In Sextas Ímpares class #132, on setting up a lead funnel for a service business, I talked about the connection between traffic channels, the capture form and routing into the sales processo comercial. The form should collect only what's essential for the first interaction: size of the problem, location, urgency or segment. Anything more turns the contact into a survey and drives away people who just want a quick answer.
As soon as a contact comes in, it needs an owner and needs to follow a flow with clear stages. A shared inbox might work in the first months of a small company, but once volume passes a few dozen requests a week, you lose track of response times. A contact left without a reply for two days loses much of the initial buying intent.
Design the processo comercial around concrete action stages: contact attempted, conversation held, proposal delivered, decision made. Avoid inventing states just to fill fields in the CRM. Each stage should tell the salesperson what to do next and show management where the business is losing opportunities. I develop this organization further in the article Como montar um funil de leads que a equipa comercial consegue trabalhar.
Qualifying and Following Up: The Context the Sales Team Needs
Lead conversion depends directly on the depth of context recorded at the first contact and on how quickly the team follows up. A phone number with no history, no identified pain point and no urgency reduces the salesperson to someone making cold, repetitive approaches.
In Sextas Ímpares class #144, with Beatriz Teixeira, we talked at length about the barriers that block conversion after a request comes in. If the salesperson forces the customer to repeat everything because there are no shared notes, the buyer immediately notices disorganization. Recording the reason for the purchase, the budget capacity and the decision timeline avoids meetings that go nowhere.
One thing I've learned not to do is mark a lead as unqualified just because there was a single call, during business hours, with no answer. Before blaming the ad or the agency, you need to confirm whether there was follow-up through different channels, whether the approach was careful, and whether the offer presented matched the original request.
When a proposal is sent, follow-up needs clear rules. A proposal can be accepted, rejected with a specific reason, put on hold due to budget, or simply go unanswered, and each of these scenarios calls for a different follow-up. I develop this system in the article Como transformar mais leads em clientes num negócio de serviços.
The sales automation with AI guide develops enquiry assignment, states and conditions for changing follow-up. It helps organise the process while preserving the context of each conversation.
How Much You Can Pay for a Lead in Your Business Model
The cost you can afford to pay per lead depends on the contribution margin of the product or service, fixed costs and the team's closing effectiveness. Deciding whether a cost per lead is high or low without cross-referencing margin and conversion rate leads to risky financial decisions.
In Sextas Ímpares class #124, on cost per lead, I explained the calculation for setting safe limits on advertising investment. The logic of the closing rate is explained in the segment at 9:50:
Reference CPL = advertising budget per customer × lead-to-customer conversion rate.
To clarify this calculation, imagine a hypothetical scenario: if margin allows 200 euros of advertising spend per new customer and the team converts 5% of leads, the reference CPL comes out to 10 euros. If the closing rate drops to 2%, the maximum ceiling drops to 4 euros per contact. The service and the margin stay the same; it's the closing team's capability that changes what you can afford to pay per contact.
Use your own historical numbers, group comparable periods, and factor in your business's collection timelines. Revenue expected in two or three months, or contracts paid on ninety-day terms, isn't money available today to fund campaigns. The survival of a growing operation depends on immediate liquidity, not on a projection on paper.
Test a specific question and keep what you learn
A test should answer a concrete question and leave information the team can use later. In the book, I called investment in traffic to learn about an offer, message or audience “buying data”. I meant funding experiments with defined limits, never a need to purchase databases of people’s contact details.
Before spending, write down the hypothesis, the outcome you will observe and the available budget. If you want to find out whether a promise is clear, a conversation with potential customers may be the first step. If you already have two page versions, you need comparable traffic and an evaluation criterion set before choosing a winner.
AI can help prepare variants and organise the report, but check the arithmetic. For example, at a cost per click of €0.20 and a conversion rate of 2%, you need an average of 50 clicks per conversion, costing €10. The relationship is cost per conversion = CPC ÷ conversion rate as a decimal. You still need to compare that amount with conversion quality and the business’s margin.
Record tests that are inconclusive or show no improvement too. Note the period, traffic source, what changed and what remains unclear. When someone proposes the same change again, they can consult what has already been tried. That documentation avoids paying repeatedly for the same lesson and gives useful context to people joining the team.
Where time and sales are lost
Diagnosing sales drops means examining every stage of the sales funnel before adjusting campaigns. Changing budget, audiences, or creatives without concrete data adds noise and obscures the real causes. Careful observation first helps identify true points of friction, ensuring later campaign changes address actual problems rather than guesses.
In Sextas Ímpares class #135, on drops in lead volume, I built a diagnostic table that I use to guide teams in identifying where the difficulties lie:
| What you observe in the funnel | What's worth checking |
|---|---|
| Lots of traffic, few requests | Whether the ad matches the page, price and offer clarity, mobile load time and form complexity. |
| Requests come in, few meetings booked | Time to first contact, channel used, number of attempts, context handed to the salesperson. |
| There are meetings, few proposals sent | Qualification criteria, real identification of need, whether the service fits the financial capacity. |
| Many proposals, few sales | Clarity of terms, time to deliver the proposal, active follow-up, history of objections. |
| Sales grow, margin barely holds | Real acquisition cost, rework in delivery, post-sale support hours, contract retention. |
These points serve as a starting point for investigation, not as an automatic diagnosis. The numbers show you where to look, but they rarely explain the cause by themselves; for that you need to read sales conversations, review a sample of forms and talk to the people on the front line. Always test one change at a time so you can measure the real impact.
Margin, Leadership and the Founder's Bottlenecks
Growth in a service business often stalls when decisions concentrate too heavily around the founder. Centralizing quoting, sales screening and support overloads the organization's response capacity, limiting speed and consistency. This dependence on one person weakens margins and constrains scalability, since the whole structure waits on a single decision-maker.
In Sextas Ímpares class #147, on seven years of entrepreneurship, I talked about a phase where I took as long as seven days to respond to potential customers. I was the sole bottleneck for validating every proposal at the agency. Opportunities kept coming in, but my limited time meant I couldn't respond in time, and that cost revenue.
That phase taught me to create clear delegation rules. The team needs objective criteria to approve quotes without constantly asking permission, and needs to know exactly when to bring in management. If every sales question has to go through the founder, the company creates internal waiting lines that cancel out the scale gains won from paid traffic.
Another central focus was prioritizing retained profit over gross revenue volume. Growing just to show big numbers forces you to inflate the team, tool licenses and coordination costs. More than doubling revenue, what matters is how much net margin is left in the account at the end of each quarter.
I also learned to protect the team in relationships with clients. When a client disrespects the agreed contractual scope or behaves destructively toward the technical staff, the relationship needs to be resolved without hesitation. Keeping toxic clients around just to hold up revenue numbers damages internal culture. I develop this view in Crescer um negócio online: o que aprendi sobre margem, equipa e IA.
Retention: Financial Sustainability After the Initial Close
Business growth depends on the value each customer generates throughout the relationship, not just at signup. High acquisition costs paired with early churn drain cash flow and block consolidation. Financial sustainability requires retaining customers long enough for their lifetime value to exceed what was spent acquiring them initially.
In Sextas Ímpares class #108, on retention strategies, I talked about communication, expectation alignment and active listening. A customer can receive everything agreed on time and still end up dissatisfied simply because they don't understand the metrics presented or the practical impact of the work done.
Retention work starts at the welcome session: clear service boundaries, a delivery calendar and responsibilities for each side. If project goals change midway through, the impact on budget and timelines needs to be formalized immediately, so frustrations don't build up based on informal assumptions.
When presenting periodic reports, detail what's been achieved, the obstacles encountered and recommendations for the next phase. In the class mentioned, I talk about continuous communication at 59:07 and about periodic experience evaluation at 1:44:39. Carefully auditing the reasons behind cancellations lets you tell apart technical team failures from legitimate adjustments in the client's own business.
Automation and AI in Support of Business Processes
Artificial intelligence and automation exist to improve internal processes that are already consolidated, reducing repetitive tasks and speeding up data reading. Applying technology on top of disorganized sales processes only amplifies the confusion and produces reports with no practical use.
On the YouTube broadcasts I demonstrate this application regularly. In Sextas Ímpares class #142, on AI tools in advertising, I showed how to process traffic metrics and connect them to dashboards. In Sextas Ímpares class #143, on the shift toward internal systems, I talked about the need to build tools tailored to concrete operational difficulties, instead of paying for generic software licenses that nobody uses fully.
In Sextas Ímpares class #145, on practical building with Claude Code, I explained that much of the result depends on preparation and architecture planning before you start generating code. A database with weak security rules or disorganized fields doesn't generate analytical value, no matter which model is used. In the article Como planear uma aplicação com Claude Code antes de começar a programar I explain the architecture method I follow.
In Sextas Ímpares class #151, on the evolution of the traffic role, I showed how AI can help with continuous campaign monitoring: profitability drop alerts, WhatsApp notifications and cost containment rules on advertising accounts. The idea isn't to use the machine to decide company strategy, but to precisely execute the safety parameters defined by leadership. I develop these approaches in Como usar a IA no negócio para além das perguntas ao chatbot.
Data integrity and compliance with the General Data Protection Regulation are non-negotiable boundaries. API connections should carry only the minimum permissions necessary for the task, and sensitive variables should never sit exposed in open repositories. Every automation we build follows careful technical governance principles, protecting both the operation and the privacy of clients.
How I Can Help Your Business
To work on customer acquisition and sales, there's SpartAds. To apply AI or automate a process, there's SpartAds.ai. If you'd rather learn and bring your own project into the discussion, check out the training programs. Choose based on the work you need done and the support you're looking for.
If you're looking for a team to take over your acquisition campaigns and structure your conversion funnel, check out the work of SpartAds for service businesses and e-commerce. If the goal is to build internal tools, connect platforms via API, or replace bureaucratic tasks with AI architecture, the path is SpartAds.ai.
If you want to deepen practical skills and work on your business in close contact, check out the Laboratório da IA and the available training programs. I run two two mentoring sessions per month dedicated to AI and online business, reviewing students' actual projects.
I'm also available to evaluate businesses for investment, providing capital, strategic guidance, team reinforcement or a combination of these. The essential criterion is demonstrable prior execution: time invested, a team involved, or own capital already put in, whether the business already has customers or no revenue yet. I don't evaluate ideas that exist only on paper.
If you want to bring me in for a talk or training tailored to your team, send the details of the invitation, the audience profile and the session's goals so my team can check availability and put together a proposal suited to your context.
Turning the diagnosis into a working meeting
A growth meeting should end with a decision, an owner and a way to check the result. Bring information about demand, enquiries, proposals, sales and delivery capacity across comparable periods. Choose one concrete difficulty to work on, and avoid changing several stages at once if you cannot distinguish the effect of each change.
Imagine, as an example, a business that received more enquiries this month but booked fewer meetings. Before changing the advertising, compare enquiry sources, time to first response and the proportion of people the team reached. Also check whether the definition of a booked meeting stayed the same across both periods.
Include the people handling those enquiries in the conversation. Requests may be poorly distributed, missing information or arriving when the proposed meeting times are unavailable. Each explanation remains a hypothesis until checked against the records. Team experience helps identify what to investigate; data helps establish how widespread it is.
Choose a change you can follow. For example, test an enquiry allocation rule over an agreed period and record who owns each request. Define what happens when that person is absent too. Assigning a name is not enough if the enquiry still has no next action.
Review the whole journey afterwards. A faster response may help, but you also need to know whether it led to a conversation, meeting, proposal and decision. Keep recent enquiries separate from those that have had time to progress. Comparing opportunities at different stages can lead you to misjudge the sales team's work.
Look at delivery as well. If the business closes more sales, check that the team can fulfil its promises and that costs leave the expected margin. A commercial improvement may expose an operational limitation that was less visible before. That is why I discuss acquisition, people and capacity together.
Keep the decision and result even when the test does not help. Knowing that a hypothesis has been checked prevents the same discussion from repeating without new information. The next meeting can then start with what you learned and the next issue to resolve, instead of another round of opinions about advertising.