How to fix: Portfolio bid strategy applied incorrectly

TL;DR

Portfolio bid strategies pool conversion signal across multiple campaigns that share one goal, which can lift campaigns that individually sit below Smart Bidding's 30 conversions per 30 days learning gate. The utility has narrowed in 2024-2025 as campaign-level Smart Bidding matured, so the question now is rarely "should I use portfolio" and more often "do I have a specific pooling reason". Group campaigns into a portfolio only when they share the same conversion action and target CPA or ROAS and each falls below the learning gate individually. Single high-volume campaigns that already clear 30 conversions per 30 days perform as well or better on a campaign-level strategy [1][2].

Why it matters

A portfolio bid strategy is a single automated bid strategy — Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value, Maximize Clicks, or Target Impression Share — applied across multiple campaigns instead of being attached to one campaign at a time [1]. The Google Ads algorithm treats the attached campaigns as one optimization unit: it pools their conversion history, learns from their combined signal, and bids toward one shared target across all members.

The narrow case where this helps is pooling under the learning gate. Smart Bidding strategies need 30 conversions in 30 days at the campaign or portfolio level to stabilize, and Target ROAS additionally needs those conversions to carry reported revenue values. A merchant running five product-category Search campaigns that each convert 8 to 12 times per month is below the gate on every campaign individually but clears it comfortably (40 to 60 conversions per month) as a portfolio. Pooling lets the algorithm exit the permanent exploration regime that a sub-threshold solo campaign is stuck in. See Smart Bidding running under 30 conversions per 30 days for the deeper mechanics of that gate.

Outside that pooling case, portfolio bidding is rarely the right answer in 2025. Campaign-level Smart Bidding has matured: tCPA, tROAS, and Maximize Conversion Value at the campaign level now handle most accounts well as long as the campaign clears the learning gate. Three things specifically push merchants away from portfolio bidding:

1. Distinct goals across campaigns. A brand-defense campaign and a non-brand prospecting campaign do not share a CPA — brand converts at 1/5th the CPA. Pooling them forces a portfolio target that is wrong for both. Each should stay on its own campaign-level strategy. The bidding strategy goal alignment article unpacks why goal-specific strategies beat blended ones.
2. Distinct conversion actions. A lead-gen campaign optimizing toward demo bookings and an ecommerce campaign optimizing toward purchases do not belong in the same portfolio — the model would average two unrelated value distributions.
3. One campaign above the gate. If a single campaign already clears 30 conversions per 30 days with reported value, pooling it with smaller siblings does not improve its performance — it constrains it under the portfolio's averaged target.

Portfolio bidding is also unavailable for some campaign types entirely: Performance Max, Shopping, Demand Gen, Hotel, Travel, and App campaigns must run on campaign-level Smart Bidding strategies [2]. Audit findings that recommend "switch to portfolio" on those campaign types are wrong on the face of it.

The other use case worth naming is centralized control over bid caps and constraints in multi-campaign rollouts. If you manage 30 location-based campaigns that should all respect the same maximum CPC ceiling, a portfolio strategy is the only place to set that ceiling once instead of 30 times. That is an operational efficiency gain, not a performance gain — measurable on team time saved, not CPA improvement.

How to fix

1. Confirm the pooling case actually applies. List the candidate campaigns and check: do they share one conversion action? Do they share one target CPA or ROAS? Does each individually fall below 30 conversions per 30 days (or 30 conversions with value for tROAS)? If any of those three is "no", portfolio is not the right tool.
2. Pick the matching portfolio type. Open the Bid strategies library (Tools → Budgets and bidding → Bid strategies) and create a new portfolio of the type that matches the goal — Target CPA for lead-gen with a known cost target, Target ROAS for ecommerce with a revenue target, Maximize Conversions / Maximize Conversion Value when there is no fixed target yet, Target Impression Share for awareness or brand defense, Maximize Clicks for upper-funnel traffic.
3. Set the portfolio target at the pooled level. Do not anchor on the best individual campaign's historical CPA — the portfolio will average across members. Use the weighted average CPA or ROAS across all candidate campaigns over the last 60 to 90 days as the starting target, then refine after learning closes.
4. Attach the qualifying campaigns. Go to each campaign (Campaign → Settings → Bidding → Use a portfolio strategy) and select the portfolio. Leave campaigns with distinct goals or distinct conversion actions on their own campaign-level strategy.
5. Wait out the learning phase. The portfolio re-enters Smart Bidding's 7 to 14 day learning window when first created and again any time a member is added or removed. Avoid budget changes greater than 20% during this window — the model is recalibrating.
6. Audit member balance after 4 weeks. Open the portfolio detail view and check spend by member campaign. If one member is consuming most of the budget and starving the others (a common failure mode for portfolios with mixed campaign sizes), split that member back out to a campaign-level strategy and keep the rest in the portfolio.

Common mistakes

• Putting brand and non-brand into one portfolio. Brand search converts at a fraction of the non-brand CPA, so the portfolio target is mathematically wrong for both members. Keep brand on its own campaign-level strategy.
• Adding a high-volume campaign to a portfolio "for consistency". A campaign that already clears 30 conversions per 30 days does not benefit from pooling. It will spend more time in learning whenever the portfolio's other members churn, and the averaged target will under-serve it.
• Treating portfolio as the default. In 2024-2025, campaign-level Smart Bidding is the default. Portfolio is a structural fix for a specific signal-pooling problem, not a generic upgrade.
• Forgetting the portfolio when changing a campaign. Bid strategy changes made at the campaign level are silently ignored when the campaign is attached to a portfolio. Make the change in the portfolio settings instead.
• Mixing campaign types Google does not support. Performance Max, Shopping, Demand Gen, Hotel, Travel, and App campaigns cannot be in portfolios. Audit findings or third-party recommendations suggesting otherwise are wrong.

FAQ

Should every account use portfolio bidding? No. Most accounts run better on campaign-level Smart Bidding. Portfolio is a targeted fix for accounts running multiple low-volume campaigns toward the same goal.

Does pooling guarantee better CPA? No. Pooling lets the algorithm clear the learning gate it was stuck under, which usually stabilizes CPA. A campaign that was already above the gate sees no improvement.

Can I keep some campaigns on portfolio and others on campaign-level? Yes — this is the recommended setup when only a subset of campaigns matches the pooling criteria.

What if my portfolio target drifts after a few months? Treat it like a campaign-level target. Adjust by 10 to 15% at a time, never more than 20% at once, and accept a fresh 7 to 14 day learning window each time.

Is portfolio bidding deprecated? No. Google still supports it actively. It has simply become more niche as campaign-level Smart Bidding got stronger.

Sources

[1] Google Ads Help — About automated bidding strategies. https://support.google.com/google-ads/answer/2979071

[2] Google Ads Help — About Target ROAS bidding. https://support.google.com/google-ads/answer/6268637