Skip to content

Native advertising beyond the big two: Taboola, Outbrain and MediaGo

Not search, not social: a reader, an article, and a headline underneath it.

Krög editorial teamAccount managers, compliance and billing

5 min read

Key takeaways

  • Native ads appear in recommendation units on publisher sites, where people are reading rather than searching or scrolling a feed.
  • You bid mostly per click and manage results by publisher site. Site-level data is the main optimization lever.
  • Headlines and images must match the landing page, and article-style pages must be labeled as advertising.
  • Budget for a longer test than on social. Native needs volume per site before it says anything.

When media buyers talk about the big two, they mean Meta and Google. Native advertising is the channel many accounts never try: the 'Recommended for you' and 'Around the web' units under articles on news and publisher sites. Taboola and Outbrain run the two largest of these networks (Outbrain combined with Teads in 2025), and MediaGo is a newer network with its own publisher supply. The mechanics differ enough from social and search that they are worth learning before you spend.

How the placement changes the ad

On social platforms, people scroll a feed of friends and creators. On search, they type what they want. On native, they have just finished reading something. That changes what works: a headline that earns curiosity without overpromising, an image that looks at home on a publisher page, and a landing page that continues the story the headline started.

Intent

Search
Stated in the query
Social
Inferred from behavior
Native
Inferred from what they are reading

Usual bidding

Search
Per click
Social
Per impression or per result
Native
Mostly per click, with conversion bidding available

Main lever

Search
Keywords
Social
Audience and creative
Native
Publisher sites and headlines

Landing page

Search
Product or service page
Social
Product page or short form
Native
Often an article-style page, labeled as advertising

The networks, briefly

Taboola

One of the two largest recommendation networks, with placements across many major news and lifestyle sites. Campaigns are built from items (a headline, a thumbnail and a URL), bid per click or with conversion-based bidding, and optimized by site.

Outbrain

The other large network. The model is similar: headline-and-image units on publisher sites, per-click and conversion bidding, and reporting by publisher. Many buyers run both, because each network has publishers the other does not.

MediaGo

A newer native network with its own publisher supply and a similar buying model. It is worth testing once you have a working native funnel on one of the larger networks, as a way to add reach without changing the creative approach.

Site-level optimization is the whole job

Native networks report performance by publisher site, and often by placement within a site. The same headline can convert well on one site and waste budget on another, so most of the work is reading that report and acting on it.

  1. Launch broad, at a bid that gets delivery, with three to five headline and image combinations.
  2. Wait until each site has spent enough to judge. A common rule is one to two times your target cost per conversion.
  3. Block sites that pass that threshold without converting.
  4. Raise bids on sites converting below target, and lower them where results are marginal.
  5. Repeat every week. The site list is where native accounts are won or lost.
1–2×Your target cost per conversion, spent on a site before you judge it

What native networks review

Native networks review every item before it runs, and they pay close attention to the relationship between the headline and the page.

  • Headlines must be accurate. Teasers that withhold the point and claims the page does not support are rejected.
  • Images must relate to the offer. Sensational or misleading images are rejected.
  • No implied endorsement from public figures who have not given one.
  • The landing page must deliver what the headline says, and restricted categories follow each network's own rules and, where required, certification.

Headlines that pass review and still get read

The craft of native is writing a headline that is specific enough to pass review and interesting enough to be clicked by someone who was not looking for you. Most rejected headlines fail in one of a few predictable ways.

Pattern

'You won't believe what happened next'

Why it fails
Withholds the point; reviewed as clickbait
What works instead
A specific promise the page keeps: 'What a 30-day energy audit finds in a typical home'

Pattern

'Ends tonight!' with no real end date

Why it fails
Invented urgency is misleading
What works instead
A real date, or no deadline at all

Pattern

A celebrity photo beside the offer

Why it fails
Implies an endorsement that was never given
What works instead
The product, the team, or neutral imagery of the topic

Pattern

Shock or before-and-after images

Why it fails
Sensational, and restricted in health topics
What works instead
An image that shows the product or the subject plainly

The presell page

Many native funnels use two pages: an article-style presell page that explains the problem and the product, then the offer page where the visitor buys or signs up. It works because the reader arrived in a reading mood. It stays compliant when the presell page is honest about what it is.

  • Label it as advertising near the top, and name the business behind it.
  • Keep every claim the headline makes, and support each one on the page.
  • Host both pages on your own domain, so the reader, the reviewer and the ad's display URL all point to the same business.
  • Make the step to the offer page obvious, with the price and terms shown before payment.

Budget and a fair test

Native clicks are usually cheaper than search clicks and convert at a lower rate, because the reader was not looking for you. That makes testing slower, not cheaper: you need enough clicks per site to read it. Plan a test budget that lets your top sites each reach the stopping threshold, and give it two to four weeks.

  • Track conversions with each network's pixel or server-side conversion API, and reconcile them with your own analytics.
  • Put the network's site and campaign macros in your tracking URLs, so you can read results by site in your own reports.
  • Refresh headlines and images on a schedule. Native fatigue shows up as a falling click-through rate.
On native, you do not choose an audience. You choose where you are read.

Written by

Krög editorial team

Account managers, compliance and billing

Written by the people at Krög who open, fund and support advertising accounts for verified businesses every day. Platform rules change often. If something here is out of date, tell us on Telegram.

Meet the team
  • Native
  • Taboola
  • Outbrain
  • MediaGo

Share this article

Platform rules change often. Check each platform's own help center for the current requirements, and tell us on Telegram if something here is out of date.

Keep reading

More from the account side.

Keep the conversation going

  • Krög AI

    Questions about this article?

    Krög AI is the assistant on this site. It answers common questions about our services, prices, verification and the platforms we support, based on what we publish here. For anything it can't answer, the team is one message away on Telegram.

    TryHow does native advertising on Taboola and Outbrain work?

    How Krög AI works
  • Telegram channel

    New articles as they're published.

    The channel carries new articles, promo codes and notes on platform policy changes, posted when they go up here. For questions, message the team directly on Telegram.

—  From the team that writes this

Accounts, handled.

We open advertising accounts for verified businesses, fund them from your prepaid balance and support them on Telegram. You keep full control of campaigns, creative and budget.

  • Exact rate confirmed in writing before you pay
  • Usually ready within 48 hours of business verification
  • A dedicated manager on Telegram