Most businesses start digital marketing by picking channels. That is backwards. You start by deciding what a customer is worth and what you can afford to pay to get one. Everything after that is execution. This is the execution guide: the exact numbers, structures, and steps to launch an engine that survives contact with your P&L.
Step 1: Set your numbers before you spend a euro
You cannot run paid acquisition without three numbers. Get them on one line before you open an ad account. First, average order value or first deal value. Second, contribution margin: revenue minus the variable costs to fulfill one order (COGS, payment fees, shipping, fulfillment). Third, your target CAC payback: how many months the profit from a customer takes to repay what you paid to acquire them.
Here is the move. Allowable CAC equals contribution margin per customer multiplied by your payback tolerance. If a customer throws off 40 euros of contribution margin per month and you can stomach a 3-month payback, your ceiling is roughly 120 euros to acquire one. That ceiling, not a channel, decides what you can run. If the math is ugly, fix the offer first: read what makes a great value proposition, because cheap traffic cannot rescue a weak offer.
- Pull AOV (or first contract value) from the last 90 days of real orders, not a guess.
- Subtract variable costs to get contribution margin per order. That is the money you actually keep.
- Pick a payback tolerance: 1 to 3 months for ecommerce with repeat buyers, 6 to 12 months for B2B with high retention.
- Multiply margin by payback months. That product is your hard allowable CAC ceiling per channel.
- Write a kill rule now: if a channel's CAC exceeds allowable for 14 straight days at meaningful spend, you pause it.
Step 2: Fix tracking, or you are flying blind in 2026
Attribution got harder, not easier. Browser-side pixels miss a growing share of conversions, and as of June 15, 2026 a single Consent Mode parameter, ad_storage, controls whether GA4 data flows to Google Ads. When ad_storage is denied, Google tags stop reading and writing the advertising cookies (including the _gcl_au cookie) that power Google Ads conversion tracking. Translation: your default setup now undercounts.
- Audit Consent Mode: confirm ad_storage, analytics_storage, ad_user_data, and ad_personalization are all explicitly set, and verify them with Google Tag Assistant on load and on consent change.
- Stand up server-side tagging for one platform first (GA4 or your main ad network's Conversions API) so you capture events the browser drops.
- Pass consent state from your banner through to the server container, so denied users are modeled, not silently lost.
- Verify server events against a pre-deadline baseline before you expand to the next platform.
- Report contribution-margin-adjusted return, not raw ROAS, so a low-margin SKU sale does not flatter the dashboard.
Step 3: Pick two channels, not seven
New marketers spread thin across every platform and learn nothing from any of them. Concentrate. Choose your primary paid channel by where buyer intent already lives. High-search-intent product where people actively look for a solution: lead with Google. Visual, impulse, or discovery-led product: lead with Meta. A sane starting split for an ecommerce or DTC launch is roughly 60% Meta, 40% Google, then let the numbers move it.
Then run one durable owned channel alongside paid: email and SMS. Paid rents attention. Email owns it, and it is where retention and repeat revenue compound. Build a welcome flow, an abandoned-checkout flow, and a post-purchase flow before you scale spend, because acquiring a customer you cannot retain just leaks margin. For the wider playbook see how to improve your growth marketing strategy and how email marketing earns its keep.
Step 4: Launch paid the way the algorithms now work
Targeting is no longer the lever. In 2026 Meta's Andromeda system reads your creative to predict the audience, and detailed interests now act as soft suggestions rather than hard filters. Google's Performance Max and broad match plus AI work the same way. Your job moved from picking audiences to producing creative volume and protecting the conversion data feeding the machine. Here is how to set it up.
- Consolidate into 2 to 3 well-funded ad sets or campaigns. Fragmenting budget across many tiny audiences starves the algorithm of learning data.
- Run broad or Advantage+ audiences. Stop stacking interests; let the creative do the targeting.
- Ship at least 5 genuinely different creative angles, ideally 15-plus assets: a UGC clip, a product demo, a testimonial, a text-led explainer, a lifestyle shot.
- Set the conversion event to your real money event (purchase or qualified lead), and fund each campaign to clear at least 50 conversions per week so the model can exit learning.
- Refresh creative every 2 to 3 weeks. If frequency climbs past 3 and performance dips, treat it as creative fatigue, not audience exhaustion.
Treat creative as a portfolio, not a bet. You are not trying to guess the winner up front. You test enough concepts that the winner reveals itself, then pour budget into it. That is why creative volume now beats clever targeting.
You do not start digital marketing by choosing channels. You start by knowing what a customer is worth, then refusing to pay more.ADGY
Step 5: Send traffic to a page built to convert
The fastest way to waste ad spend is a great ad pointed at a mediocre page. Before you raise budget, the destination has to do its job. Removing friction usually moves conversion rate more than any bidding tweak. Run this checklist on every paid landing page before you scale.
- Message match: the landing page headline echoes the exact ad that drove the click. No bait and switch.
- One primary action per page: remove competing links and the top navigation on paid landing pages.
- Offer above the fold: state the offer and the next step before the user has to scroll.
- Proof near the decision: reviews, results, logos, guarantees. See how to use social proof.
- Speed and clarity: cut load time and copy until the next step is obvious. More in landing page optimization.
- Mobile first: most paid clicks are mobile, so design and test there before desktop.
Step 6: Run a weekly cadence and let it compound
Digital marketing is not a launch. It is a loop. The brands that win review the same numbers every week and act on them, while everyone else stares at dashboards and changes nothing. Keep the loop boring and consistent. Here is the exact weekly cadence.
- Monday: pull spend, CAC, contribution-margin return, and email revenue for the prior 7 days.
- Compare CAC against your allowable ceiling. Above it at meaningful spend? Pause or fix. Do not hope.
- Cut the bottom creatives, scale the top ones, and queue the next batch to test.
- Check that one retention flow is improving: welcome, checkout recovery, or post-purchase.
- Write down one decision and one number you are trying to move next week. One, not ten.
That is the whole engine. Set your economics, fix tracking, concentrate channels, feed the algorithm creative, convert on the page, and run the loop. When you want this built and run on your numbers, not handed a list of tactics, talk to ADGY.
Frequently asked questions
What is the very first thing I should do to start digital marketing?
Math, not channels. Calculate contribution margin per customer, pick a payback tolerance in months, and multiply them to get your allowable CAC. That ceiling decides what you can afford to spend to acquire one customer on any channel. Without it you are just buying traffic and hoping.
How much budget do I need to start?
Enough to clear about 50 conversions per week per campaign so the algorithm can exit its learning phase. Work backwards from your target CAC: if it is 50 euros, that is roughly 2,500 euros a week per platform for clean signal. Below that, results stay noisy and you learn little. Fund one channel properly rather than two channels starved.
Do I still need to pick audiences and interests on Meta and Google?
Mostly no. In 2026 Meta's Andromeda system reads your creative to find the audience, and detailed interests act as soft suggestions rather than hard filters. Run broad or Advantage+, consolidate budget into 2 to 3 campaigns, and put your effort into producing 15-plus distinct creatives. Creative volume is the lever now, not targeting.
How do I know if my digital marketing is actually working?
Track contribution-margin-adjusted return and CAC payback, not raw ROAS or clicks. A campaign can post a strong ROAS while losing money on low-margin SKUs. If the profit from a customer repays acquisition cost inside your payback window, it works. If not, fix it or cut it.
Why does my tracking matter so much in 2026?
Because as of June 15, 2026 the ad_storage consent parameter alone controls whether GA4 data reaches Google Ads, and when it is denied the cookies behind conversion tracking stop firing. If you rely on default browser pixels you will undercount and make bad budget calls. Server-side tagging plus a correct Consent Mode setup is what keeps your numbers honest.