Main area
Agency
sea

Understanding Smart Bidding — when tCPA, tROAS or Maximize Conversions makes sense

Redaktion ·

Why bidding strategy makes or breaks an account

The most common reason a Google Ads account “just doesn’t perform” isn’t the wrong keyword and isn’t a weak ad copy — it’s the wrong bidding strategy at the wrong time. A shop with 12 conversions per month set to tROAS 400 % will systematically underperform because the algorithm simply doesn’t have the data. An established account doing 800 conversions per month still on Manual CPC is leaving double-digit efficiency gains on the table.

Smart Bidding is powerful, but not autopilot. If you understand the mechanics — what the algorithm needs, what it evaluates, where it fails — you pick deliberately and notice early when something is off. This article walks through the five most important strategies, the typical pitfalls, and three worked scenarios for accounts of different sizes.

The basic mechanic: what Smart Bidding actually does

In every Google Ads auction the algorithm decides in milliseconds how much to bid for this exact click. With manual bids the answer is simple: whatever you set on the keyword. With Smart Bidding the system evaluates hundreds of signals per auction — device, time of day, location, browser, operating system, prior user behaviour, demographics, query type, even the precise search term behind a broad-match keyword.

From those signals the model estimates the conversion probability for this single impression and derives the appropriate bid. A click with an 8 % conversion probability is four times as valuable as one with 2 % — and gets a four-times higher bid accordingly. That is the core: Smart Bidding does not trade in clicks, it trades in expected conversions.

What the algorithm absolutely needs

Three things are non-negotiable:

  1. Working conversion tracking — clean, no double-counting, with clearly defined primary conversions. Without conversion data the algorithm learns nothing.
  2. Enough conversion volume. Google’s rule of thumb is 30 conversions in 30 days for tCPA, 50 conversions in 30 days for tROAS. Below that, the learning phase becomes a lottery.
  3. Stability. Any major change — new tracking, new conversion value, daily budget shifted by more than 20 % — triggers a fresh learning phase, during which performance fluctuates for 1–2 weeks.

What the algorithm sees — and what it doesn’t

Smart Bidding sees everything Google knows about a user, except what’s in your CRM. The algorithm doesn’t know whether a lead ended up as a 50 € customer or a 5,000 € enterprise account. It doesn’t know whether a click came from a country you don’t ship to. It doesn’t know that your margin is 60 % on product A and 8 % on product B. You have to close those gaps via conversion values, value rules, location targeting or custom labels.

The five most important bidding strategies

Google offers a dozen strategies, but five cover 95 % of real-world use cases. Sorted from “click-oriented” to “value-oriented”:

1. Manual CPC — you set every bid yourself

You set a max CPC per keyword, Google never bids more than that in the auction. Pro: full control, transparent, no learning phase, performance swings are explainable. Con: no auction signals, no dynamic adjustment, lots of upkeep. Useful for accounts with low conversion volume (under 10/month), for brand campaigns with a clear price ceiling, and for test phases where you need clean baseline data.

Variant: Enhanced CPC (eCPC) — Google may raise your manual bid by up to 100 % when a conversion looks likely. Effectively a mini Smart Bidding with a manual anchor. Devalued since 2022 because Google loosened the cap on the uplift — the word “enhanced” obscures how far the system can actually go.

2. Maximize Clicks — most clicks within the daily budget

The system spreads the daily budget so that as many clicks as possible happen — without regard for conversions. Useful in awareness campaigns where reach matters more than completion, or in the data-collection phase of a brand-new account that has no conversions yet. Pitfall: Without a CPC cap the algorithm will burn budget on expensive generic keywords. Always set a cap.

3. Maximize Conversions — most conversions within the daily budget

Instead of optimising for clicks, the system optimises for conversions, without you setting a target CPA. You only set the budget — the algorithm handles the rest. Useful for young accounts with 15–30 conversions/month that don’t yet have a clear CPA target, or for lead-gen with a fixed monthly budget that should be spent. Pitfall: the system will spend the entire budget — even if CPA goes through the roof. If you need budget discipline, tCPA is a better fit.

4. Target CPA (tCPA) — most conversions at a target CPA

You set a target CPA (e.g. 25 €), the system tries to deliver as many conversions as possible without exceeding that target on average. “On average” is the key — individual conversions can come in significantly more expensive, the system balances them out with cheaper ones. Useful as soon as you have a stable 30+ conversions/month and all conversions are roughly equally valuable to you.

5. Target ROAS (tROAS) — maximize value at a target return

Here the system optimises not for conversion count but for conversion value divided by ad spend. You set a target ROAS (e.g. 400 % = each €1 of ad spend should bring back €4 of revenue), the system distributes budget so that this target is met on average. Prerequisite: you send a value with each conversion (order value, estimated lead value), and the value spectrum is wide enough that there is something to optimise. For e-commerce with real order values, often the best choice. For lead-gen with flat-rate values, usually overkill — tCPA is enough.

Five pitfalls that hit almost every account

Conversion tracking is the Achilles’ heel

Smart Bidding is only as good as the conversion data it trains on. A double-firing tag that reports every order as two conversions halves the algorithm’s perceived CPA — it bids too aggressively and burns budget. A missing cross-domain configuration loses a third of conversions — it bids too defensively and gives up volume. Before any switch to Smart Bidding: verify the conversion setup at 100 %, not 80 %. Enhanced Conversions and server-side tagging are no longer nice-to-have in 2026 — they’re table stakes for clean Smart Bidding performance.

Last-click skews the learning signal

If your account still runs on last-click attribution, the algorithm only sees the final touchpoint before conversion — typically brand clicks and direct-type-in equivalents. Generic upper-funnel keywords get systematically undervalued. Smart Bidding then optimises for what it sees and pushes generic volume out, even though it really contributes. Data-driven attribution has been the default in Google Ads since 2023 — if you’re still on last-click, switch before you set up Smart Bidding.

Respect the learning phase — don’t panic-tweak

After activation or major changes, performance fluctuates for 7–14 days. If you halve the target CPA after three bad days in a row, you force the system into a new learning phase and prolong the noise. Rule: don’t intervene for at least two weeks, then evaluate weekly, not daily. Make larger target-CPA changes in steps of no more than 15–20 %.

Quality Score doesn’t become irrelevant

A common myth: “with Smart Bidding, Quality Score doesn’t matter.” Wrong. Quality Score still flows into Ad Rank and decides how expensive a given conversion goal becomes. Smart Bidding compensates for weak Quality Scores with higher bids — which you see in the daily report as elevated CPCs. If you also work on ad copy, landing page and keyword relevance, you lower the real CPA at the same tCPA target.

Seasonality and sales kill the learning phase

Black Friday, summer sale, a promo weekend — anything that strongly shifts conversion rate in the short term confuses the algorithm. Solution: enter Seasonality Adjustments in Google Ads ahead of time, with the time window and the expected conversion-rate lift. The system then distinguishes between normal learning and a seasonal special effect.

Decision guide — which strategy when

| Situation | Recommendation | Why | |---|---|---| | New account, < 10 conv./month | Manual CPC + Maximize Clicks | No learning signal yet, collect data first | | 10–30 conv./month | Maximize Conversions (with CPC cap) | Algorithm starts learning, budget stays bounded | | 30–50 conv./month, lead-gen | tCPA | Enough volume, conversions are equivalent | | 50+ conv./month, e-commerce with order value | tROAS | Value asymmetry gets exploited | | Brand campaign | Manual CPC or tCPA with a low target | Cap it — Smart Bidding tends to overpay here | | Seasonal peaks | Existing strategy + Seasonality Adjustments | Don’t destroy the learning phase | | Awareness, no conversion goal | Maximize Clicks (with cap) | Clicks are the goal, conversions secondary |

Three worked examples from practice

Example 1: B2B lead-gen with 18 conversions/month

Assumptions: Lead-magnet campaign, conversion = whitepaper download, average CPC 2.40 €, conversion rate 4 %, daily budget 50 €.

  • Manual CPC: 50 € / 2.40 € = 20.8 clicks/day → at 4 % CVR ≈ 25 conversions/month at CPA 60 €. Stable but rigid.
  • Maximize Conversions: System optimises for the cheapest clicks with the highest CVR probability. In practice ≈ 30 conversions/month at CPA 50 €. The 2-week learning phase swings between 15 and 70 € CPA.
  • tCPA at 50 €: Not recommended. 18 conversions/month is below the 30-threshold, the algorithm doesn’t get enough learning signal. Realistic outcome: the system lands at 8–12 conversions/month at CPA 80 €+ because it bids defensively to keep the 50 € target.

Recommendation: Maximize Conversions until 30+ conversions run stable, then switch to tCPA.

Example 2: e-commerce shop with 200 orders/month

Assumptions: average order value 85 €, post-COGS margin 38 %, daily budget 400 €, current ROAS 350 %, conversion tracking incl. order value clean.

  • tCPA at 18 € (= current CPA): system optimises for count, ignores order value. Result: 290 orders/month, average order value drops to 72 €, revenue 20,880 €, ROAS 174 %. More conversions, less profit.
  • tROAS at 400 %: system prioritises clicks expected to bring high order values. Result: 175 orders/month, average order value 110 €, revenue 19,250 €, ROAS 401 %. Fewer orders, cleaner profit.
  • tROAS at 500 %: too aggressive. Algorithm throttles volume hard: 95 orders, revenue 11,400 €. ROAS 502 % — at half the volume.

Recommendation: tROAS at 400 % (or the current value); don’t ramp ROAS quickly — in 10 % steps every 2–3 weeks.

Example 3: small local account with 6 conversions/month

Assumptions: local service business, conversion = call click, 6 real calls/month, average CPC 1.80 €, daily budget 15 €.

  • tCPA, Maximize Conversions, tROAS: all three fail. Volume is enough for none of these Smart Bidding strategies. The algorithm sees 6 data points in the learning window — that’s statistical noise.
  • Manual CPC with targeted bids on 8–12 strong long-tail keywords plus a tight geo zone (10 km radius). Result: ≈ 8 conversions/month, CPA stable at ≈ 55 €, full control.
  • Alternative: go to Performance Max with conversion tracking incl. calls + forms + directions to artificially generate more data points. Often works, but for a tiny account it’s hard to control.

Recommendation: Manual CPC until the account hits at least 20 conversions/month. Before that, Smart Bidding is an expensive experiment.

FAQ

Are 30 conversions really enough for tCPA?
Enough to start. It only becomes truly stable at 50–80 conversions/month. Below 30, tCPA is a gamble.
Can I combine tROAS and tCPA?
Not within a single campaign. But: you can run campaigns of similar account structure with different strategies — tROAS for top products, tCPA for category campaigns, Manual CPC for brand.
How often can I change the target CPA / target ROAS?
At most every 2 weeks, in steps of no more than 15–20 %. More often forces the system into permanent learning phases.
What if the daily budget is too small?
Smart Bidding needs 20 % budget headroom above demand to learn cleanly. A budget that's gone every day by 11 a.m. pushes the system into an emergency mode that compromises learning.
Does Smart Bidding help on brand keywords?
Rarely. Brand has high CVR and low CPC — the algorithm spots that and bids too high because it knows: "I basically win this auction either way". Manual CPC with a clear cap is often cheaper here.

Conclusion

Smart Bidding is the default choice in 2026 for any account that meets the data baseline — clean conversion tracking, data-driven attribution, at least 30 conversions/month. If you have those prerequisites and still stay on Manual CPC, you’re leaving double-digit efficiency on the table.

If you don’t have those prerequisites, don’t force yourself into a Smart Bidding strategy just because it feels modern. Manual CPC with diligent keyword work beats an under-fed tCPA algorithm any day. The switch point is clear: from 30 conversions/month onwards go to Maximize Conversions or tCPA, from 50+ with order values go to tROAS — always with patience in the learning phase, well-maintained tracking and not too jumpy adjustments to the target.

If you also work on ad copy, landing page quality and Quality Score, Smart Bidding squeezes even more out of the same budget. The strategy wins the auction — the substance behind it you still have to build yourself.

Themenuebersicht

Search Engine AdvertisingGebotsstrategien