Viewability, Attention, and Invalid Traffic: The Three Quality Checks

These three numbers show up in nearly every pharma campaign report. They are useful. They are also easy to over-trust. Each one answers a narrow question, and none of them, alone or together, says the impression was valuable. This page explains what each check measures and where it stops.

Viewability: could the ad be seen?

The Media Rating Council sets the viewability guidelines most of the industry follows. The core thresholds:

Check the current MRC documents before quoting thresholds in a contract. Guidelines get updated.

The key point: viewability measures opportunity to see. An ad that was half on screen for one second passed. Nobody had to look at it. For a pharma display ad that needs to carry a brand name, a claim, and fair balance, one second at half-visible is a low bar.

Viewability also has blind spots. Some environments, especially parts of CTV and in-app, are harder to measure. "Measurable rate" (the share of impressions where viewability could be measured at all) matters as much as the viewable rate. A 75 percent viewable rate on 50 percent of impressions tells you nothing about the other half.

Invalid traffic: was the viewer human?

Invalid traffic is any impression not generated by a real person with real intent to view. The MRC splits it into two groups.

General invalid traffic (GIVT)

Traffic that can be identified with lists and routine filters. Known crawlers and spiders, data center traffic, declared bots, and obvious non-human activity. Most DSPs and verification vendors filter GIVT before or after the bid.

Sophisticated invalid traffic (SIVT)

Traffic that takes more work to detect. Hijacked devices, malware generating fake impressions, bots that mimic human behavior, domain spoofing, and fake app installs. Detecting SIVT needs more data and analysis, and it is where accreditation of the verification vendor matters.

A few practical points on IVT:

• Pre-bid filtering avoids buying invalid impressions. Post-bid measurement tells you what got through. You want both.
• Average IVT rates can look low while one path, app, or domain is heavily invalid. Ask for IVT by path and by domain.
• CTV has its own fraud patterns, like spoofed devices and server-side ad insertion abuse. Confirm your vendor measures CTV as well as web display.
• IVT credits or refunds are worth tracking. If invalid impressions are being billed, someone should reconcile them.

Attention: did anyone engage?

Attention metrics try to answer the question viewability leaves open. Vendors use different inputs: time in view, share of screen, scroll speed, audio on, player size, and in some cases eye-tracking panels used to train models. The output is usually a score or an estimated number of attentive seconds.

What attention is good for:

• Comparing formats and placements within one vendor's method
• Spotting ad-dense pages where viewability is high but attention is low, a common made-for-advertising pattern
• Adding a quality input to path and domain decisions

Where it falls short:

• Methods differ across vendors, so scores are not comparable between them
• There is no single accredited attention standard to rely on as a currency, as of this writing
• Higher attention has not been shown, as a general rule, to cause more prescriptions. Test that link with your own measurement before optimizing to it.

How the three checks fit together

Each one filters a different failure. Think of them in order: was the viewer human (IVT), could the ad be seen (viewability), and did the person likely engage (attention). An impression that fails IVT does not matter for the other two.

The problems tend to come from reading them in isolation. High viewability with low attention on unfamiliar domains suggests clutter. Low IVT with very short sessions suggests real but uninterested traffic. None of these checks look at whether the person was on your HCP list or whether the page was suitable. Those are other layers, covered in the programmatic media quality guide.

Questions to ask your verification vendor

1. Which of your metrics are MRC-accredited, and for which environments (desktop, mobile web, in-app, CTV)?
2. Do you measure SIVT, and is that accredited?
3. What share of our impressions was measurable for viewability, by channel?
4. Can you break out IVT and viewability by SSP, seller, and domain?
5. What happens to invalid impressions: blocked pre-bid, flagged post-bid, or credited?
6. If you offer attention, what inputs drive the score and how was the model validated?

For video specifically, the existing pharma video quality scorecard adds checks beyond completion rate. For path-level breakdowns you will likely need log-level data.

Practical takeaway

Ask your verification vendor for one table: viewability, measurable rate, and IVT by your top 10 SSPs and top 50 domains for the last month. Look for any row that is far off the campaign average. That is where quality is leaking, and the summary number would never have shown it.

Editorial note. Analysis and frameworks are the author's own and do not represent Acxiom or any current or former employer, client, or named platform. Examples labeled hypothetical or illustrative are not results from real campaigns. Nothing here is legal, regulatory, or medical advice.