01The Direct Answer
My agency seems to do good work, but sales do not grow. How is that possible?
Direct Answer
It is possible because a supplier is contracted, measured and paid on a segment of the path, and revenue only happens at the end of the whole path.
Every party in the arrangement can be doing exactly what was asked, competently, and the result can still be flat — because the thing that determines whether an inquiry becomes an order lives between the parties. It is what the buyer was looking for when they searched, what they were told when they arrived, who answered them and how fast, what that person asked, and whether any of that came back to the people producing the next campaign. None of those belongs to a scope of work, which is precisely why nobody is failing and nothing is improving.
02How the Numbers Stay Right
Three ways a correct indicator produces a wrong buyer
These are not failures of execution. Each one is a supplier doing well exactly what it was hired to do, inside a specification that nobody wrote.
Where the Local Success Comes From
Search: your words, not the buyer words
The pages rank for how your catalogue names the product. The buyer searches by the problem, the application or the failure — a term your catalogue may not contain at all. Ranking improves honestly while the traffic is technically right and commercially beside the point.
→
Media: optimising towards what it can see
The platform is told that a submitted form is the goal, because that is the last event it receives. It then learns to find people who submit forms. If the form asks nothing that changes how the inquiry is routed, cost per form falls while the composition of who is filling it drifts.
→
Content: right answer, wrong moment
Articles explain how to install, operate and maintain — questions asked after the supplier has been chosen. The questions asked while choosing are about risk: what happens if it fails, who else uses it, how long delivery takes. Those get answered by whoever publishes them, and it may not be you.
In all three, the supplier can present a report that is true, defensible and improving. The report is not a lie and demanding a different report will not change the outcome — the number is measuring the segment it was given.
03Local Optimum, Aggregate Outcome
Every station is running well and the line is not producing
This is a familiar shape on a factory floor. Every station has good utilisation, every operator hits their own target, and the line still does not deliver, because the constraint is between stations and no station is measured on it. Marketing arrangements fail the same way, with one aggravating difference: on the floor everyone can see the pile of work-in-progress accumulating. Here the equivalent pile is invisible — it is inquiries that were answered late, or answered by someone who did not know what to ask, or never answered at all.
The Mechanism in One Sentence
Optimising each segment separately does not optimise the path, and can quietly make the path worse.
It can make it worse because segment targets compete. Lowering cost per form pushes towards the audience that fills forms most easily, which is rarely the technical buyer with a specification in hand. Raising ranking pushes towards the terms with volume, which are usually the generic ones. Increasing publishing frequency pushes towards subjects that are quick to write. Each pull is individually rational and the sum drifts away from the buyer you actually want.
04The Context Test
One question that settles whether the supplier is the variable
The Test
If you replaced this agency tomorrow, would the new one receive better context than this one received?
Answer it with what exists in writing, not with what you would intend to explain. Is there a document stating which customer, with which application, in which words? A closed list of what counts as a good lead, signed by sales? A record of which hypotheses were tried and what they produced? If the honest answer is that the new supplier would receive the same thing, then the supplier is not the variable in this system, and replacing it changes the variable you are not testing.
The test is uncomfortable on purpose. It is the only question in this cluster that cannot be answered by attributing the problem to someone else, and that is what makes it useful.
05What the Agency Cannot See
Ask your supplier what it is blind to
This is a good question to ask in the next meeting, and it is not a trap. A competent supplier answers it precisely and with some relief, because the list is usually long and it has been carrying it alone.
The Usual Blind Spots
What happens after the form
Who called, how long it took, how many attempts were made, what was said. The supplier usually has no access to any of it and is optimising towards the last event it can see.
→
Why deals were lost
Price, lead time, specification, a competitor already installed, no budget this year. Each of those implies a different message, and none of them reaches whoever writes the messages.
→
What the buyer says on the call
The exact words used to describe the problem — which are the words that should be on the page and in the ad, and which no keyword tool will ever hand over.
→
Which orders were actually good
Margin, repeat purchase, technical trouble afterwards. Without it, the supplier optimises towards volume of orders, which is not the same as optimising towards the orders worth having.
If the supplier answers this list accurately, you are not dealing with incompetence — you are dealing with information it never received. That is a fixable arrangement, and it is fixed on your side of the table.
06The Missing Return Path
What has to come back, and what makes it usable
One return path is worth more here than any new front: the loss reason, arriving at whoever produces demand. Without it, every campaign is a first campaign, and a supplier that has been with you for years is still guessing with the same information it had in the first month.
What Makes the Return Path Work
A closed list
Free text does not aggregate and will not be read. A short set of reasons, agreed by sales, is what turns individual losses into a pattern somebody can act on.
→
Written by who lost it
Not by a manager reconstructing it later. The reason is accurate only while the conversation is fresh, and it degrades into "price" within a week.
→
Arriving on a fixed rhythm
A standing slot where the same three questions are asked. Information that arrives only when someone remembers to send it is information that arrives when things are already bad.
→
Reaching the supplier, not just the folder
The loss reasons have to land with the people writing the ads and the pages. Stored in a system that only the sales manager opens, they change nothing.
The pattern that emerges from a few months of this usually reorganises the message more than any creative round would. [EVIDÊNCIA NECESSÁRIA: uma conta em que indicadores de canal melhoraram e o faturamento não, com a distribuição de motivos de perda antes e depois de o retorno existir. É a evidência que sustenta esta seção como constatação em vez de recomendação.]
07The Next Question
This page does not tell you to keep or replace anyone
Deliberately so. Three mechanisms were described above and they call for different responses: a vocabulary gap is corrected by rewriting pages, an optimisation target is corrected by changing what the form asks and what event is sent back, a missing return path is corrected by a routine. None of the three is corrected by replacing the supplier, and two of them the supplier can correct itself once it receives what it was missing. Find out which one is operating before deciding anything about the contract — deciding first is how the previous switch happened.
There is also a version of this problem that appears when the arrangement is not one good agency but a full set of suppliers, each competent in its own front. That is the next page: why owning every piece does not produce a predictable pipeline.
What both versions have in common is the material that would let each supplier execute against the same specification — who the customer is, in which words, what counts as a good lead, what comes back when a deal is lost. Producing that material deliberately, and keeping it with the company, is the work of the Prebound Marketing method.
Prebound Marketing is cheaper before.
08FAQ
FAQ
My agency seems to do good work, but sales do not grow. How is that possible?
Because the supplier is contracted, measured and paid on a segment of the path, and revenue only happens at the end of the whole path. Search ranks for the words in your catalogue rather than the words the buyer types. Media optimises towards the last event it can see, which is the submitted form, not the order. Content answers what is asked after the supplier has been chosen. All three reports can be true and improving while what determines the order — who answered, how fast, and what they asked — sits outside every scope of work.
What can explain a good agency with flat sales?
Three mechanisms, and they often run together. Vocabulary: the pages describe the product the way your catalogue does, while the buyer searches by the problem or the application. Optimisation target: the platform was told a submitted form is the goal, so it learns to find people who fill forms, and if the form asks nothing that changes routing, the composition of who fills it drifts. Timing: the content answers installation and maintenance questions, which come after the choice, rather than risk questions, which come during it.
How do I tell incompetence from information the agency never received?
Ask the supplier what it cannot see. A competent one answers precisely: what happens after the form, why deals were lost, what the buyer says on the call, which orders had good margin. If the list is accurate and long, you are not facing incompetence — you are facing information the supplier never received, and that is fixed on your side. If the supplier cannot even name what it is blind to, that is an execution signal and belongs to a different conversation.
Which numbers show whether it is vocabulary, optimisation target or the missing return path?
For vocabulary: the words buyers actually used in recent inquiries and calls, compared with the terms your pages rank for. For the optimisation target: whether the form asks anything that changes how the inquiry is routed, and whether any event after the form is sent back to the platform. For timing: which questions arrive in the first contact, compared with what the published content answers. For the joint: the time between an inquiry arriving and the first contact attempt, and whether loss reasons reach whoever writes the campaigns.
Should I change the agency contract to pay on results?
Only for results the supplier can influence, otherwise the contract becomes a dispute about attribution instead of an incentive. It is reasonable to tie payment to the composition of what arrives — segment, application, size — and to the response the message produces. It is not reasonable to tie it to signed orders when the supplier has no access to what happens after the form, cannot influence lead time or price, and does not choose who calls the buyer back.
Does adding another front — media, automation, content — help at this point?
Usually not, and it can make the reading harder. A new front adds one more segment with its own target and one more handover with no owner, so the same mechanism now runs in four places instead of three. It also makes it more difficult to tell what caused any change, because two things moved at once. Close the return path first, with what you already have. If the mechanism persists after the loss reasons are reaching whoever writes the campaigns, then adding capacity is a reasonable next step.
My sales team says the leads are bad and the agency says the team does not call. Who is right?
The argument is unresolvable without two facts, and both are in your CRM rather than in either opinion: the time between an inquiry arriving and the first contact attempt, and the number of attempts made per inquiry. If contact is slow or shallow, quality cannot be judged yet because you are measuring follow-up, not quality. If contact is fast and persistent and the inquiries still do not fit, then the composition is the issue and it is settled by writing down what counts as a good lead.
When does the next question become "we bought everything and the problem is still here"?
When the arrangement is no longer one agency but a set of suppliers, each competent in its own front — site, search, media, CRM, content — and the flat line persists across all of them. At that point the mechanism is the same one scaled up: every piece answers a local question well, and the questions that cross all of them belong to nobody. That is a different page, because the fix is not about any single contract.
Measure the joint before scaling the segments
If every supplier is doing well and the line is flat, the constraint is between them. The specification that lets all of them execute against the same definition — and the return path that tells them whether it worked — is what the Prebound Marketing method organizes.
