How do I use the PMax Insights page to understand what's driving gains and drops in performance?

Before writing YouTube off, it helps to understand that video advertising behaves differently from click-driven formats by nature: video is immersive, but unlike other formats, people don't always act on it in the moment; the action often comes later, after the viewing session ends. That's exactly why YouTube inventory has its own dedicated conversion metrics that don't overlap with the standard "Conversions" column.

Both channels are officially defined and appear as their own rows in the channel distribution table: the Discover channel covers ads shown in the Discover feed, and the Gmail channel covers ads shown in users' Gmail inboxes. You can see them directly, alongside Search, Display, YouTube, and Maps.

Google's own answer here is direct: PMax campaigns use AI automatically distribute budget across every Google channel to maximize conversions, prioritizing channels the system predicts will perform best; you can't directly control budget by channel, but you can influence it indirectly. This isn't an interface limitation. It's a direct result of how the product is built. PMax is designed as one model that decides, in each auction moment, where to send budget, rather than a set of independently manageable per-channel sub-budgets.

The word "diluted" describes exactly what happens with blended reporting: looking only at aggregate campaign metrics (overall CTR, overall conversion rate) averages a strong Search result together with a weaker Display result into one number. Google's own recommendation is to look at each channel individually through the channel distribution table rather than judging by the blended average, since impressions, clicks, conversions, and cost are all available there broken out by channel.

Google built an official tool for exactly this decision: Shopping versus PMax experiments let you compare one of your existing Standard Shopping campaigns against a new or existing PMax campaign with similar settings, to see which one drives better results for your business. This is a purpose-built A/B test for exactly this question, not a judgment call made after the fact.

The key method here is comparing against the same period a year ago, not just the previous stretch within the current season, since that's exactly how Google's own tool for spotting seasonality is built: the year-over-year comparison checks the last 28 days of search volume against the same 28 days a year earlier, built specifically to separate a recurring market shift from something specific to your account. If a channel's rise or fall matches the same pattern from last year, that's seasonality. If there's no matching pattern from last year, it's more likely something new and specific to the current campaign.

First, it's worth establishing that video ads can grow impressions organically, without depending on a click, unlike Search, so impression growth and conversion growth are expected to diverge here more than in other channels. The "Ads using video" segment in the channel distribution table lets you view video performance separately from the rest of the inventory. That's the first step: confirm you're analyzing this specific segment, not the whole YouTube channel, which can include non-video formats too.

Worth flagging up front: Google's official tools for measuring true, causal incrementality (Conversion Lift, geo-based experiments) measure a campaign or geo region as a whole, not one channel isolated inside a single, already-running PMax campaign. There's no built-in way in the interface to cleanly "cut out" just Display or just YouTube from an overall PMax campaign and honestly measure its separate incremental value.

The mechanics here are fairly direct: without video assets, part of PMax's inventory is unavailable or only partially available, and adding video opens that inventory up fully and right away. Google's own documentation names adding video specifically as an example of an event that produces a visible shift: the time series chart pinpoint exactly when a channel-mix shift happens, for example right when video is added to a campaign. Google is essentially confirming this is an expected, documented pattern, not a side effect.

Google's own troubleshooting documentation names this pattern directly, calling it cross-channel dynamics: with last-click attribution, a channel can look like it has a low ROI when it's actually playing an important role earlier in the path to conversion, since last-click often fails to capture the full contribution of channels engaging users earlier in their journey. That's a direct official acknowledgment that a low channel ROAS under last-click attribution isn't always a diagnosis; it's often an artifact of the measurement method.