The New Performance Max: How Google PMax Changed in 2026

First published: 2 July 2026 · Last updated: 2 July 2026

1

Search themes: 25 to 50 per asset group

Doubled limit means tighter intent control. SG advertisers can now segment search themes by buying-stage, persona or product line in ways that were not possible at 25.

2

Video assets: 5 to 15 per asset group

YouTube and Demand Gen surfaces get more variant rotation. Short-form 9:16 plus 16:9 plus 1:1 in the same group is finally practical.

3

Channel-level reporting (finally)

You can now see what spend went to Search, Display, YouTube, Discover, Gmail, Maps and Shopping separately. End of the "black box" complaint that defined PMax 2022 to 2024.

4

Full search term visibility

Every triggering search term now shows in the report. SG advertisers can audit waste, build comprehensive negatives, and brief better Search Themes.

5

10,000 campaign-level negatives

Up from a few hundred. SG accounts can now block branded competitor terms, irrelevant industries and existing-customer searches at scale, recovering meaningful budget.

6

Demand Gen integration and conversion expansion controls

Demand Gen runs as a sibling, not a rival. New conversion-value-rules and audience-expansion-toggle controls let SG advertisers cap PMax's tendency to chase loose conversions.

If your SG Google Ads account has been on autopilot since 2024, you are running yesterday's Performance Max with yesterday's expectations. Google has shipped roughly two years of meaningful PMax updates between 2025 and the first half of 2026, and most of them flip assumptions that were baked into how SG agencies and in-house teams structured PMax campaigns when they first migrated from Smart Shopping. The result: a re-audited, re-structured 2026 PMax setup is producing 20 to 40 percent better ROAS on the same budget for almost every SG account we touch. This piece walks through the six material 2026 changes, what each one means for SG advertisers in practical terms, and the asset group, search theme and reporting changes you should make this quarter. We are deliberately staying out of "what is PMax" territory; for that, our existing piece on the Performance Max campaign basics still holds. This is the 2026 update layer on top.

Change 1: Search Themes Doubled to 50 Per Asset Group

Search themes are the closest thing PMax has to keyword targeting. Each theme is a phrase or short query you supply that tells Google's automation which search-intent clusters to prioritise inside an asset group. They behave roughly like broad-match keywords with intent grouping. The 2024 limit was 25 themes per asset group. From mid-2025, Google rolled the limit to 50. This sounds incremental. In practice it changes asset-group strategy completely. At 25 themes you had to mash together everything the asset group was meant to cover into one diluted intent set. At 50 you can finally segment cleanly. For an SG B2B SaaS account we run, this meant splitting a single 25-theme "growth marketing software" asset group into a tighter 50-theme structure that separately covered category research queries, comparison queries, jobs-to-be-done queries and competitor-alternative queries. Same asset group, much cleaner intent grouping. ROAS lifted 31 percent in the first 60 days post-restructure. **The SG-specific play.** Use the 50-theme limit to add SG-specific phrasing alongside generic terms. "Singapore", "in SG", "for sg companies", "near me singapore", "best in singapore" combinations against your category terms. Google's automation does not always favour SG variants on its own; supplying them explicitly through search themes pushes the system toward SG-relevant queries.

Change 2: Video Assets Tripled, From 5 to 15 Per Asset Group

PMax has been a force multiplier for accounts that bring video assets. Until 2025, you could only supply 5 videos per asset group. The new limit of 15 unlocks meaningful variant testing. Practical SG playbook for the 15-video allowance:

• 3 to 4 videos in 9:16 (Shorts, Reels-style, Demand Gen vertical surfaces)
• 3 to 4 videos in 16:9 (YouTube standard, in-stream)
• 2 to 3 videos in 1:1 (Discover, Gmail, mid-feed placements)
• 2 to 3 longer-form 16:9 (15 to 30 seconds for connected TV and skippable in-stream)
• 1 to 2 product-focused short videos (15 second product highlights for Shopping integrations)

Most SG advertisers we audit are running 1 to 2 videos in PMax, often in landscape only. Underutilising 13 of the 15 video slots is the single biggest unlocked opportunity in PMax 2026. The cost is real video production, but for SG accounts spending SGD 5,000+ a month on PMax, the math always favours producing 6 to 10 short variants. For the supply-side angle on short-form video creative, our piece on short-form video marketing 2026 covers production workflows that produce ad-ready variants at SG SME scale.

Change 3: Channel-Level Reporting Ends the Black Box

The single biggest historical complaint about PMax from SG agencies was opacity. You did not know how much of your spend went to Search vs Display vs YouTube vs Discover vs Gmail vs Maps vs Shopping. You just knew the blended ROAS. Now you know. The new channel report (under the campaign's Insights tab) shows spend, impressions, clicks and conversions per channel. Three patterns to look for: **Pattern 1: Display spending too high.** PMax tends to over-allocate to Display when conversion data is sparse. If Display is more than 30 to 40 percent of your spend without commensurate conversions, you have a creative quality issue, an audience signal issue, or both. **Pattern 2: YouTube underspending.** If your video assets are weak or absent, YouTube spend collapses. Re-audit videos against the 15-asset allowance from Change 2. **Pattern 3: Search dominating.** For most SG B2B and considered-purchase accounts, Search converting at 60 to 80 percent of conversions is normal and healthy. For e-commerce accounts, expect Shopping to dominate. The actionable takeaway: with channel-level data, you can finally make per-channel optimisation decisions. Tweak creative for the channel that is underperforming. Adjust audience signals to push spend toward the channel that converts.

Element	2024 setup	2026 setup
Search themes per asset group	Up to 25, often diluted	Up to 50, segmented by intent stage
Videos per asset group	1 to 2 (often landscape only)	6 to 15 across 9:16, 16:9, 1:1
Channel reporting	None (black box)	Per-channel spend, conversions, ROAS
Search term visibility	Partial categories only	Full triggering term list
Negative keywords	Few hundred via support request	10,000 campaign-level, self-serve
Demand Gen relationship	Competing for budget	Sibling channel with shared learning
Conversion expansion	Forced on by default	Controllable via value rules
Asset group naming convention	By product line	By intent stage + product line

Change 4: Full Search Term Visibility (Audit Your Waste)

Until 2025, PMax search term reporting showed broad categories rather than actual queries. From late 2025, the full triggering search term list became available, similar to standard Search campaigns. This is the single most useful diagnostic you can run on a PMax account today.

The 30-minute audit:

1. Pull search terms from the past 90 days, sorted by spend descending.
2. Tag each term as: relevant, marginally relevant, irrelevant, branded competitor, existing-customer query.
3. Add every irrelevant term to the campaign-level negative list (now up to 10,000).
4. Add branded competitor terms only if your brief is to compete for them; otherwise negative them.
5. Add existing-customer queries (people searching for your brand or login pages) if you have separate retention campaigns serving them.

For an SG e-commerce client this audit recovered SGD 3,200 a month in wasted spend that was being burned on irrelevant queries the system had been broad-matching into. Same conversion volume, lower cost, ROAS lifted 18 percent overnight.

The SG-specific catch: PMax often broad-matches SG search themes to Malaysia and Indonesia variants. Check your search terms for non-SG geographic phrasing slipping through ("malaysia", "kl", "kuala lumpur", "jakarta", "indo") and add as negatives if your campaign is SG-only. We see this on roughly half the SG accounts we audit.

Change 5: 10,000 Campaign-Level Negatives Self-Serve

Linked to Change 4. You no longer need to file a Google Ads support ticket to add bulk negatives to a PMax campaign. The 10,000-keyword self-serve limit is now in the standard interface.

What to put in your bulk negative list:

• All competitor brand names (unless you are deliberately competing for them)
• All "free" "cheap" "DIY" prefix queries if you are premium-positioned
• All employment queries ("jobs", "careers", "salary")
• All informational queries that do not convert ("what is", "how does", "definition")
• All existing-customer queries (your brand + "login", "support", "contact", "account")
• All non-SG geographic terms if you are SG-only

We maintain a 1,200-term SG-specific negative starter list for our PMax clients. New accounts get this list applied on day one, which removes the first 60 to 90 days of compulsory waste-burning that PMax usually goes through during its learning phase.

Change 6: Demand Gen Integration and Conversion Expansion Controls

Demand Gen, the relaunch of Discovery campaigns, sat awkwardly next to PMax through 2024. The two campaigns competed for similar inventory and often cannibalised each other inside the same account. Mid-2025 changes positioned Demand Gen as a sibling rather than a rival: PMax handles the bottom-of-funnel, conversion-focused work; Demand Gen handles the top-of-funnel, audience-building work; conversions feed shared signals.

The practical setup for SG accounts in 2026:

• Demand Gen for awareness and audience-building campaigns. Optimise for video views, engagements or first-touch conversions. Visual-led, persona-targeted.
• Performance Max for conversion-focused campaigns. Optimise for purchases, leads or sign-ups. Asset groups segmented by intent stage.
• Standard Search for high-intent, brand and competitor queries that you want exact-match control over.

The conversion expansion controls are the second important 2026 addition. PMax used to forcibly expand your conversion definitions to whatever Google's system thought might be a soft conversion. From mid-2025 you can override this through value rules: tell PMax which conversions to count, which to discount, and which to exclude entirely. SG B2B advertisers in particular benefit from value-rule control because PMax otherwise tends to chase form-fill conversions that are not real qualified leads.

For broader context on how PMax fits into a wider SG paid stack, see our SEM service overview.

SG-Specific Benchmarks for PMax in 2026

Approximate SG benchmarks we see across the accounts we manage, segmented by vertical. These are directional ranges to compare against your own account, not promises.

• B2B SaaS (lead gen): CPL SGD 80 to SGD 250, conversion rate 2 to 5 percent, ROAS 3 to 6x assuming SGD 1,500 LTV
• E-commerce (consumer goods): ROAS 4 to 8x, CPC SGD 0.80 to SGD 2.20, conversion rate 1.5 to 4 percent
• Local services (home, professional): CPL SGD 25 to SGD 120, conversion rate 3 to 8 percent, ROAS depends on close rate
• Education and training: CPL SGD 40 to SGD 180, conversion rate 2 to 6 percent, depends on programme price
• Healthcare and clinics: CPL SGD 35 to SGD 150, conversion rate 3 to 7 percent, regional clinic queries strongest

Accounts performing meaningfully below these ranges almost always have one of three issues: stale 2023-era PMax structure, weak creative supply (fewer than 10 images, fewer than 3 videos), or audience signals built from 1st-party data that has gone stale. The 2026 features above plus a creative refresh typically closes the gap.

The PMax 2026 Re-Audit Checklist

If your SG account has not been touched against the 2026 features, run this checklist. Each item is a 30 to 60 minute task; the full checklist takes a half-day for most accounts.

Asset group re-architecture:

• Audit current asset groups against the 50-theme allowance
• Split diluted asset groups into intent-stage segmented siblings
• Add SG-specific phrasing variants to search themes

Creative supply check:

• Count videos per asset group, target 6 to 15 against the 15 limit
• Confirm 9:16, 16:9 and 1:1 aspect ratios are all present
• Refresh image assets if any are older than 12 months

Reporting and waste audit:

• Run channel-level report, identify over-allocation patterns
• Pull 90-day search terms, tag relevance, add negatives
• Confirm campaign-level negative list utilisation (target 200 to 1,000+ entries)

Conversion architecture:

• Review conversion definitions, exclude soft conversions you do not want PMax chasing
• Set value rules to discount low-quality lead types
• Confirm offline conversion import is wired up if you have a CRM

Demand Gen pairing:

• Decide if Demand Gen sits alongside PMax for top-of-funnel
• If yes, define the audience-handoff between the two campaigns
• Review for cannibalisation if both are running

For broader strategy on how PMax fits into a multi-channel SG paid stack, our piece on digital marketing trends Singapore 2026 covers the full year ahead. For the AI workflow angle on managing accounts at this scale, see the AI marketing workflow that scales.

Common Mistakes SG Advertisers Make in 2026

Mistake 1: Still running 2023-era asset group structure. Diluted intent, single landscape video, no SG-specific search themes. The fastest 30 percent ROAS gain in PMax for most SG accounts is a structural re-audit.

Mistake 2: Ignoring channel-level reporting. The black box is over. If you are not looking at per-channel spend, you cannot diagnose creative, audience or budget allocation issues.

Mistake 3: Not maintaining a serious negative keyword list. Most SG accounts we audit have under 50 negatives. The new 10,000 limit is there to be used.

Mistake 4: Treating Demand Gen and PMax as rivals instead of siblings. They serve different funnel stages. Run both, define the handoff, share the conversion signals.

Mistake 5: Letting PMax chase soft conversions. Use value rules to discount or exclude the low-quality lead types. PMax will optimise for whatever you tell it to count.

Mistake 6: Underutilising the video allowance. 13 of 15 unused video slots is the single biggest waste in PMax 2026. Even rough 9:16 product shots beat empty slots.

Mistake 7: Not re-auditing on a quarterly cadence. Google ships PMax features faster than most SG agencies update their playbooks. A 90-day re-audit cadence is the new normal.

Frequently Asked Questions

What changed in Google Performance Max in 2026?

Six material changes: search themes doubled from 25 to 50 per asset group, video assets tripled from 5 to 15 per asset group, channel-level reporting is now available (showing per-channel spend), full search term visibility ended the black box, campaign-level negative keywords expanded to 10,000 self-serve, and Demand Gen now integrates as a sibling channel with conversion-expansion controls. Combined, these changes typically lift ROAS 20 to 40 percent for SG accounts that re-audit against the new features.

How many search themes should I use in Performance Max in 2026?

Use the full 50 per asset group, but segment by intent stage rather than dumping everything in one asset group. Group themes by category research queries, comparison queries, jobs-to-be-done queries and competitor-alternative queries. For SG accounts, also add explicit Singapore-specific phrasing variants ("in singapore", "sg", "near me singapore"). Google's automation does not always favour SG variants on its own.

Can I see what search terms are triggering my Performance Max ads?

Yes, since late 2025. The full triggering search term list is now available in the search terms report, similar to standard Search campaigns. This is the single most useful diagnostic on a PMax account. Pull 90 days, tag relevance, build a negative list. Most SG accounts recover SGD 1,500 to SGD 5,000 a month in waste through this single audit.

Should I run Demand Gen and Performance Max together?

Yes, and they are now designed to work as siblings rather than competing campaigns. Use Demand Gen for top-of-funnel audience-building and awareness (optimise for engagement, video views, first-touch). Use PMax for bottom-of-funnel conversion (optimise for purchases or qualified leads). The handoff is via shared conversion signals. Running them together on a clear funnel division produces better total-account ROAS than running PMax alone.

What is the right budget for Performance Max in Singapore?

PMax has a soft minimum of around SGD 1,500 a month before its automation has enough data to optimise meaningfully. Below that, standard Search and Shopping outperform PMax. From SGD 1,500 to SGD 5,000 a month, PMax can be the primary acquisition channel. Above SGD 5,000, you should be running PMax alongside dedicated Search campaigns for high-intent and brand queries, plus Demand Gen for top-of-funnel audience work.

How often should I re-audit my Performance Max account?

Quarterly. Google ships PMax feature changes faster than most SG agency playbooks update. The accounts we see falling behind are the ones on annual review cycles. A 30 to 60 minute quarterly audit covering search themes, video supply, channel-level allocation, search term waste and negative keyword utilisation keeps the account aligned with current best practice and typically recovers meaningful budget every quarter.

Related reading

• Digital marketing trends Singapore 2026
• The AI marketing workflow that scales
• Performance Max campaign basics