Why Bidding Beats Keyword Selection
When I take on a new PPC account, the single fastest lever I pull is bids. Not keyword selection. Not campaign structure. Not match types. Bidding.
Think of it this way: if you picked the best keywords in the world but the bids are wrong, you'll either get zero impressions (too low) or bankrupt the account (too high). But if you inherit mediocre keywords and set the bids correctly, the formulas will naturally increase spend on everything relevant and decrease spend on everything irrelevant. The account self-optimizes.
Keyword bidding covers a multitude of sins. Your account could be fully auto campaigns — if the bids are right, the performance will be on target. Keyword selection determines which queries you show up on. Bidding determines everything else.— Mohsin Raza, Adaptoid E-Commerce
The Problem with "Increase/Decrease by X%"
The Industry Default
- If ACOS > target, decrease bid by 10%
- If ACOS < target, increase bid by 10%
- At 500% ACOS, takes months to reach target
- Frequency-dependent — running daily compounds errors
- Why 10%? It's arbitrary — a guess
Math-Based Bidding
- Calculates the exact bid needed to hit target ACOS
- Gets you there in one move, not months of iteration
- Frequency-safe — same data = same output every time
- Based on actual keyword performance data
- No arbitrary percentages, pure math
Applying percentage-based bid rules to your current bid instead of your current CPC. The bid is a status — it's whatever it happens to be set to right now. The CPC is actual performance data from your selected time frame. Always calculate from CPC, never from bid.
The Revenue Per Click Formula
This is the foundational formula for everything in this guide. If you learn nothing else, learn this.
ACOS is spend divided by sales. At a keyword level, that's CPC divided by RPC (Revenue Per Click):
If your target ACOS is 30%, you need your CPC to be 30% of your RPC. Cross-multiply:
Your current CPC is $2.00 — you're actually underbidding. You can push the bid up to $3.00 and capture more traffic while staying within your ACOS target. If CPC was $5.00 instead, you'd know immediately it needs to come down.
The RPC formula doesn't increase or decrease by an arbitrary percentage. It calculates the exact target CPC needed to hit your target ACOS — based on how the keyword is actually performing. You can run it as many times as you want; as long as the data hasn't changed, the output is the same.
The Mental Math Shortcut
If current ACOS is 60% and target is 30%, the ratio is 0.5 — cut the CPC in half. Gets the same number as RPC. Faster for scanning the ad console; RPC is more natural in spreadsheets.
If someone reduced all bids to $0.05 yesterday, the 30-day ACOS still shows historical performance at the old CPCs. Reducing further because "ACOS is still high" would collapse the account. The bid is a status. The CPC is data. Calculate from data.
The 4 Keyword Categories
Every keyword in your account falls into one of four categories. Each uses a different formula. The first two reduce spend. The last two increase spend.
Reduce 1. High ACOS
ACOS exceeds target. Apply: New Bid = RPC × Target ACOS. This gets you to your target in one move. Run anytime — frequency-safe.
Reduce 2. High Spend, No Sales
Significant spend, zero conversions. Apply CPA-based anticipation formula. Run anytime — frequency-safe. Details in Part 3.
Increase 3. Low Visibility
Too few clicks/impressions for RPC. Step up bids 10-20% per cycle until data accumulates. Frequency-sensitive — once per cycle only.
Scale 4. Low ACOS
Performing well below target. Step up 5-25% per cycle to capture more volume. Never exceed the bid ceiling. Frequency-sensitive.
Frequency Safety: Which Formulas Are Safe to Repeat?
| Category | Method | Run Anytime? | Why |
|---|---|---|---|
| High ACOS | RPC × Target ACOS | ✅ Yes | Recalculates from data — same input = same output |
| High Spend, No Sales | CPA projection | ✅ Yes | Formula recalculates as clicks accumulate |
| Low Visibility | Step up 10-20% | ⚠ Once/cycle | Compounds on current bid — daily runs double bid in a week |
| Low ACOS | Step up 5-25% | ⚠ Once/cycle | Same compounding risk — bid ceiling prevents runaway |
Run either the decrease formulas (1 & 2) or the increase formulas (3 & 4), depending on account pacing. ACOS running hot? Focus on decreases. ACOS healthy and want growth? Run the increases. This push-and-pull gives more control than running all four simultaneously.
Optimizing Non-Converting Keywords
Keywords with significant spend and zero sales are one of the biggest account-level ACOS killers. Most people handle them wrong.
What Most Sellers Do
- Archive everything with $20+ spend and no sales (arbitrary)
- Negate the keyword entirely (lose future sales potential)
- Decrease bid by X% weekly (slow, imprecise)
- Use one threshold for all products ($200 product ≠ $15 product)
What You Should Do
- Calculate target CPA per product mathematically
- Use CPA-based formula to anticipate next conversion
- Bids step down automatically as non-converting clicks accumulate
- Threshold adapts per product — no arbitrary dollar amounts
Step 1: Define "High Spend" Mathematically
A $200 product at 25% target might have a $50 CPA threshold. A $15 product at 30% has only $4.50. Using math instead of an arbitrary dollar amount means your system adapts per product automatically.
Step 2: Calculate the Anticipatory Bid
If the keyword gets 5 more clicks without converting (now at 20 clicks):
The bid naturally steps down from $0.36 to $0.30 as non-converting clicks accumulate. No manual intervention required. If it converts on click 25, total spend will be close to the $9 target CPA.
Individually, no single low-data keyword is a big offender. But cumulatively, keywords with 1-5 clicks and no sales can constitute 10-20% of total account spend. This formula gives you a systematic way to manage all of them without blindly archiving potentially valuable terms.
Want Your Bids Optimized by an Expert?
I'll audit every bid in your account using the RPC methodology and deliver optimized bids across all four keyword categories.
When to Increase Bids
Knowing when to increase is harder than knowing when to reduce. The RPC formula can't help here — on low ACOS keywords with few clicks, the RPC value is unreliable.
Why Not Just Use RPC to Increase?
A keyword with 1 click and 1 sale at $0.20 CPC on a $20 product has an RPC of $20. Multiply by 30% target = $6.00 bid. That assumes a 100% conversion rate holds at 30x the current CPC. That's dangerous.
The Step-Up Method by Account Condition
| Account Condition | Increase % | Rationale |
|---|---|---|
| On target, gentle growth | 5-10% | Conservative push for incremental volume |
| Under pacing budget | 10-20% | More aggressive to close the spend gap |
| Severely under pacing / launch | 15-25% | Need velocity now; monitor closely for waste |
The Grace Range: Don't Touch What's Working
Use a 10% grace range around target ACOS. If target is 30%, anything between 27% and 33% is "on target" — don't touch it. Only optimize outside that range. Prevents unnecessary disruption on keywords performing within range.
When to Increase Bids on High ACOS Keywords
Sounds counterintuitive, but there are legitimate reasons:
The bid was previously set wrong (manual over-reduction)
Someone manually nuked bids. The current bid is too low relative to what the data says the keyword can afford. The RPC formula corrects upward. The "new" higher bid is still below the historical CPC that caused the high ACOS — it's not overspending, it's un-underspending.
Performance improved since last optimization
Conversion rates went up (new reviews, better images, price change), so the keyword can now support a higher CPC. The new calculated bid is above the current bid but still below the CPC that caused the old high ACOS. The data changed — the formula responds.
Placement adjustments shifted
If you reduced a non-top-of-search placement modifier, the base bid may need to increase to maintain total effective CPC across placements. Placement modifiers and base bids are a linked system — changing one often requires adjusting the other.
Strategic ranking decision
You're willing to accept a higher ACOS on specific keywords for market positioning, share-of-voice, or organic ranking benefits. This is a deliberate investment, not a mistake — but it should be tracked separately from standard profitability metrics.
Bid Ceilings & the Data Hierarchy
A bid ceiling is the maximum CPC you can afford while staying at or below target ACOS. It's derived directly from RPC:
When stepping up bids on low ACOS keywords, this ceiling is your hard stop. Never step above it.
The Data Hierarchy for Low-Data Keywords
What about keywords with 1 click and no sales? Their individual RPC is meaningless. Use the hierarchy:
If the ad group averages a $0.75 CPC at 30% ACOS, that's roughly the ceiling for any keyword within it. A keyword at $2.00 CPC with 1 click, no sales, is already over that benchmark. A keyword at $0.20 CPC has room to grow.
The "1X" bid ceiling means bids never exceed the max affordable CPC. Settings like "2X" or "3X" allow bids up to 2-3× that ceiling — useful for ranking campaigns, but dangerous for profitability. Stick to 1X unless you have a specific strategic reason to go higher.
Placement Adjustments & Bid Tandem
Here's a mistake I see constantly: people optimize keyword bids without touching placement settings, or vice versa. These must be managed simultaneously.
It's one bid with multipliers that produce one CPC. You can't optimize one without the other.
The Goal: Balanced ACOS Across Placements
If your target ACOS is 30%, you should hit approximately 30% on Top of Search, 30% on Rest of Search, and 30% on Product Pages. The spend allocation scales with conversion rates — more spend on higher-converting placements.
Why "Lower Bids → Higher ACOS" Happens
A common complaint: "I lowered my bids using RPC and ACOS went up." Here's why:
You lost Top of Search placements
Top of Search has the best conversion rates. When you lowered bids without increasing the ToS multiplier, you lost those placements. Now all spend goes to Product Pages where ACOS is worse.
You lost your best search terms
On broad/phrase/auto campaigns, different search terms have different CPCs. The expensive ones were often the best converters. Lowering the bid kept only cheap, low-converting traffic.
The fix: adjust placements in tandem
Always adjust placement settings when changing bids. And harvest your best search terms into exact match campaigns so you can bid on them independently.
If Top of Search ACOS is 15% and Product Pages is 60%, you're underspending where you convert best and overspending where you don't. The fix: reduce all keyword bids to target Product Pages CPC, then increase the ToS modifier to compensate.
How Often to Optimize
Once or twice a week. That's the standard cadence. Not daily. Not hourly. Weekly gives enough time to collect meaningful data and observe the impact of changes.
Why Not Daily?
Not Enough Data Change
On a 30-day lookback, one additional day barely moves the averages. You'll make the same decisions you made yesterday.
Step-Ups Compound
Increasing bids 10% daily means doubling the bid within a week. The data didn't warrant it.
Can't Troubleshoot
If ACOS spikes after 2 weeks of daily changes across thousands of keywords, you can't trace which change caused it.
You're not just paid to optimize — you're paid to know when not to optimize. If sales are growing, ACOS is on target, and trends are positive, sitting on your hands and extending budgets might be the right call. There's an art to doing nothing.
Five Methods for Setting Starting Bids
| Method | How | Best For | Risk Level |
|---|---|---|---|
| Amazon Suggested | Use Amazon's range as reference only | Quick sanity check | Medium — not based on YOUR margins |
| Start Low, Inch Up | Begin at $0.50-$1.00, increase 10-20%/cycle | Conservative launches | Low — slow but guarantees good ACOS |
| Account Avg CPC | Use current account average, adjust by product price | Adding products to healthy accounts | Low — grounded in real performance |
| Converting ST CPC | Use the search term's converting CPC | Harvesting keywords | Medium — watch for anomalous low CPCs |
| Ad Group Target CPC | Match the destination ad group's calculated target | All situations | Lowest — best method |
Don't blindly use Amazon's suggested bid. Amazon's suggestion is based on auction market price — it has no idea about your margins, target ACOS, or product economics. A $0.75 suggested bid might be $2.00 more than you can profitably afford, or $3.00 less than you should be spending to win placements.
Macro vs. Micro Bid Changes
This framework separates routine maintenance from reactive optimization. Most people only do one type. You need both.
Routine Maintenance
- Scope: entire account, 80%+ of spend
- Date range: 30-90 days
- Max adjustments: ±10-15% bids, ±20-33% placements
- Frequency: every 1-2 weeks
- Risk: lower (small changes across many keywords)
Targeted Interventions
- Scope: specific campaigns, <10% of spend
- Date range: 7-14 days
- Max adjustments: ±25-50% bids, unlimited placements
- Frequency: as needed, mid-week
- Risk: higher per keyword, low total impact
The Optimal Rhythm
Weekly (Macro)
Run full-account optimization with 30-day lookback and conservative limits. This keeps everything tuned without volatility.
Mid-Week (Micro, as needed)
Spot-check specific campaigns or keywords that are off. Use 7-14 day data. Apply more aggressive adjustments to small pockets of spend.
Your date range is a tradeoff between data confidence (longer = more reliable) and data relevance (shorter = reflects current conditions). Your window should begin after the most recent significant change — deal, bid optimization, stockout, or seasonal shift. You're not optimizing based on what happened. You're optimizing for what's about to happen.
It's Math. Not Magic.
Every formula in this guide boils down to one idea: your cost per click should be a calculated percentage of your revenue per click. That's it. Everything else — the four categories, the non-converting formula, the placement adjustments, the macro/micro framework — is applying that idea to different scenarios.
| Scenario | Method | Frequency-Safe? |
|---|---|---|
| High ACOS keywords | RPC × Target ACOS | ✅ Run anytime |
| High spend, no sales | CPA-based RPC projection | ✅ Run anytime |
| Low ACOS keywords | Step up 5-25% | ⚠ Once per cycle |
| Low visibility keywords | Step up 10-20% | ⚠ Once per cycle |
| Placements | Balance ACOS across all | ✅ With bids, 1-2x/month |
| Full account (macro) | 30+ days, conservative limits | ✅ Weekly |
| Specific issues (micro) | 7-14 days, wider limits | ✅ As needed |
The formulas handle 80% of the work. The remaining 20% — date range selection, optimization frequency, when to hold, when to push — is the art. And that art comes from experience, judgment, and knowing your account. Trust the math, but stay human.
- Bidding beats keyword selection. Get bids right and an account with mediocre keywords will outperform an account with perfect keywords and wrong bids.
- RPC is the only formula you need for high ACOS keywords. It calculates the exact target CPC in one move. No iteration, no arbitrary percentages.
- Non-converting keywords need math, not arbitrary thresholds. The CPA projection formula steps bids down automatically as data accumulates.
- Always manage bids and placements in tandem. Changing one without the other breaks the system. Effective CPC = Base Bid × Placement Multiplier.
- Prioritize by dollar impact, not severity percentage. A 100% ACOS on $50 spend matters less than 40% ACOS on $5,000 spend.
Bid optimization on Amazon isn't a mystery. It's math. And now you have the complete system. Use it wisely.— Mohsin Raza, Adaptoid E-Commerce