Google Just Redefined a Video Engagement: Why Your Numbers Dropped

The Monday your dashboard lied to you

You open Google Ads, glance at your YouTube and Demand Gen campaigns, and something looks wrong. Engagement rate is down. Not nudged down, but visibly down, the kind of drop that makes you check whether a campaign broke over the weekend. Your first instinct is to blame the creative, pause something, or fire off a worried message to whoever runs your ads.

Don't. On Oct 10, Google changed what counts as a video engagement, and the new definition is almost certainly why your numbers fell. Your ads probably did exactly what they did last week. The ruler changed, not the thing being measured.

This matters more for Indian D2C brands than it looks, because so much of our prospecting now runs through Demand Gen and Shorts, which are squarely in the blast radius. It's a measurement update dressed up as a performance crash. React to the smaller number instead of the cause and you'll make expensive decisions about a metric that didn't actually move. Here's what happened and what to do about it.

What Google changed, exactly

From 5 seconds to 10

Until last week, a video engagement was counted after roughly 5 seconds of watch time. Google raised that bar to a full 10 sec. The viewing behaviour hasn't changed, but the threshold a view must clear to be counted has doubled, so fewer views qualify. That's the whole mechanism, and it's why the drop shows up everywhere at once rather than campaign by campaign.

The move also lines up with a broader shift in how YouTube counts views across formats earlier this year, so if it feels like the platform has been quietly redrawing its measurement lines for months, that's a fair read. This is the version that lands inside your ad reports.

Where it bites

The higher threshold applies across the placements most D2C brands actually use for video:

• In-stream and video app promotion ads: 10 seconds watched, or a click
• In-feed video ads: 10 seconds played while muted, or a thumbnail click
• Masthead ads: 10 seconds of muted playback, or a click
• YouTube Shorts ads: 10 seconds watched, or a tap on the call-to-action
• Demand Gen campaigns, which lean on in-feed and in-stream inventory

If you run Demand Gen for prospecting, which most Indian D2C brands now do, this is where you'll feel it most. Shorts in particular, where a lot of watch time naturally sits in that 5-to-10-second band, can look like it fell hardest.

Where it doesn't

A few formats are untouched because they never counted watch time the same way:

• Bumper and non-skippable in-stream ads, which count only clicks
• Lightbox and Gmail ads, which count expansions

So if a campaign made entirely of bumpers looks stable, that's expected, not luck. Use those steady campaigns as a sanity check: if they held flat while your Shorts and Demand Gen numbers dropped, that pattern alone tells you the cause is definitional, not creative.

The before-and-after in one table

Expect engagement rate and count to fall, sometimes sharply, because fewer views now clear the higher bar. How far they fall depends on your creative, your audience, and how much of your watch time sat between 5 and 10 seconds, so treat any specific percentage you see floating around online as a guess, not a benchmark for your account. A snackable Shorts ad and a 30-second product story will move by very different amounts, and both can be perfectly healthy.

Why your ROAS and conversions are fine

Here's the part that should lower your blood pressure. This change touches how engagement is counted. It does not touch how conversions are tracked. A purchase is still a purchase, an add-to-cart is still an add-to-cart, and your revenue numbers are reporting the same reality they did on October 9.

It matters even less if you bid on Target ROAS or Target CPA. Those strategies optimise toward conversions, not engagement, so Google's bidding keeps chasing the same outcomes regardless of where the engagement line sits. The metric that got smaller isn't the one your bidding runs on.

For a D2C brand, the number that pays your bills is contribution margin per order, and that sits a long way downstream of a video engagement. If your cost per purchase, blended ROAS, and margin all held steady while engagement rate fell, you have your answer: nothing is wrong. The real risk here isn't to your performance, it's to your judgement, if you read a definitional drop as a real one and start cutting winners heading into your busiest quarter.

What to actually do this week

Re-baseline before you react

Export engagement rate and count for the 30 days ending October 9, so you have a clean "old definition" baseline on record. Then leave bids and budgets alone for about two weeks. You can't fairly compare a 5-second world to a 10-second world, and any decision you make in that window rests on a broken comparison. Mark a hold on engagement-based optimisation and tell the team why, so nobody quietly "fixes" a campaign that isn't broken.

Fix your automated rules

This is the one that quietly bites. If you have rules that pause, flag, or shift budget based on an engagement threshold, those rules don't know the definition moved. The same healthy campaign now throws a smaller number, which can trip a rule and pause something that's actually working the night before a sale. Go through every engagement-based rule and lower the thresholds so they still mean what you intended.

If you run leaner automation through the Google Ads scripts layer, the same warning applies: anything keying off engagement counts needs its numbers reset against the new baseline, and it's worth a proper audit rather than a quick patch. A tidy approach to this kind of account plumbing is covered in our Google Ads scripts work, and the principle holds whether you script it or set rules by hand, which is that automation is only ever as safe as the benchmark behind it.

Re-benchmark through December

Treat October 10 as line zero for engagement. Build fresh benchmarks from mid-October through December and annotate the date in your dashboards, so three months from now nobody mistakes a one-time definition change for a trend. Your December-versus-October comparison will otherwise look like a collapse that never happened.

Put it in plain words for stakeholders

If you send monthly reports to a client or a founder, add one line before they ask: "Google changed how video engagement is counted on 10 October, so engagement rate and count dropped across YouTube and Demand Gen. Conversions, ROAS, and CPA are unaffected." That single sentence saves a tense call later and shows you caught it rather than missed it.

The pattern worth noticing

This is the second or third time this year a platform has changed a metric's definition and let the resulting chart do the scaring. It's a reminder that engagement metrics are platform conveniences, not laws of nature, and they can be redrawn whenever the platform wants cleaner-looking data or a higher bar for what it rewards.

The accounts that stay calm through these resets are the ones anchored to money metrics, like revenue, blended ROAS, contribution margin, and cost per acquisition, because those don't get redefined overnight to make a dashboard look tidier. Soft metrics like engagement rate are useful for diagnosing creative, but they should never be the thing a budget decision hangs on. If this week nudges you to move your reporting a notch closer to the money, the metric change will have done you a quiet favour.

Bottom line

Your video engagement didn't crash. Google moved the line from 5 seconds to a full 10 sec on Oct 10, fewer views clear it, and the reports got smaller as a result. Save your old baseline, keep your hands off bids for two weeks, fix any rule that keys off engagement, tell your stakeholders in one sentence, and keep judging campaigns on conversions and ROAS, which this change can't touch. Do that and the scary chart becomes a footnote, which is all it ever was.