Your next campaign move deserves more than a hunch. Opti turns
goal-linked segment evidence into specific proposals—with the reasoning,
assumptions, and requirements in view.
Identify each ad-group, ad, asset, and slot combination. Pair each proposed pause with one eligible replacement in the permitted slot family.
Proposal · replacement mapping required Changes are not applied
DECISION / NO ACTIONILLUSTRATIVE SCENARIO
ADVERTISER GOALPreserve the advertiser’s objective
Recommend no change when the evidence does not support one.
A useful tuning engine also knows when it cannot justify an action. Make the reason visible instead of filling the gap with generic advice.
PROPOSED OPERATIONRetain the current setup; resolve the evidence gap
No campaign mutation is proposed. Missing compatible analysis is a blocker; per-slice decisions can explain no action, missing context, or conflicts.
No action proposed Changes are not applied
Demonstration only. No client results or measured uplift are shown.
Available actions depend on the goal, evidence, platform controls, and
connected-account support.
See the boundary between evidence and action.
A proposal should make its reason and review requirements easy to inspect.
Opti proposal reviewIllustrative
Example question
Why is this campaign change worth reviewing?
M
Illustrative response
Start with the saved evidence and campaign goal. Inspect the proposed change and assumptions, then check current account controls before authorizing any action.
Ask a follow-up about your connected data↗
Conversation example · no live account data A reviewable proposalIllustrative example
1
Evidence
Suitable saved segment insight
2
Proposal
Bounded target CPA change
3
Review
Current settings and assumptions
4
Execution
Separate authorized action
Review first. Authorize any action separately.
Illustrative workflow. Opti must be enabled, suitable saved evidence is required, and a proposal is separate from provider execution.
How a proposal is built.
From a saved objective to a concrete proposal, Opti keeps the evidence
and the decision connected.
01 / FIND
Start with the goal.
Use compatible saved slicing results to identify statistically
significant segments tied to the advertiser’s objective.
02 / GROUND
Ground the proposed controls.
Attempt to refresh current settings and map the segment to
operational entities. If controls cannot be retrieved, keep that
limitation, missing mappings, and inferred parameters explicit.
03 / PROPOSE
Make the move specific.
Describe the target, operation, parameters, and reasoning. Explain a
no-action decision when the evidence does not support a change.
04 / REMEMBER
Bring experience forward.
Relevant measured episodes and reusable playbooks can inform
proposals. New playbook hypotheses remain labeled as untested.
Which changes can Opti propose?
Opti’s action catalog covers campaign controls and test designs. Each
option has its own evidence and applicability requirements.
01
Budget & bidding
Campaign and shared budgets. Target CPA and ROAS. Supported bid
ceilings.
↗
02
Search & destinations
Positive keyword additions. Keyword status changes. Landing-page
tests.
↗
03
Products & Shopping
Product inclusion and exclusion. Bounded product tests. Qualified
winner campaigns.
↗
04
Creative & exploration
Matched App asset replacements. Meta creative tests. Demand Gen
alternatives.
↗
A catalog option is a potential proposal, not a guarantee of platform
support or execution readiness.
Before you act.
Does Opti automatically change my campaigns?+
The tuning flow generates proposals. Applying campaign changes
requires a separate authorized execution flow. A schema-valid proposal
is still unverified until operational and provider checks are
completed.
Is every recommendation ready to execute?+
No. A proposal can contain inferred parameters, unresolved targets, or
missing requirements. Opti surfaces those conditions rather than
treating a hypothesis as an approved change.
Does Opti need a history of past recommendations?+
No. Compatible saved slicing evidence is required for tuning; prior
episodes and playbooks are optional. When used, their relevance must
be explained. New playbook hypotheses are not measured successes.
Does statistical significance prove a change will work?+
No. Segment comparisons show historical associations, not causal lift
or future gains. Proposed changes still need appropriate review,
guardrails, and prospective measurement.
Explore Opti for your campaigns.
Explore evidence-backed campaign recommendations with Opti, MAI’s tuning engine.