Google Ads Experimentation Tools: September Testing Playbook

If you manage Search or Performance Max, you already spent August staring at the September 1 AI Max migration and asking the same question every agency asked. What is the fastest, cleanest way to prove out a change without lighting a client’s budget on fire? Google’s answer arrived on August 20. The new google ads experimentation tools are the first upgrade in years that actually maps to how agencies test.

The rollout brings three practical changes. A single A/B test can now span multiple Search campaigns and compare different budgets and ROI targets. AI Max experiments can now run with brand and location settings enabled. Performance Planner can push its own suggestions into live campaigns in one click. Below is the September playbook so you can adopt the parts that help without stepping on the guardrails that are baked into the flow.

What the New Google Ads Experimentation Tools Actually Do

Google announced three linked upgrades on its Ads and Commerce blog. Search Engine Journal walked through the practical picture in a piece by Brooke Osmundson worth reading alongside this one.

The headline item is a cross-campaign A/B test. Starting in September, you can run one experiment that changes budgets and ROI targets across a group of Search campaigns at the same time, rather than editing each campaign on its own. Google frames the purpose as showing what happens to overall results when you scale, not just when you tweak. That framing matters because most retainer clients hire you to grow revenue, not to shave a cost-per-lead by fractions.

The second upgrade removes a real blocker for agencies that separate branded and non-branded traffic. Until now, AI Max experiments made you strip brand and location settings out of the campaign before you could test. That was a nonstarter for teams running strict brand controls, and it forced awkward workarounds. The updated flow lets those controls stay on during the trial.

The third change is inside Performance Planner. You can review the tool’s suggestions, deselect any campaigns you do not want to update, and push the approved list straight into live campaigns with one click. No export, no bulk sheet, no manual load. Every change is reversible through Bulk actions if the numbers do not hold.

Where the Google Ads Experimentation Tools Fit in Your September Plan

September is already a heavy month for anyone in Search. The AI Max migration for legacy Search campaigns kicked off on September 1, and the language targeting removal is close behind. If you have not walked through those two changes yet, the writeup in our September 1 AI Max audit is the place to start, and our earlier AI Max deadline plan covers the automatically created assets and broad match transitions.

The new tools slot into that migration cleanly, but only if you sequence them right. Run the migration audit first. Confirm which campaigns actually converted and which did not. Only then plan the experiments. Testing on top of a broken migration wastes budget and blurs the signal.

Prioritize experiments where the risk is real. Budget tests belong on campaigns that already spend enough to reach statistical significance inside a few weeks. Do not run a multi-campaign budget experiment on a $50-a-day campaign and expect a clean read. If your client has a mix of small and large campaigns, group the large ones for the cross-campaign test and let the small ones ride.

Read the Fine Print Before You Run an AI Max Experiment

Google’s help documentation lists six conditions that block an AI Max experiment from being created through the new flow. Any one of them is enough to break the setup.

A campaign cannot have text customisation enabled, the setting formerly labelled automatically created assets. It cannot target the Display Network. It cannot sit inside a portfolio bidding strategy. It cannot use shared budgets. It cannot use bidding exploration. And it cannot already have another experiment running.

Five of those six also apply to the older custom experiment flow, so most teams have already run into a few of them. The Display Network restriction is the one new item. The shared budget rule is worth watching too, because the cross-campaign budget test is explicitly built to compare budget changes across multiple campaigns, yet the flow blocks campaigns that use shared budgets. Google has not explained how the two rules interact, so the documented workaround for now is to build a fresh test campaign without the legacy features.

The text customisation restriction has a September twist. The same campaigns that run automatically created assets are the ones scheduled for automatic conversion to AI Max on September 1. If a client’s campaigns just migrated, your ability to test them through the new flow depends on which legacy features carried through.

Practical Playbook for the New Google Ads Experimentation Tools

Here is a five-step rhythm you can drop into your September and October client work.

Step 1. Audit each account against the six blockers before you plan a single test. Note which campaigns qualify and which will need workarounds.

Step 2. Pick two experiments per account, no more. One should be a budget or ROI target test using the new cross-campaign format, focused on the campaigns that already spend enough to give you a clean read. The other should be an AI Max experiment on a branded or geo-restricted campaign that you previously could not test.

Step 3. Set your control arm and trial arm deliberately. AI Max experiments divert traffic and budget inside a single existing campaign rather than creating a copy. A percentage of the campaign runs with AI Max off, the rest with it on. Decide whether search term matching and asset optimisation should apply to the trial arm alone, and document the choice for your review.

Step 4. Use Performance Planner as a preview, not an autopilot. Let it show you the projected impact of budget or bidding changes. Review the campaign-level suggestions. Deselect any you would not have made by hand. Apply what remains, then log the change so you can reverse it through Bulk actions if the numbers turn.

Step 5. Report the outcome to your client in the same rhythm you use for the rest of your ads work. Two or three lines on what changed, one line on the read, one line on the next action.

Where Bidding and Budget Guardrails Fit In

Experiments show you what happens. Guardrails decide how much room you give the algorithm to move first. Before you run the cross-campaign budget test, make sure the bidding target and budget-limited signals are stable on the campaigns you plan to include.

Our bidding target optimization audit and the budget-limited bidding audit both walk through what to look at before you scale spend. Those checks matter even more inside an experiment, because a campaign that is chronically budget-limited will not read cleanly no matter how big your test arm is.

For accounts running new customer acquisition goals, keep an eye on the split described in our bidding-safe split for new customer reporting. If you are already using target CPA or target ROAS with an optional target, our rename audit playbook covers the cleanup work that keeps experiments honest.

Google Ads Experimentation Tools and PMax Sit Side by Side

The new google ads experimentation tools do not directly rework Performance Max testing, but the two campaign types share more surface area every month. Search Term Matching and asset optimisation, the two AI Max toggles the experiment activates by default, are close cousins to the automation you already accept when you run PMax.

If you are actively testing Performance Max features alongside these new tools, keep your PMax reads clean. Our PMax partners setting playbook and the AI Max for Shopping playbook both cover the operator moves that keep PMax from swamping the read on an adjacent Search experiment.

The short version is simple. Run one thing at a time per account. If a Search campaign is inside an active AI Max experiment, do not simultaneously rework the PMax feed pointed at the same audience. Wait until the experiment has a decision before you touch the other side.

What to tell your client this month. The new google ads experimentation tools give you a legitimate way to test scale without duplicating campaigns and without sacrificing brand or location controls. That is the pitch. The delivery is a short list of qualifying campaigns, two well-scoped experiments, and a review cadence that keeps everyone on the same page. If you want a second set of eyes on the plan before you spin up your first cross-campaign test, book a free consultation and we will walk your account with you.

Let’s Grow!