21 DV360 Optimization Tips to Improve Campaign Performance in 2026

DV360 optimization is not simply increasing bids, pausing placements with weak CTR, or moving budget toward whatever delivered the cheapest CPM last week.

Strong optimization starts by asking a more basic question:

Is the campaign configured, measured and supplied correctly enough for performance data to be trusted?

A line item can look inefficient because the wrong Floodlight activity is being used. A campaign can underpace because targeting is too restrictive rather than because the bid is too low. A low CPA can look attractive while hiding poor conversion quality. A premium deal can appear weak because extra targeting layers are filtering out most of its eligible inventory.

That is why the best DV360 optimization process follows a sequence:

Validate measurement → diagnose delivery → review bidding → evaluate audiences and inventory → assess creative → make controlled changes → measure again.

If you want to verify the account before making changes, start with the 86-point DV360 Audit Checklist.

Quick answer: The most effective DV360 optimization work usually starts with conversion tracking, campaign goals, pacing and eligible inventory - not with bids. Once the setup is reliable, optimize bidding, budget allocation, audience strategy, inventory quality, frequency, creative and reporting in a controlled order.

What should you check first when DV360 is underperforming?

The order matters. Do not solve an eligibility problem with a bid increase or a measurement problem with an audience change.

1. Verify Floodlight tracking before optimizing anything

Do not optimize toward data you have not validated.

Before changing bids, audiences or inventory, verify:

• which Floodlight activities are assigned;

• whether the activities represent the intended business actions;

• whether sales values or custom variables are being passed correctly where relevant;

• whether duplicate or overlapping activities can inflate totals;

• whether the conversion window and attribution logic are appropriate for the campaign.

Floodlight is not only a reporting layer. It can directly inform bidding. Google documents that custom bidding can use Floodlight conversion counts, sales values and custom Floodlight variables to score impressions.

If the wrong event is selected, DV360 can optimize successfully toward the wrong outcome.

Optimization principle: measurement quality comes before media optimization.

Official reference: Google - Using Floodlight data

2. Optimize toward the right conversion - not every conversion

More conversion signals do not automatically create better optimization.

A campaign may track:

• product views;

• registrations;

• quote requests;

• trial starts;

• purchases;

• revenue;

• offline-qualified leads.

These actions should not always have equal importance.

For lead generation, a form submit may be too shallow if only a fraction of leads become qualified. For ecommerce, optimizing only to transaction count can ignore order value. For subscription businesses, a trial start may matter less than an activated or paid user.

Where the data supports it, structure conversion logic around the business outcome you actually want.

Google's rule-based custom bidding can weight conversion events differently, and custom bidding can also use conversion values and other signals.

Optimization principle: choose the signal that best reflects business value, not simply the easiest event to generate.

Official reference: Google - Custom bidding with rules

3. Align the insertion-order KPI with the bidding strategy

The KPI used to judge campaign performance should be consistent with the outcome the bidding system is trying to produce.

If the business goal is acquisition efficiency but the campaign is structurally optimized for a different metric, reporting and optimization can begin pulling in different directions.

Before launch or during a major review, document:

1. the campaign business objective;

2. the primary KPI;

3. the conversion or outcome being used;

4. the bidding strategy;

5. the reporting metric used to decide whether the campaign is working.

If these five points are not aligned, the team can end up optimizing one metric while evaluating another.

Optimization principle: one campaign should have a clear definition of success.

4. Avoid unrealistically aggressive CPA or ROAS targets

A tighter target is not always a better target.

Google notes that when an automated strategy is asked to prioritize a specific KPI, an overly aggressive goal can restrict spending because meeting the target takes priority over exhausting the budget.

This is a common optimization mistake:

Performance declines → target CPA is lowered immediately → delivery contracts → conversion volume falls → the algorithm receives less useful data.

Instead, compare the target with:

• historical campaign performance;

• conversion volume;

• inventory conditions;

• current market CPMs;

• audience scale;

• the quality of the conversion event.

If the campaign has never demonstrated a given CPA or ROAS level, forcing the bidding system to hit it immediately can reduce scale rather than create efficiency.

Optimization principle: use targets as achievable constraints, not wishes.

Official reference: Google - Automated bid strategies

5. Give automated bidding enough stability to learn

Do not judge an automated strategy after every short-term fluctuation.

DV360 automated bidding uses multiple signals - including inventory, audiences, creative effectiveness and line-item strategy - to decide which impressions are more valuable. Google also warns that major changes to bids, budgets or targeting can trigger re-learning and temporary delivery changes.

That means repeatedly changing:

• budgets;

• bid targets;

• conversion settings;

• targeting;

• CPM limits;

can make it harder to distinguish real performance trends from system adjustment.

There is no single universal learning period that applies to every DV360 strategy, format and campaign. The correct amount of time depends on strategy, volume and the significance of the change.

Optimization principle: make material changes deliberately, then allow enough data to accumulate before judging them.

Official reference: Google - Automated bid strategies

6. Diagnose underpacing before increasing bids

Underpacing does not automatically mean the bid is too low.

A line item can fail to spend because of:

• restrictive audience targeting;

• narrow geography;

• incompatible targeting combinations;

• limited inventory;

• tight viewability requirements;

• frequency caps;

• rejected or unassigned creative;

• deal eligibility;

• budget or pacing settings.

Google explicitly notes that targeting combinations can reduce or eliminate available inventory. A line item may already be bidding on nearly every eligible impression while still having very little scale.

Before raising bids, ask:

How much inventory is actually eligible?

Then determine whether the campaign is:

1. not entering enough auctions;

2. entering auctions but losing them;

3. being restricted by campaign settings.

Optimization principle: diagnose eligibility before price.

Official references:
Google - About targeting in DV360
Google - Troubleshoot deals and line items

7. Choose pacing based on the campaign objective

Pacing determines how quickly DV360 spends the available budget.

The correct setting depends on the campaign.

A short promotional burst may need faster delivery. A month-long always-on campaign usually needs a more controlled distribution of spend. A limited deal or event window can require different pacing from evergreen prospecting.

Google's pacing options include settings designed to spend rapidly as well as settings intended to distribute spend over time.

Do not choose pacing only because it increases delivery.

Ask:

• Should the budget be available throughout the full flight?

• Is demand concentrated on particular days?

• Is early learning more valuable than even exposure?

• Is inventory scarce enough that missed opportunities cannot be recovered later?

Optimization principle: pacing is a strategy decision, not housekeeping.

Official reference: Google - Set budgets and control pacing

8. Use automatic budget allocation only when the structure supports it

Automatic budget allocation can help shift budget toward line items that DV360 expects to perform better against the insertion-order goal.

That can be useful - but only when the line items are genuinely comparable within the same optimization framework.

Be cautious when:

• line items have different strategic roles;

• certain publishers or deals must receive committed spend;

• test cells need fixed budgets;

• prospecting and remarketing should not compete freely for the same money;

• some channels have very different attribution behavior;

• a line item exists for reach or learning rather than direct-response efficiency.

Google explains that automatic budget allocation evaluates line items against supported insertion-order goals and can move budget according to performance.

Optimization principle: automate allocation when line items should compete; use fixed budgets when strategy requires protection.

Official reference: Google - Automatic budget allocation

9. Move budget toward proven line items - not simply the cheapest CPM

The cheapest inventory is not automatically the most efficient inventory.

A low CPM line item can still produce:

• weak viewability;

• low-quality sessions;

• poor conversion rate;

• high frequency;

• low incremental reach;

• weak video completion;

• unsuitable environments.

Budget decisions should be based on the campaign KPI and the quality of the outcome.

For a conversion campaign, compare CPA, conversion volume and quality - not CPM alone. For video, include completion and viewability. For reach, evaluate unique reach and frequency. For ecommerce, consider conversion value where available.

When reallocating budget, ask:

Which line item creates the best useful outcome per unit of spend?

That is a better question than:

Which line item buys the most impressions?

10. Check frequency at campaign, insertion-order and line-item level

Frequency can restrict delivery more than traders realize.

DV360 can apply frequency controls at multiple levels. Google notes that when a campaign-level cap is used, serving stops when the first applicable cap across campaign, insertion order or line item is reached.

That means a line item-level frequency setting cannot be evaluated in isolation.

Review:

• campaign frequency cap;

• IO frequency cap;

• line-item frequency cap;

• time window;

• audience size;

• overlap between line items;

• how much inventory is being lost to frequency.

DV360's Optimization view includes a Freq. limited metric showing impressions that were not bid on because a frequency cap had been reached.

Optimization principle: frequency is both a reach control and a delivery constraint.

Official references:
Google - Create a campaign
Google - Optimization view

11. Separate prospecting and remarketing logic

Prospecting and remarketing solve different problems.

Prospecting tries to create new demand or find new customers. Remarketing focuses on people who already interacted with the brand.

If both audiences sit in the same optimization structure, remarketing can dominate because those users are naturally closer to conversion.

Separate them when you need to understand:

• incremental acquisition;

• prospecting CPA;

• remarketing efficiency;

• frequency;

• audience saturation;

• budget split.

Also use audience exclusions where appropriate so prospecting is not quietly absorbing existing users.

Optimization principle: do not let warm demand make cold acquisition look stronger than it is.

Official reference: Google - Remarketing

12. Use optimized targeting deliberately - not automatically

Optimized targeting can expand beyond manually selected audience segments to find users more likely to achieve the campaign goal.

That can be valuable for scale and performance.

But traders need to understand what the selected audiences are doing.

Google distinguishes between:

• respected inputs, which remain hard constraints;

• soft signal inputs, which guide optimized targeting but do not necessarily limit delivery.

For example, geographic settings, first-party exclusions and brand-safety exclusions can remain respected constraints, while many included audience signals can act as guidance rather than strict boundaries.

This matters for reporting.

If a stakeholder believes a campaign is serving only to a named audience, but optimized targeting is expanding beyond it, the campaign strategy has been misunderstood.

Turn optimized targeting on because expansion supports the campaign goal - not simply because the option is available.

Optimization principle: know whether your audience is a constraint or a signal.

Official reference: Google - Optimized targeting

13. Review audience overlap and exclusions

Audience structures become inefficient when multiple line items repeatedly compete for the same users.

Common causes include:

• overlapping first-party lists;

• multiple remarketing windows;

• broad prospecting audiences with no customer exclusion;

• similar audience packages split across separate line items;

• acquisition campaigns serving to existing customers.

Review audience logic at campaign level, not only within individual line items.

Questions to ask:

• Are existing customers excluded from acquisition where appropriate?

• Are remarketing windows logically separated?

• Are multiple line items targeting nearly identical users?

• Is the overlap intentional for a test, or accidental?

• Is one audience receiving excessive frequency because several line items can reach it?

Optimization principle: audience architecture should make performance easier to interpret, not harder.

14. Use Inventory Availability before over-restricting targeting

Before adding another targeting layer, estimate what it does to scale.

Google's Inventory Availability report estimates impressions that match specified targeting criteria and can be refined by inventory and targeting filters.

This is especially useful before combining several restrictions such as:

small geography + narrow audience + specific device + high viewability + limited publishers + strict brand suitability

Each condition may be reasonable on its own. Together, they may leave too little inventory to support the campaign.

Use availability data when:

• launching in a new market;

• planning a CTV campaign;

• testing narrow audiences;

• expanding publisher lists;

• diagnosing underdelivery;

• applying Authorized Seller restrictions.

Optimization principle: forecast before restricting.

Official reference: Google - Inventory Availability report

15. Audit brand suitability before blaming inventory scale

Brand suitability protects campaigns from content environments that do not meet the advertiser's standards.

But stricter suitability also changes the inventory pool.

Google offers different inventory suitability levels, including settings designed for maximum reach and more restrictive options intended for brands that want tighter content controls.

The correct optimization is not:

“Open everything to get more scale.”

It is:

“Use the least restrictive settings that still satisfy the advertiser's actual brand requirements.”

If a campaign underdelivers, document:

• current suitability level;

• content-category exclusions;

• third-party verification;

• app/site exclusions;

• whether deal inventory is being affected;

• the estimated scale impact.

Do not weaken a brand requirement simply to make pacing look better.

Optimization principle: balance scale and suitability consciously.

Official reference: Google - Brand suitability

16. Review Authorized Seller restrictions and inventory quality

Inventory-quality controls can affect delivery.

DV360 supports Authorized Seller reporting based on ads.txt and app-ads.txt participation and authorization status. Google also notes that the reach impact of Authorized Seller targeting can be estimated through Inventory Availability reporting.

If scale is weak, review:

• whether only direct authorized sellers are allowed;

• whether authorized resellers are eligible;

• whether important app inventory is being excluded;

• whether the restriction matches the advertiser's actual supply-quality policy.

The goal is not to remove quality controls.

The goal is to know how much delivery they are costing and whether that trade-off is intentional.

Optimization principle: measure the cost of every supply restriction.

Official reference: Google - Authorized Digital Sellers

17. Separate open auction, PMP and premium-deal performance

Do not hide every supply path inside one blended average.

Open auction, private marketplace deals and other premium buying routes can have different:

• CPMs;

• publisher quality;

• scale;

• viewability;

• completion rates;

• audience composition;

• commercial purpose.

A premium deal may cost more but produce stronger attention, better completion or strategic publisher access. An open-auction line item may look cheaper but deliver a different outcome.

Evaluate deals against the reason they were bought.

For each deal, check:

• eligibility;

• creative approval;

• targeting;

• device;

• geography;

• frequency;

• bid/CPM;

• pacing;

• performance.

Google's deal troubleshooting tools separate filtered and auction-stage losses, helping diagnose whether a deal is failing because of frequency, budget, pacing or auction conditions.

Optimization principle: optimize supply paths according to their role - not one blended CPM target.

Official reference: Google - Troubleshoot deals and line items

18. Do not optimize creative using CTR alone

A high CTR can be useful. It is not a universal measure of strong creative.

For display or video, compare creative against the actual campaign objective.

Possible signals include:

• conversion rate;

• CPA;

• post-click quality;

• viewability;

• video completion;

• engagement;

• conversion value;

• frequency;

• landing-page behavior.

A creative can generate clicks because it is unusually aggressive or curiosity-driven while producing weak downstream outcomes.

A different creative may attract fewer clicks but better customers.

Also check the basics:

• approval status;

• correct destination;

• dimensions;

• video specifications;

• rotation;

• assignment to the correct line item.

A line item cannot serve until eligible creative is assigned and approved.

Optimization principle: optimize creative for business response, not vanity engagement.

Official reference: Google - Create a line item

19. Refresh creatives before fatigue becomes the entire explanation

When performance deteriorates over time, traders often jump immediately to audience or bid changes.

Sometimes the problem is simpler: the audience has seen the same message too many times.

Watch the relationship between:

• frequency;

• CTR;

• conversion rate;

• CPA;

• video completion;

• creative-level performance over time.

If frequency is rising while engagement and conversion performance are declining, test fresh creative before rebuilding the entire audience strategy.

For retargeting, creative sequencing can also reflect user intent:

• product education;

• proof or benefit;

• offer;

• urgency;

• reactivation.

Optimization principle: media optimization cannot permanently compensate for stale messaging.

20. Use the Optimization view as a diagnostic tool - not an autopilot button

DV360's Optimization view can surface useful delivery and performance information, including:

• pacing;

• projected spend;

• budget at risk;

• performance vs goal;

• lost impressions;

• frequency-limited impressions;

• win rate.

Use these metrics to identify where the constraint may be.

Then investigate why.

For example:

• high lost-to-budget can indicate constrained spend capacity;

• high lost-to-bid can indicate auction competitiveness;

• high frequency-limited impressions point toward exposure constraints;

• weak pacing with low potential spend can indicate an eligibility problem.

The Optimization view is useful because it organizes signals.

It should not replace campaign diagnosis.

Optimization principle: platform recommendations are inputs to a decision - not the decision itself.

Official reference: Google - Use the Optimization view

21. Make optimization changes in controlled batches and document what changed

The final tip is operational.

If you change the audience, bid strategy, creative, budget, pacing and frequency on the same day, you may improve performance - but you will not know why.

Build a simple optimization log.

For every material change, record:

1. what changed;

2. why it changed;

3. what metric should improve;

4. when the result will be reviewed.

This creates a repeatable optimization process and prevents teams from reversing each other's work.

For reporting workflows, see the DV360 to Data Studio dashboard guide.

Optimization principle: change → measure → learn → repeat.

A practical DV360 optimization workflow

The 21 tips above can be condensed into seven stages.

This sequence prevents a common mistake in DV360:

optimizing a symptom before identifying the cause.

How often should you optimize DV360 campaigns?

There is no universal schedule that fits every campaign.

The right cadence depends on:

• spend;

• conversion volume;

• campaign length;

• bidding strategy;

• inventory type;

• objective;

• how quickly market conditions change.

High-spend campaigns may need daily delivery monitoring. Strategic performance changes may require more data before they can be judged.

A useful operating rhythm is:

Daily

Check:

• spend;

• pacing;

• rejected creative;

• major delivery problems;

• unexpected spikes.

Several times per week

Review:

• line-item allocation;

• audience scale;

• inventory concentration;

• bids and win rate;

• frequency;

• deal delivery.

Weekly

Evaluate:

• KPI trend;

• creative performance;

• audience and inventory efficiency;

• conversion quality;

• optimization changes made during the previous period.

At major milestones

Run a broader audit:

• after launch;

• after meaningful learning;

• before scaling budget;

• when performance changes materially;

• when taking over an existing DV360 account.

For a structured QA process, use the DV360 Audit Checklist.

What not to do when optimizing DV360

Avoid these common patterns:

• raising bids before checking eligible inventory;

• lowering CPA targets every time performance weakens;

• changing several major settings at once;

• optimizing toward an unverified Floodlight activity;

• judging all inventory by CPM;

• evaluating creative only by CTR;

• forgetting campaign-level frequency when reviewing line-item delivery;

• assuming optimized targeting stays inside included audience segments;

• allowing remarketing to distort prospecting performance;

• removing brand-suitability controls just to increase scale.

Optimization should make the campaign more explainable, not less.

When should you use custom bidding?

Standard automated bidding is appropriate for many campaigns.

Custom bidding becomes more relevant when the business objective cannot be represented well by a standard platform KPI.

Examples include:

• giving qualified leads more value than basic leads;

• optimizing toward revenue rather than transaction count;

• weighting several conversion actions differently;

• using Floodlight custom variables;

• creating a business-specific impression value.

Google supports rule-based and script-based custom bidding and allows Floodlight, Google Analytics events and impression signals to influence impression value.

The important question is not whether custom bidding is more sophisticated.

It is whether your business has a better signal that standard optimization cannot represent.

Official reference: Google - Custom bidding overview

DV360 optimization for CTV campaigns

CTV requires extra discipline because inventory, devices, geography, publisher access and frequency can all constrain scale.

Before changing bids, check:

• CTV device targeting;

• publisher or deal eligibility;

• geography;

• audience size;

• frequency caps;

• creative compatibility;

• inventory availability;

• brand suitability.

A highly specific CTV campaign may simply have too few eligible impressions.

Also remember that optimized targeting is not available for CTV line items, according to Google's current documentation.

For local targeting strategy, read CTV Geofencing: Neighborhood & Household Targeting.

The bottom line

DV360 optimization works best when it is treated as diagnosis rather than constant adjustment.

The strongest teams do not ask:

“What setting should we change today?”

They ask:

“What is limiting the campaign, what evidence supports that diagnosis, and which single change gives us the cleanest way to test it?”

A disciplined DV360 optimization process should:

• verify measurement;

• align KPIs and bidding;

• diagnose pacing;

• protect useful learning;

• manage frequency;

• evaluate audience scale;

• understand inventory constraints;

• balance brand suitability and reach;

• assess creative beyond CTR;

• document every material change.

If the campaign setup itself has not been validated, run the 86-point DV360 Audit Checklist first.

If your team needs hands-on campaign execution, optimization and troubleshooting, read the DV360 Managed Service Guide.

Need help improving DV360 campaign performance?

Request Access

Need help improving DV360 campaign performance?

Request Access

Frequently Asked Questions

Primary Sources & Further Reading

• Google - Automated bid strategies

• Google - Set budgets and control pacing

• Google - Automatic budget allocation

• Google - Use the Optimization view

• Google - Using Floodlight data

• Google - Custom bidding overview

• Google - Custom bidding with rules

• Google - Optimized targeting

• Google - Inventory Availability report

• Google - About targeting in DV360

• Google - Brand suitability

• Google - Authorized Digital Sellers

• Google - Troubleshoot deals and line items

• Google - Create a campaign

• Google - Create a line item