This chapter is about operating an Amazon ad account that is already live: how to diagnose a campaign in funnel order rather than reflexively tweaking bids, how to judge ACoS against margin and TACOS against the product's lifecycle stage, and what the recurring review pass actually consists of — spend-ordered audits, search-term reports, red/yellow/green keyword triage, match-type trimming, and the budget-versus-bid decision. It then layers on the secondary levers: day parting from hourly exports, Search Query Performance-driven spend allocation, Sponsored Display and Sponsored Brands video as under-contested inventory, branded-term defense versus category domination, and DSP/AMC at the top of the funnel. The material is dense and numeric on diagnosis, triage, and day parting; it is noticeably thinner — framing rather than operator detail — on DSP and AMC.
The reflex when a campaign underperforms is to open Campaign Manager and move bids. The framework that governs this whole chapter says that is almost always the wrong first move. Leading Actions vs. Lagging Metrics (PPC Diagnosis Framework) separates what a seller directly controls — a main-image swap, a bid change, a copy edit, a campaign restructure — from the delayed outcomes those actions produce: keyword rank, conversion rate, click-through rate, indexing status. Effective diagnosis is knowing which leading action moves which lagging metric, and that mapping is learned from observed algorithm behavior rather than assumed.
The practical form of this is a four-metric funnel, checked strictly in order: impressions, then CTR, then conversion rate, then ACoS. Each has its own distinct set of likely causes, so a bad number in one does not imply the same fix as a bad number in another:
ACoS sits last for a structural reason: it is the funnel's output, not a root cause. As the source puts it, “ACOS should always be analyzed last, not first. It's not the root problem. It's the result of everything happening earlier in the funnel.” Crescent Kao ties the framework to two concrete failure modes — sellers who skip straight to ACoS and therefore never locate where the funnel actually breaks, and poor campaign structure that mixes match types and keyword lists in one campaign, which muddies the very data needed to run the diagnosis. Campaign structure itself belongs to PPC Campaign Structure & Bidding; this chapter assumes it is already clean enough to read.
The arithmetic that turns raw report totals into these metrics is deliberately trivial, and worth keeping in one place because it gets reused constantly — in hourly pivot tables, placement comparisons, and per-keyword triage alike. PPC Efficiency Metric Formulas (ACOS, CTR, CVR, CPC): ACoS = spend ÷ sales, CTR = clicks ÷ impressions, CVR = orders ÷ clicks, CPC = spend ÷ clicks. Format the first three as percentages and CPC as currency to two decimals. ROAS (Return on Ad Spend) is the same numbers inverted — revenue per dollar of spend instead of spend as a share of revenue — and for sellers who think that way, every rule below mirrors cleanly: compare ROAS against a break-even ROAS threshold instead of comparing ACoS against margin.
That comparison is the point. ACoS should never be judged against a universal “good” number: 40% is fine at a 60% margin and ruinous at a 20% margin. The trigger condition in ACOS-Above-Margin Optimization Sequence is specifically ACoS above the product's profit margin — “if the ACOS is greater than your profit margin, you're actually losing money each time you make a sale.” When that trigger fires, the fix runs in two ordered steps, not simultaneously. First adjust placement — top-of-search versus rest-of-search versus product pages — because placement misallocation can mask or amplify otherwise-fine keyword performance. Only then download the keyword report and work keyword by keyword. Doing keyword surgery first risks over-correcting keywords that were only underperforming because of where the campaign was buying impressions.
ACoS measures ad spend against ad-attributed revenue only. TACOS vs. ACOS introduces the metric that measures ad spend against total revenue — ad plus organic. The difference matters because a campaign can look bad on ACoS while being profitable for the business, if what it is really buying is organic rank. The recommended split: TACOS as the north-star profitability metric for a product or account, ACoS as a per-campaign efficiency metric inside it.
The strongest version of this argument in the material is agency-side. A seller stuck at a growth plateau had a team optimizing each campaign's ad-cost ratio without ever connecting spend to organic-rank movement; once PPC was managed toward organic ranking instead of campaign-level efficiency, ad spend roughly doubled while revenue quadrupled — the extra spend was buying rank, not just clicks. In the Sophie Society “Study Key” case, ACoS and TACOS were used together as a plateau diagnostic: a campaign could look efficient on ACoS alone while total business growth stalled, and watching TACOS trend down over time was the signal that spend was converting into durable rank. In a $1M-in-12-months supplement launch, $164k of monthly revenue ran at 16.6% TACOS despite specific ranking campaigns running as high as 65% ACoS. In the same launch, a broad-match version of the identical core keywords posted the best ACoS of all ten campaigns run — and budget kept going to the worse-ACoS exact-match version anyway, because campaign selection was driven by ranking impact rather than by which campaign looked most efficient.
What counts as an acceptable TACOS is not a constant, which is why Amazon Product Life Cycle (Launch, Expansion, Harvest) exists: Launch, Expansion, Harvest, each with its own expected PPC-share-of-sales, TACOS/ACoS benchmarks, and budget mix. A product in Launch is expected to run a far higher ad-spend ratio than one in Harvest, where organic sales should carry the volume. Diagnosing which stage a product is actually in — instead of applying one universal ACoS target across a whole catalog — is the framework's main use. A falling TACOS trend is the transition signal: in one 90-day launch, weekly TACOS fell from 21% to 6.5% while weekly revenue held around $12,000, read as the product becoming a self-sustaining “cash cow” — “we continue to feed it, take more market share, and continue to pull cash out of it just like a real estate asset.”
Rather than assume that transition has happened, Harvest-Phase Ranking Stress Test makes you prove it by withdrawing support in stages and watching whether rank holds. One sequence used in practice: cut ad spend by roughly 10%, end running price discounts, and raise price outright — in one case from $32.99 to $39.99 — then watch sales and rank over the following weeks. If rank survives the price increase and reduced discounting, that is direct evidence the ranking is organic-demand-driven rather than ad- or discount-subsidized. In the case observed it didn't just hold: spend was later roughly doubled again and revenue quadrupled off the validated organic base, taking the brand from single-digit daily units to 30–50 units/day. Sophie Society's version of this for Study Key was explicitly incremental — pull back a step, watch, repeat if organic holds, ease off if it sags — a repeatable dial rather than a one-time switch, applied to overall growth pace as well as to ad spend.
Against all this weekly and monthly measurement sits one long-horizon corrective. Long-Horizon Product Evaluation Window (6 Months–2 Years) is a veteran seller's rule that a new product deserves 6 months to 2 years before being called a failure, not the 2 weeks to 1 month most sellers use. Early PPC and organic performance is noisy and shaped by launch mechanics, not by the product's real long-term demand or review trajectory. Set the kill-decision clock at 6 months minimum, and resist reading a first-month launch phase as a verdict.
Optimization here is a ritual with a clock on it. 7-10 Day PPC Data-Collection Cooldown sets the cooldown: let at least 7–10 days elapse on a new or newly-adjusted campaign before changing anything. Daily sales and click counts fluctuate enough that a shorter window reliably mistakes noise for a trend. The cooldown defines when the review below runs, not just how.
Before diagnosing any single campaign, Order-by-Spend Campaign Audit says to sort every campaign by ad spend — not by ACoS, not by sales. That surfaces where budget is actually concentrating before you evaluate whether it converts, and it catches runaway broad/auto keywords and mis-sized campaigns first: “if you order by spend, you're going to find that usually most of your spend is going to be concentrated on your top five to 10 keywords.” Subtler issues are worth investigating only after the big pipes are checked.
When something looks like it dropped, Multi-Window PPC Performance Review guards against reacting to a blip: review last week, last month, and last three months side by side. Consistent underperformance across all three is a genuine problem; a dip visible only in the shortest window may be normal fluctuation. And a real dip still shouldn't be diagnosed inside the ad account alone — cross-check it against the PPC change history and then the listing's own edit history. Title edits, image changes, price changes, and stock-outs all produce PPC drops that look like ad problems and aren't. A dip that lines up with a listing edit is, in the material's phrasing, a listing problem wearing PPC clothing.
The raw fuel for the pass is the Amazon Search Term Report, which shows the actual customer search terms and ASINs that triggered your ads with their clicks, spend, sales, and ACoS. It has two uses — pruning and promotion: negate irrelevant terms and ASINs, and promote well-performing discovered terms into their own dedicated exact-match campaigns. It becomes available within 24 hours of a campaign going live, and obviously irrelevant terms should be negated as soon as they're spotted. But a high ACoS in the first few days — even 50–100% — is explicitly not a negation trigger on its own; the sample is too small. That restraint is generalized by "Buy the Data" Keyword Decision Rule, credited by one practitioner to “Ritu”: keyword relevance can't be predicted from theory, so commit a set dollar amount to a keyword before making a keep-or-kill call, scaled to product price so the spend produces at least one real sale-or-no-sale data point. Keyword testing is a paid research cost you're expected to incur, not a targeting mistake to avoid.
Once the data is in, Excel-Based PPC Report Color-Coding Workflow is the triage. Download the per-campaign reports, sort each by ACoS and then by spend, and color-code every keyword. Red — ACoS far above the margin break-even, or zero sales with cumulative spend past roughly 3× the per-unit dollar profit margin (e.g. $30 of spend on a product with a $10 margin) — turn it off. Below that threshold, a no-sale keyword is not yet red; there simply isn't enough data. Yellow — borderline ACoS relative to margin — lower the bid gradually rather than cutting. Green — profitable — migrate it into a separate, more aggressively funded campaign so it stops competing for budget with weaker keywords. The rationale for doing this in a spreadsheet rather than the console: “doing it this way in Excel is a much better strategy than just making changes directly in Amazon's campaign manager,” because bulk-editing and color-coding before pushing changes back avoids error-prone one-off edits.
Two structural cleanups belong to the same pass. Match-Type Overlap Trimming & Manual Campaign Restructuring handles keywords whose campaigns have drifted out of control — often because automation silently duplicated or restructured them. The method is deliberately manual: disconnect automation entirely, work root word by root word and then keyword by keyword, identify every campaign touching that keyword across exact/phrase/broad, find which single campaign is actually driving the winning performance, turn off the redundant ones, and then tune the survivor's placement settings rather than reallocating spend into the also-rans. In one case a single keyword was spread across 5 duplicate exact campaigns from an automation bug; in another, 3 broad campaigns ran alongside the 1 exact/phrase campaign that was the proven winner. A related simplification some sellers apply: drop Phrase match entirely, on the reasoning that Broad already covers what Phrase would catch, so running both duplicates spend across two match types chasing the same terms.
The trap underneath all of this is Automated Bidding Software Can Silently Revert Manual Changes: third-party bid tools that aren't fully disconnected can silently revert manual changes, because the automation's last-known state wins with no visible warning. Confirm the bidder is off for a campaign before doing manual work on it — otherwise a correct fix gets quietly undone and the diagnosis looks wrong when it was never allowed to persist.
Not every underperforming keyword is a bid problem, and not every keyword is winnable at all. Per-Keyword Rank Ceiling & SQP Competitor Conversion Benchmarking makes the uncomfortable case explicit: a product can have a natural rank ceiling on a specific term — stuck at #17, say — because the products above it simply convert better on that term. Pushing more ad spend past that ceiling doesn't buy rank, it burns budget. The diagnostic is a comparison, not a judgment call: check your own ad conversion rate on the keyword against Data Dive Search Query Performance's average competitor conversion rate for that same term. Meaningfully below it with no realistic listing or image fix available means you've hit your ceiling — stop pushing. At or above it means rank is still winnable and spend can rise. This is done root by root and keyword by keyword; the same listing can carry one ceiling-capped term to abandon and one strong root to double down on at the same time.
Two reporting caveats travel with that method. Headline CTR and conversion figures are often 30-day blended averages, which can hide a much stronger recent shift — CTR over the last 2 days can run more than double the blended number right after a main-image change, so check the last few days of raw data separately. And Amazon's Search Query Performance report is inconsistently populated week to week, including gaps in recent weeks; some optimization calls have to be made on incomplete data.
SQP-Based Traffic Allocation Decision Rule turns funnel-share data into an explicit spend/no-spend gate per search term. If your brand's share holds or rises all the way down the funnel — impression to click to purchase — the term has no unresolved leak, and it deserves more traffic via a dedicated exact-match or Sponsored Brands campaign with its own creative. If share drops sharply at any stage, hold off on driving traffic until the thumbnail, title, or listing content behind that drop is fixed; otherwise the extra spend just funds the leak. The Essential Candy example shows how misleading rank alone can be: generic high-traffic terms like “peppermint” and “hard candy” captured only about 0.05% of the brand's total available search volume for those terms, even though ranking for them felt like SEO success.
For deciding whether a ranking push is worth funding at all, Revenue Potential Calculation Method (attributed to John Durket) sizes the prize: monthly sales potential = search volume × conversion rate × conversion share × average selling price. Inputs come from Amazon's own reporting — keyword search volume and the #1 ASIN's click/conversion share from search-term and Search Query Performance data, plus category conversion and opportunity sizing from Product Opportunity Explorer. The worked example cited: a gift-wrap-paper keyword nets roughly $8,500/month at the #1 spot. Pair that with a SmartScout analysis of about 2M January-2026 search terms, which found the #1 organic slot taking roughly 25% of clicks and 19% of sales versus about 8% and 5% at #3 — the payoff is concentrated enough at the top that clicks won, not rank position for its own sake, is the lever to test.
Organic-Rank-to-Ad-Gap Keyword Targeting (Cerebro Cross-Reference) finds an easier class of target: cross-reference where a listing already ranks organically against where it has sponsored presence, and build ads specifically for keywords sitting in that gap. Shoppers are already finding the product on those terms without an ad having sold them on it, so the traffic is pre-primed — this is defending and amplifying strong organic terms rather than discovering new ones from a blank slate.
The widest version of this thinking is PPC Data Loop (Iterative Campaign Testing → Listing Optimization): run a spread of campaign types simultaneously and feed the resulting data into two separate decisions — which campaigns to scale or kill, and which listing assets to revise. A keyword converting poorly across multiple campaign types is as much a listing problem as a targeting problem. Chris Rawlings (Sophie Society) frames the loop as two co-equal halves: the campaign actions everyone runs (bids, budgets, targeting, negation, graduation) are “only 50% of the actual full picture,” with the other half being primary-image split tests, secondary images, A+ and Brand Story, title, and bullets driven by the same segment-level data — “the secret sauce that no one does.” A related rule of thumb: once search-term data shows which term families convert, rewrite the thumbnail, title, and copy to reinforce those winners rather than spreading emphasis evenly across the whole keyword list — treat the search-term report as the market's vote on which framing to commit to. His thumbnail framing: “the thumbnail is the package that holds your listing, it's the wrapper to the listing,” with PPC as the engine that carries that wrapper into view on “a pay-to-play marketplace” where more than half of first-page results across the first four scrolls are ads. The creative half of this belongs to Listing Content & Conversion Design and the keyword half to SEO & Keyword Strategy: Winning A9, Cosmo & Rufus; what's specific here is that ad data is the signal that tells you which asset to change. The claimed results are self-reported and single-source: a niche supplements brand taken from ~60 units/day to 500–600+ units/day over a year, peaking above 1,000 in a day, plus an attribution of the $600M Zesty Paws exit to the same process that rests entirely on Rawlings's own account.
A recurring waste of effort is pulling the wrong one of these two levers. Time-in-Budget Diagnostic (Budget vs. Bid Adjustment Decision) resolves it with a number Amazon already reports in the Budgets tab of Campaign Manager: “average time in budget,” i.e. how much of the day a campaign spent before running out. Read it against ACoS:
Reading utilization without ACoS (or the reverse) is exactly how sellers end up raising budget on a campaign that was bid-constrained all along.
That diagnostic motivates the stronger claim in Bid Management as the Real Spend-Control Lever, Not Budget Caps: once budgets are set sanely, they stop being what actually controls spend — bid management does. A well-run account shouldn't be routinely hitting its daily cap at all, because correctly set bids make spend self-limit before the ceiling matters. Some sellers apply this literally, setting budgets far above anything that could plausibly be spent so the cap is never binding, and controlling spend entirely through bids and bid adjusters. This doesn't make the budgeting math pointless — going through it is still how you learn what should be spent per day, whether or not a hard cap does any of the enforcing.
For a product with sales history, that number comes from TACoS-Based Ad Budget Formula (Existing Products): monthly ad budget = target TACoS × target total revenue (ad plus organic), then daily budget = monthly ÷ days in month. Anchoring on TACoS and total revenue rather than ACoS and ad revenue makes it a top-down, whole-business target instead of a sum of per-campaign guesses. The launch-time alternative, used when there's no history to compute a TACoS from, belongs to PPC Campaign Structure & Bidding.
Break-Even TACoS Budget Ceiling is the sanity check on top: max daily budget = break-even TACoS × average daily sales revenue over the trailing 30 days, where break-even TACoS is the product's margin after COGS, shipping, returns, FBA fee, and seller fee. Because break-even TACoS is mathematically equal to margin, this ceiling is literally “spend 100% of your margin on ads” — framed as a deliberate zero-profit growth play for aggressive volume and rank, not a calculation error. Use it to confirm that a target budget doesn't exceed what margin allows, not as a routine operating level.
Two tools automate the bid side of this. Helium 10 Keyword Tracker Bid Rule keys bids off rank rather than ACoS: in the configuration cited, raise a tracked keyword's bid 10%/day after 7 consecutive checks outside the sponsored top 3, and lower it 10%/day once organic rank reaches the top 5 — the same raise-while-unranked / lower-once-ranked logic, running without a daily manual check. Helium 10 Ads Automation is broader, automating bid management and keyword/ASIN harvesting including a recommendation tab that flags converting search terms not yet added as targets. Its pitch in the Essential Candy case study is scale: managing roughly 700 recommended keywords and about 100 competitor ASINs in around five minutes a week instead of by hand. That claim sits awkwardly beside the manual, automation-off restructuring described earlier — the material never reconciles the two.
One calendar-driven exception overrides all of the above. Tent-pole Event On/Off Ad Strategy treats major deal windows — Black Friday, Cyber Monday and similar — as a binary, per-client decision. CPCs spike and non-deal listings convert poorly, so if a seller isn't running an actual deal, ads are turned off entirely on those days rather than merely reduced: with no deal there's no realistic chance of converting efficiently against inflated, deal-driven competition. Sellers who are participating go the other way, advertising aggressively and budgeting for elevated CPCs. For seasonal and gift products, gift-related keywords get notably more expensive in Q4, and Brand Analytics top search terms are used to prioritize the most reasonable ones rather than bidding on everything — “don't boil the ocean.”
Day Parting (Amazon PPC) is rule-based bid or budget adjustment by hour of day and day of week — more spend in historically strong hours, less in weak ones. The material is unusually blunt that low-effort blog and Reddit content oversells it: it should not be the #1 priority in PPC optimization. Get bid strategy, campaign structure, and negative keywords right first, then layer day parting on top. There are two legitimate ways to execute it — free and manual, or paid and automated — plus one crude fallback that is explicitly discouraged: manually switching campaigns on and off at set times, which limits Amazon's ability to collect and use campaign data during the off hours.
The data comes from Amazon Hourly Sponsored Ads Report Export, under Measurement and Reporting > Sponsored Ads Reports, which can break campaign performance out by hour rather than by day. The binding constraint: each report is capped at 14 days. To assemble roughly a month of hourly history, run two reports (report type: Campaign, time unit: Hourly) covering 14 days each and combine the exports in a spreadsheet.
Manual Hourly Day-Parting Analysis (Pivot Table Method) — credited within Sophie Society to “Zoki” — is the free analysis. Load the export into Google Sheets, build a pivot table filtered by portfolio with hour-of-day as rows and impressions, clicks, orders, sales, and spend as values, then add the ACoS/CTR/CVR/CPC formulas from PPC Efficiency Metric Formulas (ACOS, CTR, CVR, CPC) alongside it. Then apply the same color logic used in Excel-Based PPC Report Color-Coding Workflow, one conditional-formatting color-scale rule per column because the “good” direction differs by metric: for ACoS and CPC, green at the minimum and red at the maximum; for CTR and CVR, reversed. Collapse the pivot rows down to the 24 hour buckets so every hour is on screen at once, and the table becomes a visual pattern rather than a grid of numbers — the output being something like a block of hours (4:00 a.m. to noon, in one case) that performs well from every perspective, which converts directly into bid or budget rules.
One confound to rule out before believing any pattern: an apparently strong morning may not reflect shopper behavior at all. It can be an artifact of campaigns still having daily budget left earlier in the day and getting throttled later. Check budget exhaustion first.
Hour-of-Day vs. Whole-Day Bid Adjustment Granularity insists the adjustments be made at the hour level, not for a whole day or day-of-week. Sophie Society's Oliver holds this as a rule of thumb: changing a bid or budget for an entire day alters placement-eligibility and serving behavior across the full day, and if that lowers CTR it can suppress organic rank, since Amazon's ranking weighs click-through behavior. Hour-level scoping avoids reshaping the whole day's placement mix. The bodies flag this as a claimed mechanism, not a demonstrated causal test.
Day-Parting Re-Optimization Cadence & Change-Isolation Rule governs maintenance with two rules. Cadence: re-run the hourly analysis and adjust the schedule only every 3–4 weeks — more often and there isn't enough hourly data to separate signal from noise. Isolation: don't change regular campaign bids in the same week the day-parting schedule changes, and vice versa, or a resulting swing can't be attributed to the right lever.
On tooling, Amazon Budget Rules (Native Day-Parting Tool) is Amazon's native option and is dismissed sharply — it can only increase budget within a time window, with no way to decrease it and no bid control at all, described as “an embarrassing excuse for a day parting tool.” Treat it as a blunt supplement for boosting spend in hours already known to convert, never a substitute for real hourly analysis. AdLabs (Automated Day-Parting Software) is the paid alternative, under Optimize > Day parting: copy its calculator sheet, paste in the same 14 days of hourly Amazon export, and review three tabs — days of the week, hour of day, and bid change rules — before applying anything. Its raw suggestions can be extreme (e.g. +270%) and are treated as directional signals rather than instructions. Oliver's moderation layer is to color-code each hour by severity after cross-checking the suggested percentage against that hour's ACoS and click/order volume, then type in a tempered adjustment: typically +15% to +60% for genuinely strong hours, with larger cuts reserved for hours that are both clearly bad and low-volume, so a terrible ACoS from a handful of clicks doesn't drive a big change.
The track record is reported honestly and it is mixed. One account used day parting as a hail-mary on a struggling product and tripled sales. Another, run through AdLabs, nearly halved ACoS while sales grew exponentially. A third saw sales decrease after starting; the response was an aggressive 80% bid cut in non-converting hours, with full abandonment planned if that didn't work. Day parting “is not a silver bullet” — it doesn't work for every account, and going too extreme can itself hurt performance. When it does work, though, the claim is that it can be the difference between a profitable and an unprofitable brand.
Audience-Targeting Whitespace ("Loophole") PPC Strategy states the strategic premise for everything in this section: most sellers pour PPC budget almost exclusively into keyword-targeted Sponsored Products, which leaves interest- and lifestyle-audience targeting largely uncontested. The tactic is to research the brand's actual buyer demographics and lifestyle affinities — golfers and other athletes for a topical pain-relief roll-on, for instance — and build audience campaigns aimed there instead of only defending keyword territory. Helium 10's Scale Stories mentors prescribed exactly this for In Motion Hemp, a roughly $140K/year hemp pain-relief cream business: add Sponsored Display retargeting plus interest/audience campaigns and launch Sponsored Brands video against the 3–5 historically best-converting keywords and the pain points surfaced in reviews. The logic is cheaper, more tailored traffic from where competitors aren't playing rather than out-bidding them where they are.
Sponsored Display Targeting Taxonomy (Contextual, Remarketing, Audiences) is the map. Amazon lists four targeting categories in the UI; Rawlings collapses them into three: contextual (category or individual product targeting), remarketing (behavior-based), and audiences (in-market and interest/lifestyle). The defining constraint — “sponsor display is the only ad type out of the three major ad types where you can't Target keywords it's only about targeting categories audiences and products.” Two structural rules go with it: start new SD efforts with contextual targeting and remarketing, layering audiences only once something is already performing; and keep each targeting type in its own campaign rather than mixing, so performance stays attributable to a single targeting logic. Within contextual, category targeting is flagged as a standout heading into 2025 across all three ad types, and a specific product-targeting sub-tactic is to aim at competitor listings with a lower star rating than yours — conquesting shoppers already primed to switch, which Rawlings cites at a 10% ACoS in his own account.
Remarketing has three variants — viewers of your listing, purchasers of it, and viewers of similar or competitor listings — and a counterintuitive finding: remarketing to viewers of similar products sometimes outperforms remarketing to viewers of the advertised product itself. Rawlings calls the result “weird” but tests both as separate targets rather than assuming the intuitive audience wins. Sponsored Display Remarketing Look-Back Window Testing treats the look-back window the same empirical way: rather than matching the window to an assumed usage cycle, run 30/60/90/180 days as separate targets in the same ad group and compare in the targeting tab. The best window is frequently not the intuitive one — a product with a ~60-day repurchase cycle can perform best on 90 or 180 days. A better starting hypothesis than a guess is the brand's average repeat-purchase interval from loyalty analytics (see Profitability, LTV & Customer Analytics), and windows can be tuned per loyalty segment: shorter for At Risk customers to catch them before they lapse, longer for Hibernating ones who already have.
Sponsored Display Audience Types (In-Market vs. Interest & Lifestyle) splits the third branch: in-market audiences are dynamic segments built from actual purchase and shopping activity, updating continuously; interest & lifestyle audiences are static interest-based segments (“grain-free dog food Shoppers,” “luxury Shoppers”) found by browsing categories such as Shopping Behavior. Many of Amazon's suggested segments are irrelevant noise — described as “wacky” — so the tactic is to browse and cherry-pick rather than accept defaults. That connects to a setup trap worth its own concept: Sponsored Display Prefilled Targeting Suggestions (Clear & Manually Reselect). Amazon prefills category, product, and audience suggestions by default, and Rawlings recommends X-ing them out and manually reselecting — “if you don't do that and you add a few you're actually advertising to different audiences that you didn't even mean to or didn't even select.”
The remaining setup decisions are short. Sponsored Display Ad Format & Placement Mechanics: the default is an Amazon-generated thumbnail built from the listing's primary image, star rating, coupon, and price, previewed across up to 12 placement formats; custom image or video is available but most of Rawlings's own high-performing SD ads run on the default, so upload custom creative only when you genuinely have something that highlights a differentiating feature. Unlike Sponsored Products, SD surfaces in several on-listing spots — a banner near the top of a competing listing, a placement under the bullets, and one above the reviews. Sponsored Display Landing Destination (Product Page vs. Storefront): send clicks to the product page by default; choose the storefront only if that page is already proven to convert, and advertise one product per campaign. Sponsored Display Bidding: CPC vs. vCPM: CPC (“optimized bids for conversions”) is the default and the recommendation — “brand awareness... doesn't pay my bills” — with vCPM reserved for two cases, pushing into competitive placements CPC struggles to win, and brand defense by dominating the ad space on your own listings. Rawlings argues SD behaves as a direct-response channel, not a branding one: “it's not a brand awareness type of platform it's a direct solution type of platform people are looking for Solutions right now,” with a real proof point of over $100,000 in SD revenue on an account just over a year old at campaign ACoS ranging roughly 10%–36%. Sponsored Display Cost Control (Bid Cap): leave the newer cost-control cap unchecked by default and manage spend through bidding, testing the cap only on already-proven campaigns, since capping an unproven one starves it of the data needed to optimize.
Sponsored Brands Video Ads (Pain-Point-Driven, Low-Budget) covers the other under-used format: short video in search results, consistently under-used relative to the real estate it wins. The prescription is low-budget and informal — studio polish isn't required — built around pain points surfaced by review mining and aimed at the brand's top 3–5 converting keywords rather than broad discovery terms. Before/after and emotional-transformation framing suits it. One sequencing rule: when review mining has identified the niche's top sticking point, make the reassurance addressing it the first call-out in the video — leading with “waterproof” when waterproofing complaints dominate, rather than opening with a generic feature. The payoff is framed as incremental placement and impressions more than immediate ROAS. Pain-Point Hyper-Targeting Advertising is the copy-side counterpart: niche the messaging to one pain point so specific the reader feels “personally attacked” and recognizes themselves instantly. Specificity is the conversion mechanism, and the pain point has to come from real customer language in reviews and search terms, not a guess.
Finally, three ways to point these campaigns at things already known to work. Campaign Chaining (Cross-Ad-Type Target Reuse) duplicates proven targets across ad types — pulling top-converting search terms or ASINs out of a Sponsored Products search-term report and setting them up as targets in a new Sponsored Brands video campaign — described in one walkthrough as “the most important targeting method” for new SB video. It specifically reuses targets with an established track record, not candidates from keyword tools. Related budget guidance: treat automatic campaigns as a narrow harvesting input at roughly 10% of ad budget rather than a set-and-forget default, moving the rest into deliberately chosen exact-match and product-targeting campaigns once targets are validated. Offensive Competitor-Brand Targeting ("Sniping") scales product targeting up to a whole competitor brand — target every ASIN a competitor owns, once existing campaign data confirms your offer converts well against that brand; Rawlings frames it as “sniping” or “siphoning,” built on an established conversion edge rather than speculation. And two bid modifiers segment by buyer rather than placement: Audiences Bid Adjustment (Purchase-Intent Segments) covers three Amazon-defined intent segments (shoppers highly likely to purchase based on recent activity, shoppers who clicked or added to cart, and past purchasers) — start all three at zero adjustment, let the campaign run, then concentrate spend on whichever converts best. Business Buyer Placement Modifier, introduced by Amazon in 2025, raises bids specifically to win impressions in front of Amazon Business buyers, who are described as ordering larger quantities and repurchasing more often — a lever for reaching a higher-LTV segment rather than for driving rank or raw click volume.
Branded search is the one place in this chapter where the material openly contradicts itself, and the disagreement is worth holding in view rather than resolving prematurely.
One side is Branded Keyword Ad Spend Non-Incrementality: money spent bidding on your own brand name mostly captures shoppers who searched for you directly and would likely have bought anyway — “Problem with that is it's not incremental.” It was raised while auditing Essential Candy's spend allocation, where a large share of the agency's budget was going to the brand's own name; mentors treated that, alongside broad-auto spend on unwinnable head terms, as one of two structural leaks. The prescribed alternative is symptom- and ingredient-based long-tail terms — “nausea,” “ginger candy” for a ginger-candy brand — which reach shoppers who don't already know the brand.
The other side is Branded-Search PPC Defense (Conquesting Defense): competitors run offensive brand-targeting campaigns against category leaders, and the defense is to keep your own branded bids high enough to win the auction on your own name even at CPCs the traffic alone wouldn't justify. The explicit instruction is to monitor for competitor presence on your branded keywords and, when found, raise bids defensively rather than optimizing them for ACoS — the objective is traffic retention, not efficiency. This is distinct from ordinary branded-versus-generic budget separation; it's a bid-escalation response to an attack.
Category CPC-Inflation Domination Tactic pushes that posture further and generalizes it beyond brand terms. It runs two things together: heavy defense of branded search, and deliberately bidding up core generic category keywords specifically to inflate the CPC floor, accepting breakeven or loss on those terms. Raising the cost floor makes the category more expensive for smaller competitors, and the margin hit is treated as the price of category-wide share of voice. Unlike a time-boxed launch bidding push, this is an ongoing posture. The operator framing from the SpotMinders/Jungle Powders case is unambiguous: “we don't really look at ACoS, we want to make sure that we dominate it.” That team tracks ACoS only on long-tail efficiency keywords and exempts branded and core category terms from ACoS review entirely, judging them on share of voice instead. The transferable rule is to segment keyword tiers by objective rather than by match type — hold long-tail keywords to a normal ACoS target, and judge domination keywords on a different axis.
One compliance boundary applies to all competitor-brand work. Brand-Name Boundary: Listing Content vs. PPC Targeting: Amazon draws a hard line between using a competitor's brand name in visible listing content and using it as a PPC targeting term. You may not put another brand's name in your title, bullets, or A+ content — that risks trademark complaints and listing suppression — but the same name remains legal to target with sponsored ads, since ad targeting doesn't place the term in your own content. Practically, brand-name matches surfaced by keyword tools should be filtered out of the listing-facing keyword list and kept in a separate PPC target list. Worth knowing too: a listing can surface for a competitor's brand name purely through algorithmic association between converting search terms and similar products, without ever targeting it deliberately. Account-risk consequences of getting this wrong sit in Reviews & Account Health.
Above all of this sits programmatic. Amazon DSP (Demand-Side Platform) reaches consumers on Amazon-owned and partnered properties beyond on-site search — Twitch, Prime Video, partnered news sites, even non-Amazon inboxes like Outlook — and campaigns are typically seeded with audiences defined in AMC rather than plain keywords. In a four-stage funnel model (top, mid, bottom, and cross-sell), DSP's real value sits at top and mid-funnel awareness, because bottom-funnel conversion is already served more cheaply by Sponsored Products. Access historically required a high monthly minimum direct with Amazon; agencies now aggregate multiple clients' spend under one commitment so individual sellers can buy into a shared seat far cheaper. Self-serve automation software can run DSP, but templated execution is described as achieving only a baseline result compared with audience work built from real queries. The Find My tracker brand's approach is instructive: DSP was stood up with a dedicated team whose targeting is aligned to the same strategy as the brand's PPC manager — the same branded-defense and category-domination goals — rather than run as an independent silo.
Amazon Marketing Cloud (AMC) is the data layer under that: up to 25 months of shopper behavior — product-page engagement, cart abandonment, repeat purchases — queried to build custom, persona-based audiences layered on top of ordinary keyword or competitor-ASIN targeting (narrowing a “yoga mat” campaign to women aged 20–30, for example). Those audiences go two ways: back into Sponsored Products as a targeting filter (abandoned-cart audiences, high-converter lookalikes, competitor-engagement lookalikes), or exported into DSP to reach the same personas off-Amazon. AMC also functions as an omnichannel data room — upload first-party data such as email lists to match against Amazon shoppers or build lookalikes, query multi-touchpoint journeys, or look at geographic sales concentration to localize spend. It adds no incremental ad spend by itself, since it only narrows targeting already in place, but it demands ongoing weekly review rather than set-and-forget.
The measurement rule that keeps DSP from being killed for the wrong reason is New-to-Brand as the DSP Success Metric: judge DSP on whether it reaches consumers who haven't previously engaged with the brand, not on short-term profitability or ROAS. A DSP campaign that only retargets existing engagers duplicates what cheaper Sponsored Display already does, so evaluating it on immediate profit undersells campaigns doing genuine top/mid-funnel expansion. If new-to-brand numbers are low, redirect that budget to Sponsored Display rather than dropping DSP altogether. The metric also appears in Brand Analytics generally and in Ad Console campaign reports, where its growth percentage works as a leading indicator of acquisition and can be cross-checked against new-to-brand ROAS/ACoS.
Be aware that this is where the chapter is thinnest. The DSP and AMC material is framing-level: no worked AMC query, no stated spend threshold at which a shared seat makes sense, and no numeric benchmark for what a healthy new-to-brand share looks like. Treat it as a map of what exists and how to judge it, not as an operator's runbook of the kind the day-parting and triage sections provide.
Five-Layer PPC Tech Stack is a useful closing frame for the whole chapter: scaling PPC means assembling a connected stack across five layers rather than leaning on one tool — research (Helium 10 Black Box, Jungle Scout, Data Dive), campaign management (Amazon Ads Console, Perpetua, Pacvue), optimization (Teikametrics, Cortile, Ad Badger), analytics (AMC, Amazon Marketing Stream, Amazon Attribution), and conversion (PickFu, managed A/B experiments, Canva). The application is to map your existing tools onto the layers and close gaps one at a time, treating it as a pipeline: research feeds campaign structure, campaign results feed optimization rules, and analytics and conversion data feed back into targeting and creative. That pipeline is the same loop this chapter has been describing; scaling it across a brand family is the subject of Scaling, Omnichannel & Brand Growth.