Google Ads Bidding Strategies 2026: Why Manual CPC and Target Impression Share Still Print Money

Manual CPC — the underrated one

Everyone says Manual CPC is dead. Google says it. Bloggers say it. YouTube experts repeat it. Supposedly ML has long outplayed you, and setting a click price by hand is like driving from Kyiv to Lviv on a horse cart: technically possible, but why.

That can be true — for a big account with thousands of conversions a month and airtight tracking. In those environments tCPA and Max Conversions genuinely do the job better than a human. Fair enough.

But for 8 out of 10 accounts I actually see in the wild, Manual CPC still often beats tCPA. Here's why.

tCPA (like Max Conversions and tROAS) feeds on conversion data. Not enough data, or dirty data — and the strategy is firing blindfolded. Google won't tell you "your conversion tracking is broken, I'm pausing this." Google will just keep spending your budget on the learning phase — a month, two, three. The learning phase doesn't finish by magic; it finishes when the model is confident enough. On low volume, that can be never.

Manual CPC is different. You set a bid, and you see what it does. Click too expensive, no conversions — lower it. Click cheap, ROAS 400% — raise it. It's boring, it's manual, it's daily work. But you see money, not "insufficient signal for optimization."

When Manual CPC clearly wins:

• A new account or new campaign with no conversion history. tCPA has nothing to feed on. You'll pay Google to "learn" and it will never reach the point where it starts paying back.
• An account with a questionable landing page. Bad landing = low conversion = the model can't find a pattern = bids swing between the ceiling and the floor. On Manual CPC, at least you're not letting Google experiment on your dime.
• Small budget, high attention. One person, 30–50 keywords, willing to look at everything once a day. Manual CPC gives you control no automation will, because ML needs a volume of data you simply don't have.

This isn't a strategy for life — it's a strategy for a specific phase. Run Manual CPC when either (a) you don't yet have enough data for tCPA, or (b) you're an expert with attention and the account is lean. Otherwise — look at automation. Just without the illusion that automation solves problems you never showed it.

Honest about the downsides so no one comes back with complaints later: Manual CPC has more downsides than upsides. Time. Attention. Hourly work. Every keyword has to stay in your line of sight. You can't take a two-week vacation and not look at bids. You're not using Google's real-time signals — device, time, audience, browsing history — and those genuinely help when there's enough data. The downsides are real. They just matter less in your specific phase than the control does.

Target Impression Share — two non-obvious use cases

This is where I disagree most with the conventional wisdom. Open any bidding guide and Target IS is described as a "defensive crutch for brand." Full stop. As if it's a strategy for cowards who can't do ROI.

That's half the truth. The other half: Target IS is a serious growth lever, if you know where to apply it.

Use case #1: brand. Defense against competitors.

Straightforward. Someone's bidding on "your name" — you need to be on "Absolute top," 90–95% impression share. Budget is small (brand queries convert high and are usually cheap), but the defense is critical. One slip and a competitor grabs a warm audience that was already looking for you. Target IS "Absolute top of page" + 90%+ and a hard CPC cap — and breathe. Classic, everyone knows it.

But that's only half. The other half is more interesting.

Use case #2 (the one everyone misses): high-converting non-brand keywords.

Imagine a campaign where 5–10 keywords deliver 500% ROAS. CTR 12%, CVR two-to-three times the account average, customer LTV through the roof. You know this from data, it's stable, it's been tested.

What most people do: slap tCPA on it and hope. And here's the trap — your target CPA is usually set by account average, but on these top-converting keywords the real auction CPC is higher. tCPA looks at the auction and decides "too expensive for my target" — and silently hands the impression to a competitor who does the math on this group separately. You're literally losing impressions you should have won. Impression Share on that group is 40% and nobody's raising the alarm, because tCPA is "optimizing." Sure. Just not toward that.

Target IS with a high target (75–90% Top of page) on those keywords is literally "don't miss a single impression, because the economics pay for themselves." A 500% ROAS survives an inflated CPC. You maximize coverage where you already know you'll win. That's not defense — that's a growth lever.

Conditions for use case #2:

• At least 30 days of data on these keywords, CVR 2–3× the account average.
• ROAS or CPA good enough that you confidently say "give me more."
• Hard CPC cap. Without it, Target IS will push you into the ceiling in one bad auction day.
• Willingness to check economics every two weeks. If it breaks — roll back to tCPA or Manual CPC.

Google doesn't advertise this use case because it requires understanding your own account. Automation across 500 keywords won't see it. But you — you can.

The 7 bidding strategies in 2026 — a quick tour

A quick pass through all seven so you understand the landscape. Manual CPC and Target IS I already covered above — so the rest, short and to the point.

Manual CPC. You set the click price by hand, Google doesn't interfere. Wins on small accounts, new campaigns, weak landings, test launches. Loses on high volume and when you're not in the account every day.

eCPC (Enhanced CPC). Your bid ± 30% at Google's discretion. A hybrid that's mostly dead now — Google is gradually removing it from new campaigns. If you still have it and it works well, don't touch it. Don't build a new launch on it.

Maximize Clicks. Goal: maximum clicks for your budget. Fine for awareness campaigns, content, remarketing audience seeding. For conversion campaigns — almost always garbage: volume without quality = draining budget. Max Clicks does respect geo — it won't pull traffic from countries you didn't target. But inside your geo it will floor the gas on the cheapest queries: Search Partners, broad match sprawl into off-topic themes, cheap mobile clicks in dead hours, low-CVR informational queries. Clicks will come. Conversions won't. Always set a max CPC cap, or you'll watch your budget bleed into traffic that will never buy.

Maximize Conversions. Goal: maximum conversions for your budget, no cap on CPA. Works when conversions are equal in value and the budget is fixed. Risk: without a CPA target Google will happily pay $200 per conversion if it thinks that's the best available. And if your tracking lies — Google optimizes toward the lie.

Target CPA (tCPA). Smart bidding classic. Google officially says 15 conversions in 30 days as a minimum. Reality: 15/mo is the absolute floor and the strategy stumbles on it. The working threshold for stability is 30+ conversions in 30 days, steady week over week. Works brilliantly on accounts where you already know what you're willing to pay.

Target ROAS (tROAS). Goal: revenue per dollar spent. Google says 50 conversions with value in 30 days. Realistically — 100+. For e-commerce with real basket variance, mandatory if you have the data. If 90% of your orders are $50–70, stay on tCPA — tROAS won't give you anything extra.

Maximize Conversion Value. "Give me as much revenue as possible, don't cap the ROAS." For aggressive e-com growth when you're willing to overpay for volume. Risky, needs daily oversight.

This isn't a "pick one forever" guide. It's a toolkit. Strategy has to change with the phase of the account — and the most common mistake I see is a client who picked tCPA on day one and is still sitting on it years later, even though the budget and structure have long since become something else.

Decision tree: 4 questions, 1 strategy

Reading seven descriptions is one thing. Picking one is another. So — four questions, built on Google's real thresholds and the common sense Google doesn't document.

Answer honestly — not the way you wish it were, but the way it is. If on the first two questions you already have data or tracking problems, no smart bidding will save you. Fix the foundation first, then talk strategy. That's real money.

Two caveats. First: the thresholds (15/mo, 30/mo, 14 days) are reference points, not laws. In long-sales-cycle niches (real estate, B2B SaaS) different numbers apply — there it makes sense to count micro-conversions. Second: the tree gives you a starting strategy. After 2–3 weeks, look at the data and, if needed, move up the deck.

Real thresholds: when to switch

Google's guides love writing "once enough data accumulates." That's vague bullshit you can hang an account on for years. Here are the concrete thresholds I actually work with.

Manual CPC → eCPC (if it's still launching for you). 40+ conversions per month, steady over 14 days. Earlier — it's just ± 30% on your bid with no signal behind it.

Manual CPC / eCPC → Maximize Conversions. 60+ conversions per month, clean tracking, you don't care about a specific CPA and you do care about volume. Often this is an intermediate stop before tCPA — give it 2–3 weeks, lock in the "natural" CPA Google reaches, and only then switch.

Max Conversions → Target CPA. Google officially says 15+ conversions in 30 days as a minimum. That's the absolute floor — the strategy wobbles there, learning phase drags. Personally I recommend 30+ in 30 days for stability (twice Google's official minimum). And — target CPA has to be within ±20% of the CPA the campaign is actually delivering on Max Conv. Setting "I want it twice as cheap" isn't optimization, it's sabotage: Google will just stop serving.

tCPA → tROAS. At least 30 conversions in 30 days with proper value tracking (100+ preferred). And value has to be genuinely varied — if 90% of your transactions are $50 and 10% are $52, tROAS gives you no edge over tCPA, it's just the same model with an extra layer of complexity.

Learning period after a strategy change. Google officially says 7 days. Don't believe it. Realistically — 14–21 days, and the first week often performs worse than the old setup. Judging results before day 14 is self-deception. Don't switch back after three panicked days. That's the most expensive mistake I see accounts make.

When to go back to Manual CPC. Automation has been running for 2 months, CPA has crept up with no explanation (season, competitor, landing all unchanged) — don't be afraid to pull it, put Manual on, and figure out what's broken. If learning phase fails to finish a third time in a row, volume isn't enough for that strategy. Period. Smart strategies are not a one-way door.

5 symptoms your strategy is wrong

Quick diagnostic. If you see two out of five — time to think about switching.

1. CPA has crept up with no explanation. Was $30, now $60, no fewer conversions but each one costs twice as much. That's tCPA or Max Conv that found an "expensive" segment and won't let go. CPC 3–5× the expected for a specific keyword — same story: automation is "experimenting" on your money, and most often nothing comes of it. Pull it to Manual, figure out what happened, come back with a new target.

2. Impression Share collapsed without a budget change. Was 60%, now 20%, you changed nothing. That's tCPA rolling bids back because it decided conversions were going to get more expensive. Often it's wrong. Google saying "bids too low" is effectively the strategy quietly refusing to work. Either raise target CPA by 15% or go back to Manual.

3. Campaign burns its budget in the first 4 hours of the day. This isn't really the bid strategy itself (ad scheduling and delivery method are campaign-level settings that work with any strategy). But Max Conversions without caps often amplifies the problem: Google pushes traffic in the morning when auctions are cheap, while your audience buys in the evening. Fix — turn on standard delivery, tune ad scheduling to audience behavior, and if that doesn't help, temporarily switch to Manual/eCPC with an hourly bid schedule.

4. Volume drops after the "learning phase." Google said "training finished, it'll get better." It got worse. Standard scenario when data was on the edge. Either go back to Manual, or rethink structure (more conversions in one campaign = more signal).

5. Your best-converting group sits in "Limited by budget" for weeks. That's literally money on the table. Move that group into its own campaign, put Target IS with an aggressive 75–90% Top of page target on it (the second use case from the section above) or Max Conv Value with an expanded budget. Never leave that group choked inside a shared campaign with average keywords.

One symptom — could be noise. Two or more — time to look at the strategy with a clear head. And separately: if Google Ads recommends "raise target CPA" or "increase budget" every week — that's not advice, that's a sales pitch. Verify against your own metrics, not against Optimization score.

3 typical mistakes with smart bidding

Smart bidding can be very powerful, and on most accounts with clean data it genuinely works. But there are three mistakes that give smart bidding a reputation as "Google's bullshit," when in fact the bullshit isn't the strategy — it's how it gets turned on.

Mistake one: tCPA / tROAS on dirty tracking. The biggest and most expensive one. You send Google an event called "conversion" — and 40% of it is duplicates, 20% bots, 15% cross-device that's actually one person. Or you piped GA4 conversions into Google Ads through an imported goal with cross-domain losses. Google optimizes toward that slop, brings you traffic that looks like slop, and ROI tanks. You wonder why tCPA delivers $80 instead of $40 when GA4 shows $35. Smart bidding is only as smart as the data going in. Before turning any Smart on — verify: (a) deduplication, (b) cross-device is on, (c) offline conversions come back from CRM. Half the people skip this step — then act surprised.

Mistake two: tCPA without a big enough base, or aggressive tROAS at launch. Classic: an account with 3 conversions a week, someone sets tCPA — and wonders "why doesn't smart bidding work." It works. There's just nothing for it to work with. Or the tROAS version: you want 500%, you set tROAS 500%, Google looks at the data, sees an average ROAS of 250% — and serves almost nothing. You think the strategy is broken. It's not. You set an unrealistic target. My rule: the first tROAS = your current actual ROAS × 1.0 (no bump). Give it 2 weeks for the learning phase to stabilize. Then, if the strategy holds volume and ROAS, tighten by +10% at intervals of at least 2 weeks between steps. Starting straight at × 1.1 is the most common way to strangle volume in the first week.

Mistake three: changing the target every 3 days. Monday — CPA $50, you set target $40. Wednesday — CPA $55, panic, you drop it to $35. Friday — impressions collapse, you raise to $45. Google doesn't have time to learn any of your targets. Good point on mechanics: learning phase doesn't reset on every tiny change, but on substantive ones (bid strategy change, target change of 15–20%+, significant budget change, big structural rebuilds). But in the scenario above every one of your changes is exactly that: 20%+ up and down. You live in an eternal learning phase and pay for it daily. The rule: pick a target — don't touch it for 14 days. Minimum. If after 21 days it's worse — then roll back. Earlier — only in a catastrophe: CPA up 3× or volume down 3×.

Bonus on the third: don't run one strategy across the entire account. "We're all on tCPA" sounds solid, but it's not a strategy, it's laziness. Brand campaign — Target IS "Absolute top" 90%+. Non-brand traffic on broad match — Max Conv with a hard CPA cap. Top-converting group — its own campaign on Target IS with an aggressive 75–90% target. Retargeting with decent data volume — tROAS. Different campaigns do different jobs. One strategy for all of them is like driving nails, screws, and staples with the same hammer. Sometimes it works. Often it breaks the furniture.