Banner traffic has a reputation problem. We check the common myths against viewability and invalid traffic standards, then lay out how to run a first display test without wasting budget.

Main points

  • The first banner ad in 1994 drew an estimated 44 percent click rate, and legitimate click rates sit around a hundredth of that, which measures lost novelty, not lost value.
  • Under the viewability standard, a display impression counts only once half the ad stays on screen for one unbroken second; video needs half the ad on screen for two seconds.
  • Invalid traffic splits into GIVT, filtered through routine list checks, and SIVT, which needs advanced analytics and is disclosed only at campaign total level.
  • Rising click volume with a normal cost per click and flat conversions is the classic pattern of invalid traffic, and it gets billed like real traffic.
  • A fair first display test needs a fixed cap, pre-set success criteria, several sizes and positions, and per-placement sub IDs for tracking.

Display traffic gets dismissed before it gets tested. Affiliates repeat the same objections, banner blindness, junk clicks, bots, and then spend their budgets on channels that feel safer. Some of those objections describe real risks, but the industry has measurement standards that address most of them, and the failures that remain usually come from testing display the wrong way. Here are the common beliefs, checked against what the standards actually say.

Myth 1: Banner Blindness Means Display Ads Cannot Convert

Banner blindness is real as a description of attention. People notice and remember less of anything they did not go looking for, and an ad that interrupts a task competes with that task. What the myth gets wrong is the conclusion. The first banner ad on the web ran in 1994 on hotwired.com for AT&T and drew an estimated 44 percent click rate, while legitimate click rates sit around a hundredth of that figure. That collapse measures novelty wearing off, not the death of the format. Every mature channel settles at lower engagement than its debut.

The practical mistake is judging display by click-through rate alone. Display is an interruption channel. Most impressions build recognition and only a small share produce clicks, so a low CTR tells you the format is behaving the way mature formats behave. The number that decides whether display can convert for you is cost per action against the offer payout, and you cannot calculate that without running impressions.

Myth 2: Cheap Clicks Automatically Mean Profit

Cheap is display’s headline selling point and its most common trap. A click is an expense, not income. Cheap clicks only pay when a real person lands on the page, understands the offer, and completes the action.

The invalid traffic framing makes the risk plain. Invalid traffic gives no warning before it shows up. It hides inside accounts whose numbers look fine on the surface: click volume rising, cost per click normal, conversions flat. That gap between activity and outcomes is usually where the waste sits, and every fake interaction gets charged at full price, exactly as though a genuine prospect had clicked.

Accidental clicks add to the problem. Google’s ad traffic quality team lists clicks caused by poor ad placement, such as ads positioned too close to navigation buttons, among the invalid traffic it filters. A user who meant to tap a menu button and hit your banner was never a prospect. Before celebrating a low cost per click, divide spend by conversions and compare that number to your payout. If the math fails, the cheap clicks were expensive.

Myth 3: Display Traffic Is the Lowest-Quality Source

Display is not a traffic source. It is a format, and the same banner can run on a major news site, a hobby blog, or a mobile game. Calling display low quality confuses the format with the worst publishers on it.

Platforms do enforce quality floors. Google’s ad traffic quality team pairs automated filters and machine learning with human investigations to find and remove invalid traffic, usually spotting it while it happens or soon after. Invalid interactions are stripped from campaign metrics and billing before the billing cycle closes, and when invalid traffic turns up after an invoice, credits are issued where appropriate and possible. The Media Rating Council has accredited those defenses for particular Google services and metrics, which means an independent auditor has examined how the filtering works.

What decides quality for your campaign is narrower: which placements you allow, which publishers appear on your buy, and how well the offer matches the audience. All three are under your control.

Myth 4: One Banner Size Is Enough to Judge a Source

A single creative in a single placement measures that placement, nothing more. Viewability depends heavily on position. Under the industry standard, a display impression only counts as viewable once half or more of the ad stays on screen, unbroken, for a full second. An ad below the fold can serve thousands of impressions and never qualify, while the same creative above the fold qualifies on most page loads.

Placement affects cost and quality at the same time. Google attributes a share of invalid clicks to poor placement, which means a badly positioned banner can inflate your click count and your suspicion of the channel at once. A fair verdict on display requires several sizes in several positions, judged on cost per action, before you decide the channel cannot work.

Myth 5: You Need a Big Budget to Test Display

Structure matters more than scale. The affiliates who burn money on display are the ones who spread spend across every placement, measure nothing, and quit when the total looks bad. A small test with clear rules produces better information than a large one without them.

Set a fixed cap you can afford to lose entirely, define what success means before spending, and change one variable at a time. If the test hits its cap without a single conversion at a workable cost per action, you have learned something real about that offer and that placement mix. Display also competes with other low-cost formats for the same test budgets, and the discipline that works for push or native applies here too. The difference is that display gives you placement-level controls most channels lack, so a losing pocket of traffic can be cut without abandoning the whole buy.

The Facts: What Viewability Means for Your Impressions

The Media Rating Council is an independent nonprofit created in the 1960s to oversee broadcast ratings. It now writes the measurement rules that decide how the industry tallies impressions, clicks and viewability across digital advertising, and vendors that measure impressions or viewability can apply for accreditation, which means an independent audit of their methods and reporting.

Under the MRC’s definitions, a valid impression reaches an actual person in a spot where it can actually be seen, with identified invalid traffic taken out of the count. For display, half or more of the ad has to stay on screen for one unbroken second. Video gets a longer window: half or more of the ad on screen for two unbroken seconds. The MRC retired its display advisory from November 2012, clearing the way for buying and selling on viewable impressions, and it set a waiting period for trading in-browser video viewability that ran through June 30, 2014.

Two cautions follow. First, viewable means an opportunity to see, not a guarantee that anyone looked. Second, vendors disagree at the margins. The MRC ran a reconciliation study across 11 accredited measurement vendors to identify the sources of those discrepancies, expects variances to narrow to plus or minus 5 to 10 percent, and required vendors to adapt their methods within 60 days. If two dashboards disagree on viewability by a few points, that falls inside the expected range rather than proving either one broken.

The Facts: Invalid Traffic and the GIVT and SIVT Split

The MRC began work on invalid traffic guidelines in 2014 and 2015, and the final addendum to the Invalid Traffic Detection and Filtration Guidelines, known as IVT 2.0, came out in June 2020. Every MRC-accredited organization has to follow it. The standard is careful not to treat the word fraud as a catch-all and splits invalid traffic into two categories that call for different responses.

GIVT SIVT
How it is caught Routine filtration from lists and standardized parameter checks Advanced analytics, multi-point corroboration and significant human intervention
Typical examples Known bots, spiders and crawlers, data-center traffic, pre-fetch activity, 0x0 and 1×1 ad sizes, sessions that cannot render images Taken-over devices, paid clicking operations, automated traffic that imitates natural mouse movement, faked location data
Detection requirements A mandatory floor that every accredited measurement organization must apply Optional and separately accreditable, with findings disclosed only at campaign total level

The list-based side runs on shared resources: the International Spiders and Bots List, which the Alliance for Audited Media keeps current for the IAB Tech Lab, and a directory of data-center addresses that the Trustworthy Accountability Group operates. On the hosting side, the MRC also calls for filtering traffic from the three biggest providers, which it names as Amazon AWS, Google and Microsoft.

SIVT is the harder category because it is engineered to look human. That is also why the standard requires campaign-level disclosure only: transaction-level detail would let bad operators reverse-engineer the detection logic and route around it.

Pro tip

When a network or platform claims fraud protection, ask which categories it filters and whether its invalid traffic detection holds MRC accreditation for SIVT. GIVT filtration is the baseline every credible vendor should already meet.

How to Run a Fair First Test on Display Traffic

Work through the same sequence every time:

  1. Fix the success metric before spending. Know the offer payout and the action you are paying for, then decide the cost per action you can accept. Click-through rate never decides anything by itself.
  2. Cap the budget and pre-commit to decisions. Decide in advance what happens at the cap: stop, or continue only if cost per action beats your threshold.
  3. Tag every placement separately. Consistent sub IDs by placement, size and creative tell you which pockets of traffic convert, which is the whole point of Sub IDs: Tracking Which Traffic Actually Converts. Without that split, the entire buy averages into noise.
  4. Run several sizes and positions. One rectangle above the fold and one strip below it measure two different things. Give each a fair sample before cutting.
  5. Watch for invalid traffic signals. Click volume climbing while conversions stay flat, clicks arriving in evenly spaced bursts, and geography that does not match your targeting all deserve investigation. Our guide to spotting fraud and invalid traffic in CPA campaigns walks through the diagnostics.
  6. Use the platform’s own remedies. Platforms such as Google filter invalid traffic out of metrics and billing, and advertisers can request an investigation into suspected invalid traffic. Ask before you assume the worst.

Display traffic does not need defending so much as it needs honest measurement. The standards exist, and the myths survive because most people quit at one size, one placement and one untracked budget. If you want a framework for sizing that first test, see 7 Ways to Test a New Traffic Source Without Burning Your Budget, and if you are still choosing between display and other cheap formats, Push or Native? Choosing Traffic for Your First CPA Offer explains how the alternatives compare on cost, volume and intent.

Frequently asked

What counts as a viewable display ad impression?
The industry standard counts a display impression as viewable once half or more of the ad stays on screen, unbroken, for a full second. Video gets a longer window: half or more of the ad on screen for two seconds.

What is the difference between GIVT and SIVT?
General invalid traffic is caught through routine, list-based checks such as known bots and data-center ranges. Sophisticated invalid traffic is built to look human, requires advanced analytics to identify, and is disclosed only at campaign total level to protect the detection logic.

How do I know if my display clicks are fake?
Watch for click volume climbing while cost per click looks normal and conversions stay flat, clicks arriving in evenly spaced bursts, and geography that does not match your targeting. Platforms let you request an investigation into suspected invalid traffic.

Do I need a large budget to test display traffic?
No. You need a fixed cap, success criteria set before you spend, and tracking granular enough to cut losing placements. A small structured test teaches more than a large unstructured one.

This article is general information for marketers, not legal advice. Advertising, privacy and consumer protection rules vary by country and change over time, so check the current rules and your network’s terms, and get legal advice where needed before running campaigns.