Digital marketing

What digital marketing is and which channel to start with

Digital marketing means getting one specific message to one specific person, in a channel that person is already in. The whole job is three decisions: what, to whom, and in which channel; and the channel is the last decision, not the first.

  • Lesson 1 of 8
  • Beginner
  • Free, no signup

The five channels that mean something for a small Iranian business

A channel is a place the audience is already in. Content is not on this ring because it is not a channel, it is the fuel that burns inside each of them.

Your business the channel is the last decision
  • Search
  • Social
  • Messaging
  • Email and phone list
  • Paid advertising

No business needs all five, and someone starting all five together will not know at the end of the quarter which one worked. Advertising is not a sixth channel either; it only buys more room inside the first three.

Last checked: Facts and tool names in this lesson are re-checked against their sources on this date.

What exactly does digital marketing include

Digital marketing means getting a message to someone through a channel that person is already in. The important word there is channel. A channel is a place where the audience actually is and where you have permission or a route to reach them; not a technique sold inside a course, and not a tool installed in your browser toolbar.

For a small Iranian business the real channels are close to fixed, and there are no more than five. Search, where a person is looking for something themselves and you are the answer to that search. Social, where they spend time and are not looking for you yet. Messaging, where they talk to you one to one. Email and phone numbers, where you hold the list yourself. And paid advertising, which uses money to buy more room inside the first three.

Content is not on that list, deliberately. Content is not a channel but the fuel that burns inside each of them: one piece of writing is an article in Google, a carousel on Instagram and a newsletter in email. People who treat content writing and digital marketing as the same thing usually make the content first and then look for somewhere to put it, which is the order reversed.

A black target with one arrow at the centre and three rings lit red, green and blue beside a small megaphone

What is the difference between an owned and a rented audience

This split is the organising idea of the whole path. An owned audience is one where the route to them is in your hands: a phone number you hold, an email list, a site on your own domain. A rented audience is one where another company decides whether you reach them: Instagram followers, Telegram channel members, and yes, a Google ranking too.

I put the Google ranking on the rented side deliberately, because most places count it as an asset. The asset is your site; its ranking is rented, because somebody else decides it and that decision can change this year. Owned and rented are not separated by audience size, but by what is left for you when the landlord changes the rules.

Rented is not bad and nobody is telling you to stay away from it. A rented audience is exactly where new people are found; nobody hands you an owned list. The right position is this: build on owned, borrow from rented. Which means every contact in a rented channel has one small goal, and that goal is not the sale; it is one step towards a place you own. A number, an email, a visit to a page that belongs to you.

And one warning that never appears when this idea is being sold: owned is not free. When you hold your own list, the trouble of delivering the message is yours too. The work Instagram was doing for you at no charge is now your work, and like any infrastructure work, when it breaks it breaks quietly.

The audience you rent and the audience you own

Both are needed and they do different jobs. Rented finds new people, owned keeps the people already found.

Rented audience

  • Followers, channel members, a Google ranking
  • Another company decides how many of them see you
  • The rules can change without notice
  • This is where new people are found

Owned audience

  • Phone numbers, an email list, a site on your own domain
  • The route to the audience is in your hands
  • If that platform closes tomorrow, this stays
  • The trouble of delivering the message is yours too

Owned does not win outright: an email list you never write to effectively does not exist a year later, and delivering the message on your own channel brings its own technical trouble. What ruins the work is having only the rented side.

Where to start when the budget is close to zero

The short answer: start from the data already sitting in your phone, which no competitor can reach. Take your last twenty customers and write beside each name where they found you. Call history, WhatsApp messages, Instagram direct, a friend recommending you. This takes an evening, and the result nearly always surprises somebody, because the channel we spend the most time on is usually not the one most customers came from.

The next step is harder because it subtracts rather than adds: take that one most frequent channel seriously for three months and drop the rest. Dropping means actually nothing. Someone holding Instagram, Telegram, a site and a video account at the same time holds all four at half strength, and at the end of the quarter still does not know which one worked.

The third move is building one owned place whose address you can hand out. It does not have to be a full site. One page saying what you do, for whom, where you are and how to reach you is ahead of an Instagram profile with no address, because it cannot be lost and nobody can close it.

And the move with more effect than all three above, which no course can sell you: answer messages the same day. Most small businesses saying their advertising did not work did not have an advertising problem; the person who messaged waited two days and went elsewhere. Replying fast is free, and that is exactly why nobody is selling it to you.

The first four moves when the budget is near zero

None of them needs money, and the second is the hardest, because it subtracts rather than adds.

  1. Write down your last twenty customers

    Beside each name: where they found you. Your call history and messages already hold this.

    1

  2. One channel only, for three months

    Keep the most frequent one and actually drop the rest, not at half strength.

    2

  3. One owned place with an address

    One page is enough: what you do, for whom, where, and how to reach you.

    3

  4. Answer the same day

    The freest move and the most effective one. Which is why nobody sells it.

    4

Twenty customers is a small number and the result is not precise; it is only enough to show which channel you have been underrating. If your business is seasonal, even those twenty can mislead you.

Where does SEO sit in all this

SEO is one of those five channels and the slowest of them. Its job is that when someone types a question into Google, you are the answer; not that you go to them. That single difference separates its effect from the other channels: an SEO audience is already looking for something, so it sits closer to a purchase, but you do not set how many of them there are and in the first month it is close to zero.

The thing an agency usually will not say: for many small businesses SEO is not the bottleneck. If you get three customers a month from Instagram and have no page whose address you can hand out, SEO is not the first job. If you answer messages within two days, SEO does not fix your problem and only walks more people into the same waiting queue.

SEO is the most correct choice when three conditions hold together: what you sell is searched for in Google, you have a site that can be worked on, and you can wait a few months. The depth of this channel, from keyword research to technical SEO, is in the SEO path.

How do you know it is working

With one number chosen before you start. The words "before you start" carry the whole weight: a number picked after seeing the result is always the one that happened to look good. If views went up and calls did not, the definition of success must not switch to views halfway through.

For a small business that one number is nearly always the count of people who actually made contact: a message, a phone call, an order. Followers, views and likes get serious only when they connect to that one number, and in most months they do not.

The road between someone seeing you and someone buying has several steps, and every step is a place to leak. That road is called a funnel, and the next lesson is about nothing else: how to see where people are dropping out, and which leak to fix first.

The fast path, with AI

If you ask a language model which channel is best for your business, you get an answer written for everyone, which is exactly why it is useful to no one. What actually works is to hand it your own evidence instead of a question, and to bind it with two constraints: it may not name a channel that is absent from the input, and it must say what you have to drop.

  1. Write twenty lines, one per recent customer: where they found you, what they bought, and how long it took them to buy. Leave out names and numbers; "customer 1" is enough, and this cannot be undone later.
  2. Add two more numbers that exist nowhere on the internet: how many hours a week you genuinely have, and how much you can spend on ads per month. If the second is zero, write zero.
  3. A cheap fast model of the Flash class is enough, because this pass is counting and sorting rather than judgement. The current pick is listed in <a class="text-link" href="/en/ai/">the AI section</a> of this site. Keep the expensive model for when you genuinely have to choose between two options.
  4. Run the recipe below. If the output proposes a channel that is not in your twenty lines, reject it there and ask again; that single clause is what separates this recipe from a generic answer.
  5. Leave the output alone for three months. If you ask the model again in week three you will get a different answer, and that difference came from the sentences being rewritten, not from new data.

Copy-ready recipe

These are the last twenty customers of a {type of business} in {city}:

{twenty lines: where they came from / what they bought / how many days it took}

My time: {number} hours per week. Ad budget: {number} per month.

Rule: name only channels that appear in these twenty lines. Do not propose any other channel, even if you think it is better.

1) Count how many customers came from each channel and write beside it what those customers bought.
2) Pick one channel for three months and say in one sentence why that one.
3) Write what I have to stop doing to free up those {number} hours. Name them.
4) Give me a weekly plan that fits inside those {number} hours; if it does not fit, cut it yourself.
5) Say which single number I should look at after three months, and what its value is right now.
6) Do not write any average or industry standard rate. If a number is not in my input, do not write it.

Before you trust the output: The output of this recipe is a hypothesis, not a result. Twenty customers is not enough for any serious statistic, and if your work is seasonal those twenty can point you at the wrong channel. More importantly, you have to review clauses three and four yourself: the model does not know which task you cannot drop and which hour of the week is genuinely free.

AI in this kind of work

In this job AI is useful in one specific place and dangerous in another, and unfortunately the dangerous one is where everybody goes first. Useful: sorting the evidence you already hold and have no patience to read. Dangerous: asking it for a decision or an opinion. The model does not know the market in your city, and when it does not know it guesses, and its guess looks like an answer.

Tools that actually help

  • Claude It holds a constraint well, and in this job the constraint is the most important part of the recipe: a channel absent from the input must not be named. Iran is on neither of the two Anthropic supported country lists; we read that on the Anthropic page itself.
  • Gemini It understands Persian itself, so twenty Persian lines need no translation first. Google own page says the app works in over 230 countries and territories and more than 70 languages, and Iran is not on that list.
  • ChatGPT It works for this same counting and sorting pass, and most people are already comfortable with it. We make no claim about access from Iran: the OpenAI supported-countries page returns 403 to this server, and we do not write a claim with nothing behind it.

Where it backfires

Two specific risks. First, the very thing most people do first: asking for an opinion. Put your own plan in front of a model and ask whether it is good, and the answer is usually yes. This is not politeness, it is a known pattern in how these models behave, and OpenAI writes avoiding exactly this sycophancy into its own model spec as an explicit goal, which is to say it does not deny the tendency exists. If you genuinely want criticism, instead of asking what it thinks, ask it to write three reasons this plan fails. The second risk is the data: twenty customer lines are personal data about real people, and whether your conversation is used for training depends on the plan and the settings. Anthropic states its rule for consumer products on its own page. Strip names and numbers before you paste.

Sources: OpenAI Model Spec: avoid sycophancy Anthropic: supported countries and regions Google: where Gemini Apps are available Anthropic: is my data used for model training

Where this advice stops

This lesson is a map of the channels, not the skill of any of them. Three things are left out on purpose. The mechanics of paid advertising, from building a campaign to controlling the spend, are the later lessons of this path. We have not written the internal rules of each platform because they change month to month and any number written today would be wrong within months. And the whole access question: some of the tools treated as standard in foreign sources are unreachable from Iran or refuse an Iranian account, so any advice assuming everything is open is half an answer here. One boundary larger than all three: picking a channel does not change the answer to why you and not the cheaper option. If you do not have that answer, the right channel only shows more people faster that you do not have it.

From our own work

We learned that owned is not free from our own email. This site messages leave from our own server, and to actually arrive, the outgoing Postfix on this machine is locked to IP version four: in the config file, inet_protocols and smtp_address_preference are both set to ipv4, because the version six address on this machine has no reverse record and receivers reject or junk a message on that single point. The signing is done by a separate service on the same server. The practical consequence is this: when the WordPress mail function returns a success value, nothing has been proven yet; the real answer is in the mail service log. A rented channel has none of this trouble, and that is exactly the price you pay for ownership.

Real follow-up questions

How is digital marketing different from traditional marketing?

In two things: it can be measured and it can be started small. You cannot show a city billboard to twenty people and see how many called; a post or an online ad you can. The rest of the work is the same marketing and has not aged: you still have to know what you are saying to whom.

Should I build a site first or start on Instagram first?

If you have neither, start where your customers already are; for most small Iranian businesses that is Instagram or a messenger, not a site. But build one simple page on day one anyway, a page whose address you can hand out, because your social profile can be limited or closed and that page cannot. The usual right order is: the rented channel to get found, the owned page to stay.

Can you get a result with no ad budget at all?

You can, but slower, and you pay in another currency: time. With no budget your only levers are repetition and speed of reply, and both cost hours. Someone with neither money nor a few hours a week is better off reactivating past customers than starting a new channel; the cheapest customer is an old one.

Keep learning