How to Set Up a Taboola Campaign From Scratch (2026)

Most people open their first Taboola campaign builder, pack the US, Canada, and Australia into one campaign because they all speak English, point the ad at a Shopify product page, set a $30 daily budget, and wonder three days later why nothing converted. Every one of those decisions is wrong, and I can fix all four before you ever hit submit.

This is the setup, screen by screen. Not the strategy lecture, not the optimization playbook. The actual build: which fields to touch, which to leave alone, and the specific numbers that separate a campaign Taboola can scale to $50,000 a day from one that quietly bleeds out in 72 hours.

I'm Marcel Sattler. I founded native-advertising.net, and since 2015 I've deployed more than $100M across Taboola, Outbrain, Newsbreak, MGID, Yahoo Native, Mediago, and RevContent for DTC, lead-gen, and affiliate operators. The build below is the exact one my team runs on a fresh account.

What to prepare before you open the campaign builder

Three things have to exist before you touch the "create campaign" button, and skipping them is the most common reason a first launch fails.

First, fund the account. You need a network account or at least an account with a rep, not a self-service account. With those, there's a top-up or credit line: you load money before you can spend it, and that balance is the first thing you see when you log into the backend. If all you have is self-service, stop and get a real account first. Reach out to Taboola or book a strategy call with us, because the self-service path costs you the controls you'll need within the first week.

Second, get tracking firing. There's a tracking tab in the Taboola backend where you set your pixels, and the setup changes completely depending on infrastructure. A Shopify store tracks differently than an arbitrage funnel running through Tonic. Test it, watch the events fire, and use the Taboola Pixel Helper Chrome extension to confirm the pixels load in the right order. We cover the deeper version of this on the tracking setup guide.

Third, prepare your ads in advance. My recommendation is three headlines and three images, which Taboola combines into nine ads. Do not write headlines inside the builder on launch night. Marketers who improvise their creatives in the campaign screen consistently underperform the ones who walk in with all nine ready to upload.

How the account structure actually works

Taboola is a traffic source like Meta, TikTok, or Instagram, and like every source it has its own behavior. The single biggest structural difference: you run a few very large campaigns, not hundreds of tiny ones.

There's almost no ad fatigue on Taboola. Where someone on TikTok told me it's nearly impossible to run one video at high spend for a full year, on Taboola a campaign can live and stay profitable for a very long time. That low fatigue is why you don't fragment into dozens of $5/day campaigns. You build a handful of strong ones and feed them.

How strong? Budgets of $50,000 to $100,000 per day on a single campaign are doable here, and doable profitably. One campaign at $50,000 a day is a normal target, not a stunt. You can run the whole operation on one, two, or three big campaigns spending $50k, $60k, $70k, $80k a day rather than spreading yourself across a hundred small ones.

That structure only works if the funnel underneath it is a full funnel. Don't send the ad straight to a Shopify product page. Put a landing page, an advertorial, in between to warm a completely cold reader from the click into a buyer. The full-funnel structure is what carries the spend; the campaign settings just point traffic at it.

Split every campaign by country and device

Here are the two rules people break on day one. Both come down to one variable: the bid.

The bid is the CPC, the price you're willing to pay for a single click, and on Taboola you set it manually. Different countries carry different bids, so packing the US, Canada, and Australia into one campaign because they're all English-speaking wrecks you. Take the extreme version: the US versus India. Set the bid high enough for the US and it's insane for India; set it low enough for India and you never win an auction in the US. One campaign per country, every time. The US gets its own, Canada gets its own, Australia gets its own.

Device is the same story. Desktop carries a higher bid than mobile in general, so desktop never shares a campaign with mobile. You can combine mobile and tablet together, but keep desktop separate. If you select all three device types in the builder, Taboola itself throws a warning notification at you.

• One country per campaign, no exceptions, because bids vary by geo.
• Mobile and tablet can share a campaign; desktop always gets its own.
• Splitting this way is also how you scale cleanly, because each campaign has one clean lever.

For more on how device splits play out in the data, see the device targeting breakdown. The discipline here is what makes the Taboola agency work scalable instead of chaotic.

Build the campaign: naming, brand, objective, schedule

Now the builder itself. Log in, switch to campaigns, create a new one.

Name it with a structure you can read at a glance. Mine is date, then country, then device, then product or a label, for example a January 2nd campaign for US mobile gets the date, "US," "mobile," then the product. The brand name field matters more than people expect, because it shows in your ads. If you run an off-brand or third-party pre-lander, put the pre-lander's brand here. If my health offer's editorial is "doctor's recommendation," I type that, not "Marcel's knee sleeves."

Set the marketing objective honestly, because this is what the algorithm optimizes toward. Online purchases for ecommerce, lead-gen for lead-gen, whatever your real goal is. Your conversion goal should be account default if you've set tracking up correctly, meaning your final conversion is marked as final and intermediate events like add-to-cart are not.

For scheduling, the best-practice daypart is 6 a.m. to roughly 11 p.m. or midnight, copied across every day of the week. Few people convert at 2 or 3 a.m.; plenty convert during a normal working day. It's situational, but it's the right call in eight of ten cases. Set the correct time zone and leave the status active.

Targeting and bidding: what to leave broad

Location first. Include the country, and for a US campaign include the whole country rather than a regional state, unless you have a specific reason to exclude one. You can exclude a state like Arizona if you don't want it, but the default on a fresh account is the entire country open.

Now the part people get wrong because they treat Taboola like Meta. Leave the optimization layers broad at launch.

• Connection type: go fully broad. Wi-Fi-only is a relic; revisit it only in optimization.
• Operating system and browser: leave open. Even if you have KPIs from Facebook or TikTok, native behaves differently, so don't pre-narrow.
• Audiences: do not use them at the start. People assume they work like Facebook audiences. They don't. They're okay-ish in the US and weak elsewhere. Keep audiences completely broad on a new account and bring them in later for optimization once you have real data.

The bidding strategy is the field people skip and shouldn't. Focus on maximizing conversions to push CPA, cost per acquisition, as low as possible. Don't set a target CPA on day one; that comes after you've collected data. Set a daily budget of at least $100, somewhere between $80 and $100 to start. Don't start at $30. And note the threshold: spend more than $250 a day and Taboola gives the campaign an additional push. For the deeper logic, see the bidding strategy guide.

If your account is older and the newer target-CPA option isn't enabled, use bid control and set a realistic manual CPC, not the sky-high default the builder sometimes pre-fills. Use the optimize setting, and add any UTM or tracker parameters you need here. We run Volume internally as our tracker, but any solid tracker works.

The click ID that decides whether your data is real

Tracking is where data-driven decisions live or die, and the single technical detail that breaks the most setups is the Taboola click ID.

When a reader clicks your ad, Taboola automatically appends its click ID, a unique click identifier, to your URL along with your UTM parameters. You don't add it manually; Taboola does it on click. The click ID passes automatically from the ad to the editorial. The problem starts after that.

You have to carry that click ID through the rest of the funnel by hand. In an ecommerce setup, it has to reach the Shopify product page and the Shopify thank-you page. That sounds trivial and usually isn't, because you're running a cross-domain setup: the editorial lives on one domain, the Shopify store on another, and unless you're on Shopify Plus the checkout sits on yet another URL. Lose the click ID across any of those hops and your conversions stop reporting correctly in the Taboola interface, which means every optimization decision after that is built on broken data. If the cross-domain handoff is fighting you, reach out before you scale spend on top of bad tracking.

Build the ads: variations, focal points, and the 60-character rule

On the ad panel you'll see options like the AI wizard, variations, one-by-one, and upload-as-feed. The AI wizard is new and I'm not convinced yet, so start with variations: paste your landing page URL, add your prepared media, and drop in your prepared headlines.

Set the focal point on every image. Hover the picture and click focal point, because Taboola serves different placements at different ratios, four-by-three, six-by-five, sixteen-by-nine, two-by-one. If you don't set the focal point manually, the platform crops your image wherever it wants across those placements.

Headlines have a hard practical limit. Keep at least one headline under 60 characters so it stays fully visible on every Taboola placement. A 70-character headline triggers a warning and may get cut off on some placements. It's not a reason to panic, but always have one clean sub-60-character headline in the mix.

Two CTR levers you should never skip: always fill in the description, and always use the CTA button. Both lift click-through rate, which is exactly what you want in the launch phase. Then submit. Approval normally takes one working day, and there's a support chat in the left corner of the interface if review stalls.

Expect a high failure rate in the launch phase

Before you judge the campaign, understand what the launch phase is. It's the very beginning, and it carries a high failure rate by design. Whether it's your first campaign or your fifth, most early ads won't win.

That's normal. I've run native for more than ten years and not every campaign I launch succeeds either. The job in the launch phase isn't to be right; it's to lose small and cut fast. Keep launching different angles, different ads, different campaigns to hold momentum.

The optimization discipline that follows is its own topic, but two moves start immediately. Keep three to five ads live at any given time, not 15 or 20; that's the rhythm that gives the algorithm room to optimize in your favor. And block publishers aggressively, cutting a bad placement straight from 100 to zero rather than nudging bids in the early days. There's more in the campaign optimization guide.

Watch the full breakdown

Where to go from here

The buttons aren't the hard part. The hard part is connecting every dot, the top-up account, the click ID across three domains, the country-and-device splits, the 3x3 ad structure, without a single miss, then holding your nerve through a launch phase that's supposed to fail before it works. Get one piece wrong and the other nine don't save you.

If you'd rather not learn that on your own budget, book a strategy call and we'll tell you whether your account and funnel are ready for native. You can see what scaled setups look like in our case studies, or read deeper on the Taboola agency page if you're running ecommerce or affiliate offers.