How to Optimize a Taboola Campaign: A Step-by-Step Framework

If you came to Taboola from Facebook and tried to test a campaign with $100, you already lost. Taboola needs far more money to gather data than social ads do, and treating it like a Facebook ad set is the fastest way to kill a profitable angle before it ever proves itself. This is the campaign-level optimization process we run on live Taboola accounts, step by step.

I'm Marcel Sattler, founder of native-advertising.net, and since 2015 my team has deployed more than $100M across Taboola, Outbrain, Newsbreak, MGID, Yahoo Native, Mediago, and RevContent for DTC, lead-gen, and affiliate clients. Taboola is one of the biggest native networks on the planet, and the optimization sequence below is exactly how we take a fresh campaign from broad launch to a tuned, profitable buy.

Why your Taboola campaign structure starts with device splits

Before you touch a single bid, the structure has to be right. We separate by device type from day one: desktop, mobile, and tablet each get their own campaign. Sometimes it makes sense to combine mobile and tablet into one campaign, but the default is three separate campaigns.

There's a reason for the split. Desktop skews to an older audience, so if your product targets people who are a bit older, desktop traffic is absolutely worth it. Mobile is where most of the cheapest traffic lives — every iPhone and Android phone out there. Tablet is the smallest slice of traffic, and while it can be very profitable when the right people are on iPads, the volume is just small.

One thing worth knowing if you're coming from social: the iOS 14 tracking changes that wrecked Facebook do not affect native ads. We're on iOS 16 now and native tracking still works the way it always did. That's a real advantage Taboola has over the social platforms.

If you're running DTC or dropshipping, the device split matters even more, because purchase behavior on desktop versus mobile can look completely different for the same offer.

How much budget you need to optimize a Taboola campaign

Here's the number people get wrong. For every campaign, start with at least $100. Three campaigns — desktop, mobile, tablet — means $100 on desktop, $100 on mobile, and $100 on tablet before you make a single decision.

You cannot test Taboola the way you test Facebook with $100 across the whole thing. You need way more money to test everything out, and the data only becomes actionable once you've actually spent. Underfund it and you'll be making optimization calls on noise.

Inside each campaign, we run three headlines and three images, which gives nine ads per campaign. Nine ads for desktop, nine for mobile, nine for tablet. That combination is what lets the first optimization pass tell you something real about which creative angle is working.

We also start completely broad. Taboola has targeting options and audiences, but we usually select none of them at launch. Broad is the best way to start, and it's the only way to scale later — a tight audience might work at first, but it caps out the moment you try to push volume. Start broad, then exclude as the data comes in.

Step 1: Optimize the ads after the first 100-150 in spend

Let the campaign run for one or two days. Once it has spent roughly 100 to 150 (EUR or USD in that range), you can make your first adjustments. Open the section showing your nine ads and look at how spend is distributed.

Say one ad spent $97, another spent $48, and a third spent $7, and your target CPA is $25. After a few days, if an ad has gathered real spend with no conversions, turn it off. The other campaigns and ads keep running.

This first pass is simple. You're looking at two things only: where you have good click-through rates, and where you're actually receiving conversions. That's the very beginning of the process — kill the dead creative, keep the rest alive.

If you're optimizing toward leads instead of purchases, the same logic applies on a lead-gen account: spend with zero conversions after a fair test gets cut.

Step 2: Optimize the site (publisher) section

Next, switch to the site section. This is the overview of every publisher site where your money went, each with its own CTR, conversions, and spend. You'll see, for example, site one with its CTR, a conversion, and exactly what you paid.

A word on CTR, because people misread it. A high CTR is usually good — but a CTR that's too high is a big problem. There's no universal number here. What counts as a good CTR depends on your country, your product, and your service. If you're unsure, ask your Taboola rep directly; they'll tell you the average for your country and your vertical. As a rule of thumb, a 9% CTR is way too high, and that placement should be turned off.

Now the bid. When a site is gathering conversions but they're too expensive — say your target CPA is $25 and a site is delivering three conversions at a $35 CPA — you have two moves.

Use Smart Bid, then override it at the site level

I recommend running the Smart Bid campaign type. The bid adjusts itself, but understand what it's optimizing for: it is not chasing the cheapest results, it's chasing the most conversions it can get from your budget. On a bad conversion day, the algorithm will still try to spend the whole budget while finding the most conversions it can for that money. It optimizes for conversion volume, not for budget efficiency.

That's why the site section is so powerful. Taboola gives you a plus and minus control on each site so you can override Smart Bid manually. For the site delivering a $35 CPA against a $25 target, I'd cut the bid by -30% for the first few days, then check back.

A few days later, go back into that site and read what the change did. Two outcomes matter:

• The CPA came down. Good — the adjustment worked.
• You're now getting zero traffic from that site. You pushed too hard. Instead of -30%, dial it back to about -20%.

This is how you overrule a generally excellent Smart Bid to optimize for your long-term results. For affiliate offers especially, that means optimizing your margin and your profit at the publisher level — not just chasing whatever the algorithm finds cheapest today. The site section is one of the most important parts of the whole Taboola workflow.

Step 3: Optimize by device and operating system

After sites, move to devices. Within a desktop campaign you'll typically see Mac and Windows broken out, along with the OS versions delivering conversions.

The pattern to watch for: Mac shows zero conversions while Windows shows 10, but the campaign keeps spending on Mac. When you see that, turn off Mac OS. You're not guessing — the data is telling you where the conversions live, so stop funding the OS that isn't producing.

Step 4: Optimize by browser

Browser is the next layer, and it's an underused one. You'll see Chrome, Safari, Edge, and the rest, each with its own CPA and CTR. On desktop campaigns especially, the spread can be dramatic.

Take a real example: Edge delivers 20 conversions at a $21 CPA, Chrome delivers 15 conversions at a $31 CPA, and Safari delivers zero. Taboola lets you optimize at the browser level, so here I'd turn Chrome off — or at least tighten it hard — because Edge is producing more conversions at a cheaper price. Safari, with zero conversions, comes off entirely. The browser breakdown alone can move a campaign from breakeven to profitable.

Going one step further with a tracker

Everything above is the basic optimization you can do inside the Taboola Backstage web interface, and it's a genuinely good overview for most campaigns. On very big campaigns, we go one step ahead and use our own tracker.

The tracker gives us far more data than Backstage shows. We can see which network and which mobile provider each user is on. Sometimes a clear structure appears — mobile provider A converts at a higher rate than mobile provider B — and we switch off the weaker one. There's a lot of optimization potential the moment you can see beyond the platform's native reporting.

The rule of thumb: as you spend more, it always makes sense to layer the tracker on top of Backstage to find what else you can exclude and where you can push results higher.

Watch the full breakdown

Is your account a fit for the same play?

This process scales from a first $300 test across three device campaigns up to the largest buys we run with a tracker bolted on. If your Taboola campaigns are spending without a clear cut-down sequence — or you're still testing on social-sized budgets and wondering why nothing converges — the structure above is where to start.

If you'd rather have a team run it for you, book a strategy call and we'll look at whether your account fits this exact play. You can also browse our case studies to see the results, or work through the rest of the video library and resources for the deeper optimization moves on Taboola and beyond.