Main reference

Bid Strategies Cheat Sheet — Ecommerce

Built from Google's documentation, the Croud UK Q2 2024 playbook, Savvy Revenue (Andrew Lolk), Brainlabs, Tinuiti, Optmyzr and Smarter Ecommerce.

📎 Companion documents: ROAS Target Calculator · Campaign Reorganisation Playbook · Andrew Lolk transcript · Croud PDF notes

Caveats — what's verified vs rule-of-thumb

A few claims here are practitioner consensus, agency folklore or industry estimates rather than Google-verified fact. They're useful as defaults, but you should know which is which before quoting them:

Claim	Confidence
"Smart Bidding needs 30–50 conv/month to settle" — Google's actual minimum is 15 conv/30d for tROAS and tCPA. The 30–50 figure is where it stabilises in practice.	practitioner
"Portfolio bid strategies share learning across campaigns" — agency folklore that Google's docs don't support. Real benefit is volume aggregation; Smart Bidding already learns at the query level account-wide.	myth
"+14% conversion value lift from tCPA→tROAS" and "+18% unique queries from Smart Bidding Exploration"	Google-published
2025 ecommerce ROAS ~2.87:1, CPA up ~12% — reported in industry roundups, primary source not re-verified.	industry-reported
"New customer multiplier 2.5–3.5×" — case-study range from agency blogs. Use your own (CLV ÷ first-order AOV) ratio.	rule of thumb

Part 1 — Foundations

Quick decision tree

Start at the top, take the first branch that fits.

Under £2K/month spend OR <30 conversions/month?

→

Manual CPCSmart Bidding can struggle on small budgets in learning

Brand campaign?Branded search defending your own name

→

tROAS 800–1500%Or Target Impression Share 90–95% absolute top, with max CPC cap

Pure traffic / awareness — no conversion data yet?

→

Maximize ClicksWith a max CPC cap so it can't go feral

Have conversion values and 15+ conversions/month?

→

Target ROASThe default for ecommerce in 2026

15+ conversions/month but no reliable values?

→

Target CPABridge to tROAS once values are clean

Scaling / plateau-breaking / peak ramp / post-incident reset?

→

Max Conversion ValueTemporarily, then return to tROAS

Video, display, or pure brand awareness?

→

CPV or CPMCPV for video views, CPM for display impressions

Default for ecommerce: Target ROAS. Everything else is either a stepping stone, a temporary unstuck-the-account move, or an exception.

Complete strategy taxonomy

Smart Bidding (auction-time, AI-driven)

Strategy	Use when	Avoid when	Volume floor
Target ROAS	You have conversion values and want a profitability floor. Default for ecommerce.	No value tracking, or target set fantasy-land far from reality	Google min: 15/30d. Reliable: 50+
Target CPA	You have counts but values are unreliable	You can pass real values (use tROAS)	Google min: 15/30d. Reliable: 30+
Max Conversion Value	Scaling / plateau / peak ramp / post-incident reset	Steady state (no efficiency floor — see Mistake #2)	15–30+/month
Max Conversions	Lead-gen with no value tracking, growth-only goal	Ecommerce — use a value-aware strategy	15–30+/month
Enhanced CPC (eCPC)	Historical only	Effectively deprecated — removed from Search 2022, Display/Shopping 2024	N/A

Non-Smart automated

Strategy	Use when	Avoid when
Target Impression Share	Brand defence campaigns. 90–95% absolute top with max CPC cap	Performance-driven campaigns — it ignores conversions
Maximize Clicks	Pure traffic / new campaign with no conversion data	You have conversion data — use Smart Bidding

Manual & video

Strategy	Use when
Manual CPC	New accounts, <£2K/month, <30 conv/month, surgical control
CPV	YouTube video — pay per view
CPM / vCPM	Display + YouTube brand awareness

Prerequisites — Google's official guidance

Strategy	Tracking	Conv. volume	Conv. value	Notes
Target ROAS	Required	Min 15/30d, prefer 50+	Required	100+ very reliable
Target CPA	Required	Min 15/30d, prefer 30+	—	Lower floor than tROAS
Max Conv Value	Required	15–30+	Required	No efficiency floor
Max Conversions	Required	15–30+	—	No efficiency floor
eCPC	Required	N/A	—	Effectively deprecated
Manual CPC	Optional	Any	—	Best for small/new

Part 2 — Ecommerce playbook

PMax vs Standard Shopping in 2026

Both. Run them side by side.

PMax averages +18% conversions at the same CPA vs Shopping Google. Accesses Search, YouTube, Display, Discover, Gmail, Maps from one campaign.
Standard Shopping still wins on branded and high-intent product searches, with cleaner search-term reporting.
2025–2026 PMax updates: channel-level reporting, full search term reporting, negative keywords (up to 10K), 50 search themes per asset group.

Recommended architecture

Search — Brand
Search — Non-Brand
Shopping Standard — Brand
Shopping Standard — Non-Brand
Performance Max — Prospecting (with brand exclusions)
Performance Max — Retargeting

~6 campaigns total, grouped under 2–3 portfolio bid strategies.

2025 ecommerce benchmarks (reported, not verified): Blended ROAS ~2.87:1, median CPA up ~12% YoY to ~$24. Use as a rough sanity check, not as a target — derive your tROAS from your own contribution margin, see the ROAS Calculator.

Margin-tier segmentation — the highest-ROI structural move

The problem: a single tROAS across mixed margins is almost always wrong. A 50%-margin accessory and a 15%-margin bulky item can't profitably share a target.

The fix is segmentation by margin tier using custom labels in Merchant Center:

Populate custom_label_0 with high-margin, medium-margin, low-margin
Use Merchant Center attribute rules to auto-assign based on cost_of_goods_sold vs price — scales with the catalogue, updates automatically
Create separate Shopping or PMax campaigns per tier
Each tier gets its own tROAS reflecting that tier's actual contribution margin (see ROAS Calculator)
Group per-tier campaigns under 2–3 portfolio bid strategies

Sources align: Savvy Revenue explicitly endorses margin-tier segmentation. Croud's position is more nuanced — "consolidate campaigns, inject margin as a value signal via cost_of_goods_sold" is their default, but their PMax guidance says "if profitability is key, separate campaigns to target low and high margin products with more aggressive ROAS targets on the latter." Use segmentation as the default if you don't have a reliable path to push per-SKU margin data into Google/SA360.

Feed quality — the silent performance lever

Per Croud, feed quality is one of the biggest performance levers for Shopping and PMax.

Use a feed management tool (DataFeedWatch, Channable, Productsup, Feedonomics, Shopify Feed for Google)
Push cost_of_goods_sold into the feed — this unlocks margin-aware bidding via SA360 and POAS tools
Use Google's structured title/description fields where available
Run a Zombie SKU rescue (see Advanced)

Part 3 — The 6 Smart Bidding mistakes Savvy Revenue

An audit checklist you can run against your own account. Source: Andrew Lolk, Savvy Revenue. Full transcript banked here.

1Running campaigns limited by budget

Two failure modes.

Failure A — capped budget creates artificial ROAS

When a campaign is "limited by budget", Smart Bidding cherry-picks the most profitable clicks. You see ROAS at 700% against a 500% target and feel great. Increase the budget, ROAS drops, you panic. Nothing was wrong — Google was artificially efficient because it was constrained. You're not choosing the performance, you're accidentally getting there.

Fix: set the ROAS target you actually want. Budget should be 20–50% higher than current daily spend so it's not the constraint.

Failure B — using budget as a spend control lever

Lowering daily budget when you need to spend less forces all the same cherry-picking problems. To reduce spend correctly:

Lever	Effect	Use when
1. Increase ROAS target	Long-term, less immediate	Strategic / sustained
2. Negative seasonal bid adjustment	Immediate	3–7 days only
3. Decrease max bid limit	Immediate	Underrated lever
4. Lower daily budget	Immediate but breaks Smart Bidding	Genuine emergencies only

2Using Max Conversion Value instead of tROAS

Max Conv Value gives Smart Bidding only one constraint: budget. With variable demand, that creates the wrong behaviour:

What Max Conv Value does

Low-demand days: raises CPCs to spend the budget.
High-demand days: lowers CPCs to stay within budget.

What you actually want (tROAS)

Buys every profitable auction at the target. Spends more on high-demand days, less on low. Doesn't care about volume.

The 3 scenarios where Max Conv Value is correct:

Hard budget client (corporate, can't move budget between months)
Bypassing a tROAS target that's choking spend (temporary unstucker)
Resetting after a tracking incident or at the start of peak season

In 99% of ecommerce cases: use tROAS.

3Only using campaign-level targets

Per-campaign tROAS feels logical but starves Smart Bidding of data when campaigns are small.

Fix: combine campaigns with similar targets into portfolio bid strategies. The primary benefit is volume aggregation above the learning threshold — small campaigns that would starve on their own cross the line when pooled with siblings.

⚠ Important — what portfolios do NOT do: Agencies routinely claim portfolios "share learning" or "cross-pollinate signals" between campaigns. Google's official portfolio documentation does not use that language. Google separately documents that Smart Bidding already learns at the query level across your entire account — that pooling happens with or without a portfolio. Portfolios aggregate volume and unify targets; they don't unlock a separate hidden learning mechanism.

What a portfolio actually gives you (verified against Google docs):

One shared target applied across all assigned campaigns
Aggregated conversion volume toward the learning threshold — the real reason to use one
Min and max bid limits (varies by strategy type)
Portfolio-level top-signals reporting in the Shared Library

Practical implication: if a campaign already has 30+ conversions/month on its own, forcing it into a portfolio doesn't necessarily improve performance — the portfolio optimises to a portfolio-wide average, which can mask underperforming pockets. Both Optmyzr (Fred Vallaeys) and Savvy Revenue have noted this.

Use a portfolio when

Multiple small/low-volume campaigns. Multiple campaigns governed by one identical target. You want shared bid ceilings/floors.

Avoid a portfolio when

Each campaign already has 30+ conv/month. Campaigns have genuinely different economics (different margins, intents).

Sweet spot for ecommerce: 2–3 portfolios. 500 vs 600 vs 590 is not a meaningful difference. Consolidate aggressively. Only split when targets differ by 30%+.

4Static ROAS targets

Setting tROAS in January and never touching it. Reframe: tROAS is a communication tool. Lower it to tell Smart Bidding "explore, find me volume". Raise it to say "tighten up, be efficient".

Situation	Direction	Why
Before peak season	↓ Lower	Get into expansion mode before demand spikes
After a sale/promo	↑ Raise	Smart Bidding is bidding off inflated sale data; tighten
Growth flatlining	↓ Lower	Too efficient, sitting in a small auction pool
Account never getting volume	↓ Lower temporarily	Force exploration; revert when volume returns

Cadence: 3–7 strategic adjustments per year. Not weekly. Targets are not set-and-forget.

5Ignoring Projected ROAS and Average Target ROAS

The mistake that separates experienced managers from everyone else. Most people compare last 30d actual ROAS to current target. Wrong.

Two metrics most managers don't use:

Projected ROAS — what Smart Bidding is actually steering on. Actual ROAS doesn't include conversions still arriving from lag. Hover over actual ROAS and Google shows projected. If projected matches your target, Smart Bidding is doing its job — even when actual looks low.

Croud independently confirms this: "Conversion delay is a key consideration when assessing the success of smart bidding."

Average Target ROAS — the time-weighted average of your target across the reporting window. If you raised the target last week but the period is 30 days, the "target" the algorithm has been chasing is the average, not your current setting.

The correct comparison: Projected ROAS vs Average Target ROAS. If they match, Smart Bidding is doing exactly what you asked. Don't intervene.

6Tinkering too much

Yes, you should make changes. No, not every week. Tinkering looks like:

Changing the target weekly
Daily seasonal bid adjustments
Weekly max-bid-limit fiddling
Daily data exclusions for normal variance

Andrew Lolk: "I've seen advertisers who were constantly making changes, went on vacation, and performance got better."

The most powerful Smart Bidding skill is reading what the algorithm is doing and having the confidence to do nothing. That's expertise, not neglect.

Bonus mistake: Avoid ad-group-level ROAS targets. In 99% of cases, they're micro-constraints that just limit Smart Bidding. Use campaign or portfolio level instead.

Part 4 — Advanced levers

Portfolio bid strategies — beyond the basics

Already covered in Mistake #3, including the "portfolios don't magically share learning" point. Two extras:

Bid limits inside a portfolio protect against CPC creep. Savvy Revenue cite this as one of the highest-ROI interventions they make. Availability varies by strategy: tCPA portfolios expose min/max CPC bid limits; some tROAS / Shopping configurations expose max CPC at the campaign level instead.
Multi-target Portfolio Bid Strategies in Search Ads 360 (Q3 2025) — flagged by Brainlabs as a meaningful advancement.

Seasonality adjustments

Tell Smart Bidding to expect a temporary change in conversion rate (promo, sale, flash event). You enter the expected % change in conversion rate during the window.

Use only for short events: 1–7 days, max 14
Input is the expected conversion rate change (e.g. "+30% during the sale"), not a CPC change
Don't use for regular seasonal cycles Google already knows about (Christmas, Black Friday)
Maintain a calendar of upcoming promo events
Available for Search, Shopping, Display, PMax

Secondary use — spend control (Savvy Revenue): apply a negative seasonality adjustment for 3–7 days to temporarily reduce spend without touching budget or targets. Niche trick, not the primary purpose.

Data exclusions

Tell Smart Bidding to disregard data from a specific time window — used when something broke (tracking outage, site downtime, pixel firing wrong).

Safety net only, not routine
Don't extend beyond the actual incident window
Critical after a tracking incident — otherwise Smart Bidding learns from bad data and degrades for weeks

Conversion Value Rules

Adjust the value Google sees for a conversion based on conditions: location, device, audience, new vs returning customer.

Multiply value for new customers — set the multiplier to your actual (CLV ÷ first-order AOV) ratio, not a default
Reduce value for low-margin segments
Boost value for high-LTV segments

Smart Bidding Exploration Google 2025

Biggest bidding update in a decade per Brainlabs. Google can temporarily lower your effective ROAS target to explore new search queries it would otherwise miss.

Currently: Search campaigns with Target ROAS only
Google reports: +18% unique search query categories, +19% conversions
Test in well-established campaigns. Use Drafts & Experiments for A/B testing.

Value-Based Bidding — the 2025–2026 default shift

Industry has moved from tCPA to tROAS as the default. Google data: advertisers switching from tCPA to tROAS see median +14% increase in conversion value Google.

The next evolution is POAS (passing profit, not revenue, as the conversion value) — covered in ROAS Calculator §6.

Zombie SKU rescue Croud

Products without enough impressions for Smart Bidding to learn on get further deprioritised — a vicious cycle. Fix: temporarily move zombie products into a dedicated campaign with more aggressive bidding to force them to serve and gather data. Once they have a baseline, re-merge.

The learning period — what resets it

Typically 7–14 days, needs ~50 conversions or 3 conversion cycles.

What resets learning

Bid strategy change · Significant budget change · Adding/removing campaigns from a portfolio · Pausing & unpausing · Changing conversion actions

How to behave during learning

Don't tinker. Adjustments max 10–20% per change. Watch projected ROAS, not actual. Wait the full 14 days.

Part 5 — Transitioning between strategies

Move	When ready	How
Manual CPC → Smart Bidding	30+ conv/month, ~£3K+/month	Max Conversions or eCPC for 2 weeks, then move to tCPA/tROAS
Max Conversions → tCPA	Stable conversion volume	Set tCPA at recent average CPA, not a fantasy lower number
tCPA → tROAS	Conversion values reliable	Set tROAS at recent average ROAS, then increment
Max Conv Value → tROAS	Spend stabilised after a reset	As above

Rules for all transitions: Move in 10–20% steps max · Allow full 14-day learning per change · Don't change anything else simultaneously · Watch projected ROAS, not actual.

Part 6 — 2025/2026 updates

Update	What changed	What it means
Smart Bidding Exploration	Google can lower effective ROAS to find new query categories	Test in stable Search + tROAS; +18% unique queries reported
AI Max for Search	Broader AI keyword matching	Pair with Smart Bidding; review search terms regularly
PMax channel-level reporting	You can finally see where PMax is spending	Audit PMax inventory split
PMax negative keywords (10K)	Add brand or irrelevant terms	Fix brand cannibalisation
PMax 50 search themes	Steer toward intent	Use sparingly
Manual CPC UI refresh (2026)	New pro-level controls	Manual CPC still relevant for small/new accounts
Most bid adjustments sunset	Only device exclusions, tCPA device modifiers, seasonality, data exclusions still work	Stop expecting old bid modifiers to do anything

Bid Strategies Cheat Sheet — Ecommerce

Caveats — what's verified vs rule-of-thumb

Part 1 — Foundations

Quick decision tree

Complete strategy taxonomy

Smart Bidding (auction-time, AI-driven)

Non-Smart automated

Manual & video

Prerequisites — Google's official guidance

Part 2 — Ecommerce playbook

PMax vs Standard Shopping in 2026

Recommended architecture

Margin-tier segmentation — the highest-ROI structural move

Feed quality — the silent performance lever

Part 3 — The 6 Smart Bidding mistakes Savvy Revenue

1Running campaigns limited by budget

Failure A — capped budget creates artificial ROAS

Failure B — using budget as a spend control lever

2Using Max Conversion Value instead of tROAS

What Max Conv Value does

What you actually want (tROAS)

3Only using campaign-level targets

Use a portfolio when

Avoid a portfolio when

4Static ROAS targets

5Ignoring Projected ROAS and Average Target ROAS

6Tinkering too much

Part 4 — Advanced levers

Portfolio bid strategies — beyond the basics

Seasonality adjustments

Data exclusions

Conversion Value Rules

Smart Bidding Exploration Google 2025

Value-Based Bidding — the 2025–2026 default shift

Zombie SKU rescue Croud

The learning period — what resets it

What resets learning

How to behave during learning

Part 5 — Transitioning between strategies

Part 6 — 2025/2026 updates

Part 7 — Further reading

Google official verified primary source

Savvy Revenue (Andrew Lolk) best ecommerce source

Croud UK in this folder

US agencies + tooling

POAS / margin-based bidding