How to Run Ads: A Step-by-Step Plan for Business
A step-by-step plan for running ads: pick a platform, set the objective, build tracking, test, read results, and scale what works without wasting budget.
ADS Beast editorial teamPublished 9 min read
To run ads, pick one platform where your buyers already spend time, set a campaign objective that matches the action you want, build one campaign with a few ad sets, add creative and tracking, publish, and leave it alone long enough to collect data. Then scale the winners.
In short
- Running ads is a sequence: platform, objective, structure, creative, tracking, launch, reading data, scaling. Skipping a step usually costs money later.
- Start with one platform and one campaign. Two platforms at once split your attention and your data.
- Tracking comes before launch, not after. Without it you are buying impressions and guessing at results.
- Most campaigns need a quiet period after launch. Editing during it restarts the learning your platform is doing.
- Budget decisions come from your own cost per result, not from benchmarks you found online.
Step 1: Choose one platform where your buyers already are
The right platform is the one your customers already open without being asked. Go where demand exists instead of trying to create it from scratch.
If you sell to other businesses, LinkedIn is a reasonable first stop: minimum daily budget is 10 USD, minimum audience size is 300 people, and LinkedIn recommends audiences from 50,000. If your product is visual and your buyers are younger, TikTok or Meta make more sense. If people search for what you sell by name or by problem, Google Ads fits, and there is no fixed minimum budget there. Microsoft Advertising reaches people on Bing and Windows surfaces, and most ad reviews there finish within 48 hours.
Pick one. You can add a second channel once the first produces a cost per result you can live with.
Step 2: Set the objective to match the action you want
The objective tells the platform what to optimize for, and the platform will find people likely to do that thing. Choose wrong and you get cheap clicks that never turn into customers.
Meta offers six objectives: awareness, traffic, engagement, leads, app promotion, and sales. The old "Conversions" and "Messages" objectives are gone, so map your goal onto what exists. Google and TikTok use similar logic under different names.
A practical rule: if you want purchases, choose sales or conversions, not traffic. If you want phone calls or form fills, choose leads. Traffic is for when the visit itself is the goal, for example sending people to a page where they book a call themselves.
Step 3: Build the campaign structure before you touch creative
Structure decides how your budget gets distributed and how fast you learn. Keep it simple at the start.
- Create one campaign with the objective you chose.
- Add a small number of ad sets, each testing one variable: an audience, a placement, or a geography.
- Put two to four ads inside each ad set so the platform can rotate creative.
- Set the budget at the level that matches your structure. If you want the platform to distribute spend across ad sets, budget at the campaign level. If you want to control each audience separately, budget at the ad set level.
- Leave room for the platform's minimums. TikTok requires more than 50 USD per day at the campaign level and more than 20 USD per day per ad group. Meta's minimum varies by country, currency, objective, and payment method, and Ads Manager warns you when your budget falls below it.
One campaign with three ad sets teaches you more than three campaigns with one ad set each, because the platform can move budget toward what works.
Step 4: Write creative that matches the platform, not your brand deck
Ad creative is the part you control most directly, and usually the part that decides the result.
On TikTok, vertical video that looks like something a person filmed performs better than a repurposed TV spot, because the feed rewards creative that holds attention. On Meta, check your text against Meta's ad policies before submitting; most ads are reviewed within 24 hours, sometimes longer. On Google, most ads are reviewed within one business day. On Microsoft Advertising, most reviews finish within 48 hours.
Keep claims specific and verifiable. If you cannot prove a number in the ad, cut the number. Before you send anything to review, run the text past the platform's rules yourself; a rejected ad costs you a day and teaches you nothing about your audience.
Step 5: Set up tracking before the first dollar is spent
Tracking is what turns a platform from a bill into a feedback loop. If purchases, form fills, or add-to-carts are not recorded, you cannot tell which ad set produced them.
On a store, install the platform's tracking so purchases and add-to-carts are recorded, then connect your product catalog if the ad platform supports shopping formats. For Meta specifically, the pixel and the Conversions API deduplicate by matching event name and event_id; there is no toggle to switch on. In Google Analytics 4, event data is retained for either 2 or 14 months, IP addresses are not stored, and attribution models are last click and data-driven. Google Ads counts conversions within 30 days of a click by default, and that window is configurable from 1 to 90 days.
Decide what counts as a result before launch. "Purchase" and "lead" are results. "Click" is a step on the way, not a result.
Step 6: Launch and leave it alone long enough to learn
The most common way to waste an ad budget is to keep editing the campaign. Every significant change resets the learning the platform has done.
On Meta, an ad set exits the learning phase after roughly 50 results in a week following your last significant edit. That number is a guide, not a finish line, but the principle holds everywhere: the platform needs a stable campaign to find your buyers. Set a review date before you launch, and during that window only intervene if something is broken, such as a disapproved ad or a broken link.
TikTok rewards attention in the first days of delivery, so watch early performance rather than changing settings hourly. Hitting pause and restart is also an edit.
Step 7: Read the results and decide what to do next
Judge the campaign on the action you set as the objective, not on clicks or impressions.
| What you see | What it usually means | What to do |
|---|---|---|
| Spend with no results | Audience, offer, or tracking is wrong | Check tracking first, then the offer |
| Results but cost per result too high | Creative or audience is not matching | Test new creative before changing audience |
| One ad set far ahead of the rest | The platform found a working combination | Move budget to it, pause the weakest |
| Results only at high spend | You are buying the expensive part of the audience | Test narrower targeting or a different placement |
| Results on day one, nothing after | Creative fatigue | Rotate in fresh creative |
Your break-even cost per result comes from your own numbers: price, margin, and how often a lead becomes a customer. If you do not know those, calculate them before you scale, not after.
Step 8: Scale what works instead of rebuilding everything
Scaling means adding budget to the campaign or ad set that produced results, not creating a new campaign that starts from zero.
Increase budget in steps and watch whether cost per result holds. If it climbs, you have hit the point where the platform has to reach less interested people. When that happens, expand the audience or add a new creative angle rather than pushing the same campaign harder. Once one platform is stable, add the second channel, and give it its own learning period.
If you want the setup, checks, and reporting handled in one place while you keep control of the budget and the launch decision, AI campaign setup covers how a campaign gets prepared, paused, and reviewed before it goes live.
Platform differences worth knowing before you start
The mechanics are similar across platforms. The rules, minimums, and review times are not, and those decide how you plan your first weeks.
| Platform | Budget rule | Review time | Notes |
|---|---|---|---|
| Meta | Minimum varies by country, currency, objective, and payment method | Most within 24 hours, sometimes longer | Six objectives; learning phase around 50 results per week |
| Google Ads | No fixed minimum | Most within one business day | Conversion window 30 days by default, 1 to 90 configurable |
| 10 USD per day minimum | Not published in the same terms | Audience minimum 300 people, 50,000 recommended | |
| TikTok | More than 50 USD per day per campaign, more than 20 per ad group | Not published in the same terms | Rewards creative that holds attention |
| Microsoft Advertising | Not published in the same terms | Most within 48 hours | Reaches Bing and Windows surfaces |
Two more things that trip people up. Google's Local Services Ads charge per lead and the money-back guarantee has been retired; the badge is now called Google Verified. And Google Ads has had no "Local" campaign type since 2022, so ads on Maps run through search with addresses or through Performance Max.
How to run ads for free
There is no way to buy reach for nothing, because platforms charge for delivery. What you can do without a media budget is use free distribution: post organically, ask customers to share, list your business in free directories, and publish content that ranks in search.
If you want paid reach later, start with the smallest budget the platform accepts and let the results tell you whether to add more. That first small spend is also the cheapest way to learn how the platform's auction treats your audience.
Common mistakes and how to spot them early
Mistakes in paid ads are usually structural, not creative. They show up in the numbers before they show up in your bank account.
- Editing the campaign every day. If your edit history is longer than your results history, you are the problem, not the audience.
- Running two platforms with one budget. Each platform gets too little data to optimize.
- Sending all traffic to the homepage. Send it to the page that matches the ad.
- Judging on clicks. Clicks are cheap on bad placements and expensive where buyers are.
- Skipping the policy check. A disapproved ad stops delivery and tells you nothing about your audience.
Your next step
Write down your break-even cost per result, then launch one campaign on one platform with tracking in place and a review date on the calendar. If you want the campaign prepared and checked before you approve the launch, see how campaigns are built and audited with AI.