Zero to Profit: Optimize a New Native Campaign on Taboola

A brand-new Taboola or Outbrain account in 2023 starts with zero conversions, no audience history, and no smart algorithm doing the work for you. Facebook and Google will optimize toward a purchase on your behalf. Native does not. You optimize by hand, layer by layer, in a specific order — and if you do it out of order you burn budget chasing the wrong number.

This is the exact sequence we use to take an account from scratch to its first profit. It is not the same play as fixing a campaign that already has data. Here there is no data. You build the data, then you tighten the funnel one layer at a time, and only at the very end do you flip the campaign type from collecting data to defending margin.

Why a new native account can't optimize for the final conversion

Marcel Sattler, founder of native-advertising.net, has deployed more than $100M across Taboola, Outbrain, Newsbreak, MGID, Yahoo Native, Mediago, and RevContent since 2015, and the first thing he tells anyone launching a fresh account is this: you cannot optimize on the final conversion on day one.

A new account has zero conversions in it. The pixel is empty. You have no specific targeting to lean on, so in the beginning you are effectively shooting at everyone. Asking the platform to optimize toward a lead or a sale when it has never seen one is like asking a stranger to guess your favorite meal — there is nothing to learn from.

So you optimize for the steps before the final conversion. Marcel has run media since the early Facebook days more than a decade ago, when even Facebook couldn't optimize directly on the purchase and you fed it add-to-cart events first. Native is the same idea, just done manually. You climb the funnel from the top: click-through rate first, then everything downstream of it.

One structural note before the layers: one product rarely means one campaign. One product across mobile/tablet and desktop, in the US and Canada, is already four campaigns — and each one gets optimized on its own. That is normal and recommended. See our Taboola agency page and Outbrain agency page for how we structure this at scale.

Layer 1: optimize the ad CTR with 9 creatives per campaign

The first lever is the click-through rate of the ads themselves — how many people who see the ad actually click it. This is the single most important number in the whole account, because the right image paired with the right headline acts as a multiplier on everything below it.

To find that pair fast, build each new campaign with 3 images and 3 headlines. Taboola and Outbrain combine them for you, so 3 × 3 gives you 9 creatives out of the gate. That is enough variety to surface a clear winner without splitting your budget into dust.

There is no universal "good" CTR I can hand you, and anyone who quotes one is guessing. The number swings with country, niche, safety category, device, and whether you run a whitelist. The same creatives can post very different CTRs on desktop versus mobile versus tablet. High CTR from push traffic, for example, is usually bad CTR — clicks that never convert.

The practical move: ask your Taboola or Outbrain account rep what the average CTR looks like for what you sell. Now you have a benchmark. Underperforming it means your ads are the problem — rework the copy, swap stocky images, sharpen soft headlines, fix a headline that promises the wrong thing. Our native ad headlines blueprint goes deeper on the copy side.

Layer 2: split-test 3 advertorials at 33.33% each

Once the ad earns the click, the advertorial has to hold it. In the beginning, run 3 advertorials per campaign, each taking 33.33% of the traffic, split-tested in the background.

These are not three versions of the same page with a swapped word. Make them genuinely different bets:

• An aggressive angle, a soft angle, a neutral one
• A short advertorial of roughly 600 words
• A medium one
• A long one of roughly 1,000 words

You are watching which advertorial earns the better click-through and keeps the conversion behind it. A high advertorial CTR alone is a trap: if the page is too pushy or makes a false promise, people click through and nothing converts. You want the balance — strong click-through and a real conversion rate behind it.

For a baseline, a good advertorial lands around a 12% to 18% click-through at higher spend. Two or three percent usually means the advertorial is weak. As soon as one advertorial clearly wins, kill the other two and concentrate traffic on the winner — less to test, profit sooner. The high-converting advertorial formula covers the structure we use.

Layer 3: prune sites and sections — start broad, narrow down

A fresh campaign serves across thousands of publisher sites and sections. This site-level control is unique to native — Facebook and Google don't hand you a placement list like this — and it is where a lot of the manual work lives.

The rule is: start broad, then narrow down. Don't pre-judge placements from your armchair. We have seen the weirdest sites convert and obvious ones flop, so let the data decide which sites stay.

The mechanics differ by platform:

• Outbrain: turn losing sections off.
• Taboola: turn them off, or boost and de-boost them with different bids per section.

Use a whitelist or block list to keep the campaign on premium publishers and off low-quality push traffic if you want to avoid it. And cut on relevance, not just numbers — a men's product running on a women's magazine rarely makes sense. More on this in our piece on whitelists and blacklists.

Layer 4: devices, browsers, and OS — only with enough spend

Below sites comes the device layer, and you can slice it finely: by operating system (run iOS only, or Android only on mobile) and even by browser. It sounds picky, but the differences are real — combinations like Android + Chrome or iOS + Safari often outperform, and you switch off the device, OS, and browser combinations posting higher CPAs.

The same applies to geo. Run one country per campaign, not several stuffed together, because countries carry different bids and different traffic strength. Inside a country you can also cut weaker regions and concentrate on the profitable ones.

A hard gate on this layer: don't optimize devices, browsers, and regions on a $100–$200 campaign. There isn't enough data to trust the splits. Wait until you've spent something like $1K–$2K on that campaign, then the numbers are real enough to act on. Our device targeting guide breaks down the mobile-versus-desktop reads.

Layer 5: bids — last, and optional

Bids come dead last, and they are optional. Yes, you can lower a bid, watch how much reach you keep and how many auctions you still win, and shave it down a cent at a time.

But fighting for the last two cents on a bid is the worst use of your attention. The real gains are upstream — improving the CTR of the ads and the advertorials moves the campaign far more than scraping pennies off the bid. Work the bid only after the layers above are dialed in. See our bidding strategy breakdown before you touch this.

The campaign-type switch: max-conversion to target-CPA

Here is the move that separates "collecting data" from "making money," and most of the related campaigns on this site don't cover it the way this one does.

You launch a new account on a max-conversion campaign — its job is to gather as many conversions as possible and fill that empty pixel. Once you have the data, you switch to a target-CPA campaign: you tell Outbrain or Taboola the price you'll pay per conversion, and it buys as many conversions as it can at or under that price.

Two rules make this switch work:

1. Have the data first. Run max-conversion until the pixel is fed, then switch. A target-CPA campaign with nothing behind it has nothing to optimize against.
2. Give the algorithm room. Set a target the account has actually hit before. If your real CPA has been $40, don't type in $20 — it won't work. Set $40 and let it work. Typing $10 when history says $50 just stalls the campaign.

This is the full optimization of a from-scratch native campaign in a nutshell: feed the pixel on max-conversion, climb the funnel layer by layer, then defend margin with target-CPA. Our zero-conversions diagnosis is a useful companion if your pixel still isn't filling.

Watch the full breakdown

Is your new account a fit for the same play?

If you are staring at a fresh Taboola or Outbrain account with zero conversions and a budget you don't want to torch on guesswork, the order above is what gets you to first profit — ads, advertorials, sites, devices, geo, bids, then the target-CPA switch. The expensive mistake is optimizing the bottom of the funnel before the top has data.

Book a strategy call and we'll map your product across the right campaign splits and tell you which layer to attack first. If you're running DTC or dropshipping, lead-gen, or affiliate offers, the layers stay the same — only the targets move.