Performance Max campaigns and when to trust them
Decide whether Performance Max deserves your budget by testing what it adds rather than what it reports since the two are often different numbers.
One campaign across every surface
Performance Max spends a single budget across search and Shopping and YouTube and Display and Discover and Gmail and Maps. You supply a conversion goal and a set of assets and a product feed. Google decides the rest. Your Google representative will recommend it and they are not wrong to.
The trade is explicit. You hand over placement and keyword and audience control in exchange for an optimiser with more data than you have. Where that trade pays depends almost entirely on whether your conversion signal is clean and whether you can verify what the campaign actually added.
What you give up and what it costs
You give up visibility first. Reporting shows far less about where the money went than a standard campaign does and you cannot see the split between channels with any precision. Search themes are suggestions rather than keywords and asset group insight is thin. If your reporting must explain spend to a board then accept that before you start.
You also give up the boundary around your own brand. Left alone Performance Max will happily serve against people searching your company name and against people already in your remarketing pool. Both convert well and neither is new business.
In money the campaign needs assets rather than fees. Images in several ratios and a few videos and headlines and a feed kept accurate. Where that production or management is bought in service fees raised in India attract GST at 18% on top of the fee and sponsored ads management for Google and Meta publishes its scope so you can see which of those pieces are covered.
Who should not run it
Do not run it without trustworthy conversion tracking. Performance Max optimises relentlessly towards whatever you declared a conversion. If that event is a phone icon tap or a page view then it will find thousands of cheap worthless ones and the report will look superb.
Lead generation businesses should be cautious in general. Given a form fill goal it finds the cheapest form fills available which in India often means low intent Display traffic and a sales team buried in junk. If you run it for leads then feed qualified leads back as the conversion event or do not run it.
Do not run it on a new account with no conversion history. There is nothing for it to learn from and the exploration happens on your money. A single product business on a small budget should also skip it since there is no catalogue to work with and a standard search campaign gives the same demand with full control. Ecommerce sellers with an incomplete or inaccurate Merchant Center feed should fix the feed first because the feed is the campaign.
How to run it without losing the plot
The controls that remain are worth using properly. Set the conversion goal to the action that represents money and use value based bidding where your margins differ by product. Apply account level negative keywords and a brand exclusion list so the campaign cannot claim your own name. Keep a dedicated brand search campaign running separately at a low cost per click.
Structure asset groups by the way your catalogue actually sells. Separate margin bands or categories so bidding can treat them differently. Feed audience signals as a hint and accept that the system moves beyond them.
Then give it a defined window of several weeks without edits. Every significant change restarts its learning. Decide the test length in advance and leave it alone.
Judging it honestly
The number that matters is total new customer revenue against total spend across the whole account. Look at the account rather than the campaign. If Performance Max is adding sales then blended acquisition cost improves when you fund it. If it is reallocating credit then the blended figure stays flat while the campaign dashboard improves.
The metric that flatters is the campaign's own return on ad spend. It absorbs brand searches and remarketing and existing Shopping traffic and reports all of it as its own achievement. The cleanest test available is a geographic holdout. Exclude a comparable region for a few weeks and compare total orders between the regions rather than platform attribution.
The common mistake
The expensive error is switching Performance Max on beside existing campaigns and judging it on what it reports. It reports beautifully because it is being fed the easiest traffic in your account. Standard campaigns go quiet. Total revenue stays flat. Budget then moves towards the campaign that cannibalises rather than creates and the business pays a premium for customers it already had.
The second error is treating poor results as a bidding problem. When a Performance Max campaign underperforms the cause is almost always the conversion signal or the feed. Both sit outside the campaign and neither is fixed by adjusting a target.
The trust test
Trust Performance Max when you have clean conversion values and a catalogue worth optimising and a way to verify incrementality. Those three conditions together describe a reasonably mature ecommerce account. Without them the campaign is not dishonest. It is simply reporting on a goal you defined badly and it will do that with great efficiency.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.