A letter worth opening.
Delivered by a person.

If you sell something complicated, your market rarely fits inside the category you put on the homepage. A tool sold to a software team might also help a service business running the same workflow. A product bought by a founder in a small company might be evaluated by an operations manager in a larger one. The job is to discover those differences before we write.

We carry the constraints into the search. Geography, delivery capacity, implementation effort, deal size, prerequisites and the situations you cannot handle all matter. A beautiful market map is useless if every company on it needs a product you have not built. We want an audience your business can serve when somebody answers.

Give a group of people a product and they will usually name the markets they already know. That is a reasonable starting point. It is also an easy place to stop. The familiar ideas feel convincing, so nobody spends the afternoon looking for the less obvious company with the same underlying problem.

I kept coming back to this in the Market Scan work. The hard part was making the system continue after the easy answers. It had to follow adjacent uses, study competing products and look at the companies buying those products. Then it had to explain the connection. Another industry name, on its own, adds almost nothing.

We expand the candidate markets and test the reasoning behind each one. Who has the problem? What creates it? What do they do today? Where does your product fit into that process? What evidence would make us take the idea seriously? Those questions turn a loose brainstorm into something we can inspect and improve.

Some ideas survive with strong evidence. Some remain plausible and need a small, deliberate test. Some fall apart when we read the source. We keep those outcomes separate. Uncertainty belongs in the working file, where it can influence a decision.

We study competitors because they leave clues about the market. Their product pages, customer stories, integration pages and public descriptions show which problems they choose to explain. A customer logo can suggest a company to investigate. A detailed case study can reveal a workflow that the logo never could.

We follow the relationship carefully. Did that company buy the competing product? Was it a partner, an investor, a trial user or simply mentioned in an article? What was the actual use? A logo cloud cannot answer all of that. We go back to the surrounding material before turning a brand into a sales hypothesis.

The same care applies to an unusual customer. It might reveal a market nobody on the first call mentioned. It might also be a one-off project with conditions your business cannot reproduce. We look for another example and for the reason the use case exists. One attractive anomaly is a research lead.

This is where the business file earns its keep. We can compare a competitor's promise with your actual capability. If the promise depends on something you do not provide, we do not smuggle it into your letter. We either find a different angle or put the market aside.

Broad labels hide different businesses. A manufacturer, a distributor and a software vendor can all describe themselves with the same industry word. They buy differently. They have different margins, workflows and people responsible for the problem. A list built from the label alone mixes them together.

We describe the type of company we mean in practical language. What does it sell? Who does it sell to? How is the relevant work performed? Which condition makes the offer useful? We also describe the businesses that look similar but should be excluded. That second description saves a surprising amount of wasted research.

Related ideas are grouped when the buying logic is genuinely shared. We split them when the reason to buy changes. This keeps the research readable without flattening useful differences. You should be able to look at a market and understand why its companies belong together.

The result guides the search for actual organisations. It also gives the writer a coherent audience. A letter to a company operating a delivery fleet has to address the work of operating that fleet. Dropping the word logistics into a generic paragraph does not get us there.

We then ask who owns the problem inside each type of company. Who experiences it? Who can evaluate a solution? Who controls the budget or can bring the decision to the person who does? In a small business, those may be the same person. In a larger organisation, they may be several people.

This was one of the recurring problems in my Market Scan notes: people with relevant responsibility are called all sorts of things. Searching for founder and chief executive everywhere feels efficient until the search misses the person actually running the work. A senior title does not prove that the letter belongs on that desk.

We build the vocabulary for each market. We look at how companies describe their teams, responsibilities and current vacancies. We compare the titles used by similar businesses. We use those findings to search for the person, then check whether the responsibility still matches.

A name and a title are separate facts. So are the name and the employer. People move. Companies reorganise. We want the source behind the match and the date it was checked. If the evidence conflicts, the record needs another pass before the envelope gets a name.

Once the audience is clear, we look for companies that actually fit it. Each candidate should carry the reason it was selected. The company website, relevant public evidence and our notes stay attached to the record. Otherwise a carefully researched audience turns into a spreadsheet nobody can explain a week later.

For the person, we collect the first and last name, the relevant role and the relationship to the company. A middle name is included only when it is known and useful for identification. We do not invent one to make a row look finished. The name we print must be the name we can support.

We also seek a reachable business telephone number and the physical mailing address. Where an email address is available and relevant, it can sit in the record too. The physical letter needs a real receiving location, so an email match cannot substitute for an address check.

A main office switchboard and a person's direct business line are different contact routes. We label what we have. The same goes for a headquarters address, a local office and an address that appears to exist only for correspondence. The delivery team needs that distinction more than it needs a full-looking table.

Finding an address online is one step. Establishing that it is a sensible place to deliver this letter is another. We check the office, street address, suite or floor where available, and the connection between the location and the company. We look for signs that the company moved or that the intended team works elsewhere.

We also consider how the building receives visitors and correspondence. Some offices have a reception desk. Some have a central mailroom. Some need a delivery arranged in advance. Those differences affect the route and what can honestly be recorded as a handoff.

For this service, the research ends in a delivery-ready record: company, commercial reason, person, role, contact route, address and delivery notes. Where something material is missing, we resolve it or replace the candidate. The unresolved field is a task, not an invitation to guess.

We review the final audience with you before production. You may know that a company is already a customer, that a conversation ended badly or that a territory is unavailable. That knowledge belongs in the business file and in the exclusion list. It should reach the courier before the courier reaches the door.

The source notes tell a fairly physical story. I was connecting research tools, looking at the gaps between them and writing the instructions that would make the next pass useful. A service could enrich a list. Another could search for a contact. I still needed something that could explain which list to build and why.

In the early work, I described a flow that joined business research, competitor research and the client's call recording in a single workspace. That was the point: the call should change the understanding of the business, and that understanding should change the campaign. I wanted the call, the research and the campaign to stay connected.

Later, I wrote about rebuilding the research through several versions. The challenge kept moving. Finding more ideas created a need to organise them. Organising them created a need to distinguish evidence from a convincing paragraph. Finding the companies exposed the vocabulary problem around roles and titles.

I kept connecting tools. The questions stayed the same. Why this market? Why this company? Why this person? What are we going to say? If the answer collapses when somebody asks for the source, the system has produced more work for us.

A short request to research a business leaves a lot for a model to improvise. What counts as evidence? How far should the search go? What should happen when two sources disagree? When is a market too weak to pursue? If those questions are unspecified, the answer changes with the model's assumptions.

I spent time writing down the world each research step operates in, the task it has to complete and what it has to hand to the next step. The July notes describe an orchestrator coordinating 15 sequential sessions, followed by three reviewing judges. Each session had a job, and the reviews sent weak work back for another pass.

The useful principle is the handoff. A researcher should leave enough context for the next researcher to continue intelligently. A reviewer should be able to send back a weak section with a specific reason. A writer should receive a business file and an audience brief that answer the questions a letter depends on.

I kept finding the same problem: a paragraph sounded finished, then fell apart when I checked the claim. That sends the work back to research. A better sentence cannot supply a missing fact.

The researcher is trying to establish what is true and what is worth testing. The writer is trying to make a particular person understand an offer and take a sensible next step. These jobs share material, but they make different demands on it. We keep the evidence visible through both.

One of my writing experiments made this painfully clear. I gave a model the writing instructions without the client knowledge those instructions depended on. It could produce something that looked like a letter. That did not solve the missing-context problem. It had no solid account of the sender, the buyer or the reason for the approach.

That lesson is built into this campaign. The writer receives the approved facts about your business, the market logic and the facts about the recipient that actually matter. We use specific evidence where it earns attention. We remove invented familiarity, decorative compliments and claims the file cannot support.

Review is another job again. We check whether the argument fits the recipient, whether the proof supports it and whether the request is easy to answer. When the underlying fact is weak, polishing the sentence is the wrong repair. The research goes back for another pass.

A campaign contains several versions of the truth as the work improves: the original material, our first interpretation, a corrected audience and an approved letter. We retain enough of that trail to understand what changed. The version on the printer must match the version you approved.

The recipient record connects the company and person to the final letter and the selected physical object. It also carries the address and delivery instructions. That connection matters when the report arrives. You should not have to work out which copy was sent to which office from a pile of filenames.

We use missing information as a stop signal at the point where it matters. An unverified name needs checking before personalisation. An uncertain office needs checking before routing. A claim without support needs checking before approval. Fixing an empty field is much cheaper while the letter is still on a screen.

You can follow a selected company from the research through to its letter and delivery record.

After delivery, the campaign has new information. A recipient can be interested, decline, name another person, explain that the timing is wrong or simply say nothing. These outcomes mean different things. A replacement contact can improve the next approach. A clear objection can expose a weakness in the offer.

We add the response and delivery evidence back to the campaign record. We distinguish what the courier observed from what the recipient actually said. A receptionist accepting an envelope does not establish that an executive read it. A quiet week does not tell us why there was no reply.

When a response challenges a market assumption, we review the assumption. We can change the audience, the angle or the next question. The point of keeping the research is that we have something specific to revise. A campaign should leave the business better informed about its market, including when an idea fails.

Market Scan is the research engine inside this service. The physical campaign adds writing, production, delivery and reporting around it. The longer development story lives on its own page, with the original sequence of problems and revisions. You can read that before you send us anything.

One of the practical problems in the development notes was vocabulary. A company can belong to the market you want and describe itself in words you never searched. In the gambling example, the notes list igaming, real money gaming, gambling and casino. A search for one label gives you one view of the market. It does not settle who belongs in it.

That is why the research expands the names before it expands the list. We collect the language companies use about themselves, the language their customers use and the names used in specialist directories. Then we check which words actually describe the buyer we want. Similar words can lead to very different businesses.

Take a supplier selling an operational service. A software vendor serving an industry may have an entirely different reason to buy from a company operating inside that industry. Putting both into a broad category hides that difference. We separate the commercial situations, then decide which ones deserve a letter.

You should be able to read a market description and recognise a business that could buy. If the description covers almost everyone with a website, the work needs another pass. The objective is a search we can execute and an argument we can defend when a company reaches the shortlist.

In one development note, the input supplied by the client comes first, the landing page second and the material found through search third. That is a useful way to understand the problem. Websites can lag behind the business. A founder can be selling something different from the offer still sitting on the home page.

We keep those disagreements visible. If the brief describes a new product and the website describes the old one, we ask which facts the campaign may use and what evidence supports the new offer. We do not blend the two into a fluent paragraph and hope nobody notices.

A sales call can contain the detail that changes the audience. A lost proposal can reveal a condition that makes the service a bad fit. An analyst can point out that a competitor's customer looks attractive but needs something you cannot deliver. These belong in the business file because they change who should receive the letter.

This also gives you something specific to correct. You can tell us that a market is interesting but the timing is wrong, or that a product claim needs narrower wording. The next research pass then starts from the correction. It does not rediscover the same mistake under a different heading.

A useful scan reduces work as well as finding it. We may remove a company because the relevant function is outsourced, the product needs infrastructure it does not have, or the decision sits outside the delivery area. Those are different reasons. Keeping them separate makes the next search better.

We also separate an unsuitable company from an unreachable person. The business may be an excellent fit, yet the intended recipient has moved or the building accepts deliveries only through a central room. That is a delivery problem to resolve before printing. It does not automatically invalidate the market.

For a physical campaign, this is where research meets the route sheet. We need a workable entrance, the correct office, an acceptable receiving person and enough geographic concentration to use courier hours well. Grouping the offices into workable routes keeps the research useful to the delivery team.

We group the addresses into city routes and brief the couriers before dispatch. They need to know which office to enter, who the letter is for and how the building receives deliveries. The research goes all the way to the door.