How we plan and run a campaign

The same four steps for our own brands and for brand partners. Planning happens once for each product and market. Launches follow a weekly rhythm, and measuring and optimizing happen every working day. None of it is clever, but each decision is made against a number written down in advance, and each change carries the name of the person who made it.

Send a campaign brief See the services

The process in four steps

Each step ends with something written down, and the next step starts from it: a plan, a tracking setup that has been tested, a morning note, a line in the change log.

Plan
We review the product, the landing page and the unit economics, then write down the target cost per result, the stop-loss threshold and the markets. No money is spent before that document exists.
Launch
Campaigns follow one naming convention on TikTok and Meta. Tracking is tested end to end on a phone, and ads go into review a day before the start date.
Measure
Each morning we read yesterday's figures in each account's own time zone and check them against the brand's records, where the brand shares them daily.
Optimize
Budgets, bids and ads change only against thresholds agreed in advance, and only by the person who owns the account.

What each step involves

The steps above are the short version. This is what they look like in practice, with the numbers we use.

Plan

Planning starts with the product and the page it will be sold on. Where it is practical we order the product ourselves, go through the checkout or the form on a phone, and read the terms a customer would read. Anything that would fail a platform review or disappoint a buyer is raised at this point.

Then the unit economics. Take a product sold at 60 euros with a 55 percent gross margin and 8 percent of orders returned: a sale can cost about 30 euros in advertising before it starts losing money. The target cost per result sits well under that line, partly because a return also costs shipping and handling that the margin figure does not show. The stop-loss threshold, meaning the spend at which a test is paused if it has produced nothing, goes into the same document.

Markets come last. They are picked from the countries the brand can ship to or serve, and a first test rarely covers more than two.

Launch

The structure mirrors the plan: one campaign per market and optimization event, with ad groups on TikTok and ad sets on Meta underneath. Campaigns, ad groups and ad sets follow the pattern market_product_angle_date. An ad group called DE_desklamp_homeoffice_20261012 tells anyone on the team where it runs, what it sells, which idea it tests and when it started. The same names are carried into the UTM parameters (utm_source, utm_medium, utm_campaign, utm_content), so a visit in the brand's own analytics can be matched to the ad behind it.

Before anything spends money, tracking is tested end to end. Someone on the team places a test order or submits a test form on a phone, then checks that the event reaches TikTok and Meta twice, from the pixel and from the server, under one event ID so it is counted once.

Ads go into platform review at least a day before the planned start. A rejection then costs a day of fixing, not a day of delivery.

Measure

The morning review covers the previous day as each ad account defines it. This sounds pedantic until it goes wrong: an account set to UTC-5 closes its day at 06:00 CET, one set to UTC+1 at midnight, and merging them quietly moves Tuesday's sales into Monday.

Attribution windows are fixed per account and printed on every report. Our default on both TikTok and Meta is a 7-day click window, with 1-day view conversions shown in their own column where the platform allows it. Changing a window changes the numbers, so it only happens with the brand's agreement.

Platform figures will not match the store or the CRM, and nobody should expect them to. A platform usually credits a conversion to the day of the click or view; the store records the day of payment. TikTok and Meta can both claim the same order. Some conversions are modeled for people who declined tracking, and refunds never reach the platform at all. We compare against the brand's records daily where the brand shares its data that often, and at the monthly reconciliation in any case. The point is to find the gaps that these reasons do not explain.

Optimize

Only the media buyer who owns an account, or whoever covers for them that day, changes budgets and bids. Analysts and the creative team see every number but do not move money. Each change to a budget, bid or status is logged with the person, the time, and the old and new value.

New creative goes into each active campaign every one to two weeks, sooner on TikTok when frequency climbs. A test stops when it reaches its budget, its stop-loss or the end of its agreed period, whichever comes first. Most first tests miss the target. What we want to know is whether the gap is closing.

Rules are written before a test starts, each with an explicit threshold and the name of the person who set it. A typical set:

An ad group or ad set that has spent twice the target cost per result without a conversion is paused.
An ad whose cost per result is 50 percent over target is paused once it has spent three times the target cost per result.
After three days in a row at or under target, the budget goes up by about 20 percent.
Every campaign keeps a daily spend limit, so a broken rule cannot overspend.

A typical working week

Ads deliver around the clock. The decisions about them follow a weekly rhythm, which launches and public holidays sometimes shift.

How a normal week is organized
Day	What happens	Who
Every working day	Morning review of the previous day, account by account, in each account's own time zone. Anything past a stop-loss threshold is paused; anything unusual is flagged to the account owner.	Media buying, analytics
Monday	Weekly summary sent to each brand partner. The week's test plan is agreed: which hooks, pages or markets, and with how much budget.	Media buying, analytics and operations
Tuesday	When a creative round is due, the new ads go into review on both platforms. Landing page changes are published and their events tested again.	Creative, media buying
Wednesday	First look at the week's launches. Nothing is judged yet unless it has already hit a stop-loss.	Media buying
Thursday	Creative briefs for the following week, written from what the numbers show. Open reconciliation questions followed up with partners.	Creative, analytics and operations
Friday	Budgets and daily spend limits set for the weekend. No new tests after midday, so nobody discovers a broken pixel on Monday morning.	Media buying
First week of the month	Monthly reconciliation of spend and results with each brand partner, ending in the agreed result count for invoicing.	Analytics and operations

Principles we work by

Six rules for the moments when there is a choice to make. Most of them cost us some speed or some revenue, which is why they are written down.

Numbers someone else can check

A figure in our report should be one the brand can find in its own records, give or take a difference we have explained. If a number cannot be reconciled, no budget decision rests on it.

A name on every decision

Thresholds, budget changes and paused tests all have an owner. When a partner asks why spend dropped on Tuesday, one person can answer, and the change log says the same thing.

Policies are read before planning

For every product we read the TikTok and Meta policy pages for its category, and the local rules in each target market, before a single ad is planned. A result that depends on bending a rule ends in a restricted ad account, and the result goes with it.

The ad and the page make the same promise

If the ad says delivery in three days, the page says so too, near the top. No countdown that resets on every visit.

Small tests first

New products and new markets start with a test budget and a date on which we decide. Budgets grow after the numbers do.

Stop early, and say why

When a campaign is not working and we cannot see how to fix it, we stop and write down what we learned. A brand partner hears it from us first, in plain words, with the figures attached. For our own brands the note goes into the same file, because the next test starts from it.

Advertising standards

What we refuse to run, and how a brand and its landing page are checked before we agree to anything.

What we will not run

This applies to brand partners and to our own brands alike. We decline anything that involves:

Products or services that TikTok or Meta prohibit in the target market, or restrict unless the advertiser holds an authorization the brand does not have
Health, income or appearance claims the brand cannot prove
Fake testimonials, invented reviews, or actors presented as customers
Deceptive pricing, such as a subscription hidden in the small print or a discount from a price that never existed
Cloaking, or any destination different from the one the reviewer saw
Campaigns aimed at people under 18, or age-restricted products shown to anyone under the legal age in that market
Landing pages without a clear privacy notice and a way to contact the business

How a brand and its page are reviewed

Before an agreement is signed, a member of our analytics and operations team, which also handles compliance, goes through the brand, the product and every page an ad will link to. The review is written down and uses the same questions each time.

If an answer is no, the brand can fix it and we look again. If it cannot be fixed, we decline the work. The review is repeated whenever the product or the page changes. These are the questions:

Is the product allowed in each target market on both platforms, and does it need a license or an age restriction there?
Can the brand back every claim in writing?
Are the price, delivery terms, subscription terms and refund policy visible before payment?
Does the privacy notice cover the pixel and server events, and is there a consent banner where the market requires one?
Does the page show a business name, an address and a working contact?

When TikTok or Meta rejects an ad

Rejections happen to careful advertisers too, sometimes over a single word or frame. Our response follows the same order every time.

Read the policy reason the platform gives, and the policy page it points to, before changing anything.
Find the trigger. It may be a line of voice-over, a word on screen, an image, or something on the landing page, which gets checked as carefully as the ad.
Fix the ad or the page and submit it as a new version, with the reason noted in the account log.
Appeal only when we believe the decision is wrong, and explain why in the appeal. If the appeal is turned down, that is the answer.

We never try to get around review. No cloaking, no changing a destination after approval, no rewording a claim to slip it past a reviewer, and no fresh ad account to rerun something that was rejected.

Account-level restrictions are handled the same way. We stop spending in the affected account, read the reason in full, and bring the brand in if the product or the page is part of the problem. Repeated rejections on one product are a reason to review the product again, not to keep resubmitting.

Send us a product to plan around

Include the product, the landing page, the markets and the result you would pay for. You will hear from us within two business days, with a first view of the numbers or a clear no.

Get in touch