Amazon Share of Voice: What to Change When the Number Moves

Every other number in your ad account moves because you moved something. Share of voice moves because somebody else did. You can hold spend flat, leave bids alone, ship nothing, and still watch your share of a head term fall four points in a week because a competitor launched a variation and started buying the top of search on Tuesday.

That makes it the most useful competitive signal Amazon hands you and the easiest one to misread. Two failure modes show up over and over. The first is treating every dip as an attack, bidding into a query that was never converting well, and turning a measurement question into a spend problem. The second is watching share erode for a quarter, filing it under seasonality, and finding out at the quarterly review that a new entrant now holds the click share on your best term along with the purchase history that comes with it.

Both are avoidable with a review that takes about two hours to set up. What follows is which of Amazon's reports carries which signal, how to separate a real shift from noise before you spend against it, and the six patterns that tell you whether the fix is bids, creative, or the offer. A newer question now sits beside all of it: which brands AI shopping assistants recommend for the same queries. That is a separate measurement with its own section below, and it should not be blended into the ad numbers.

The definition side of this is settled ground and we have already covered it. Share of voice versus share of shelf explains what each metric counts and where the two get confused. Competitor pricing sits outside this piece on purpose. Price needs its own decision loop with margin floors and inventory cover attached, and mixing it into the ad review produces changes nobody can attribute afterwards. If that is what you came for, start with dynamic pricing versus repricing and the Amazon pricing strategy guide instead.

Quick answer

• What it is for. Catching a competitor taking one of your best terms while the response is still cheap, and stopping yourself from spending against a drop that was never competitive in the first place.
• Amazon publishes no single number called share of voice, so the skill is knowing which report answers which question.
• Brand share of impressions, clicks, cart adds and purchases, in the Search Query Performance report. Tells you where in the funnel you are losing a query.
• Search term impression share and rank, for Sponsored Products. Tells you how you stand against every other advertiser bidding on that term.
• Top-of-search impression share, at campaign level. Measures your own eligibility, so a drop here is usually your budget, bid or stock rather than a competitor.
• Read the denominator first. The larger number your share divides into is what tells you whether a drop was caused by competitors or by you.
• Act only on a confirmed move: past a threshold you wrote down in advance, surviving a second consecutive week, and not explained by a change in query volume or in your own ASIN mix.
• AI share of voice is a separate shelf. It counts which brands ChatGPT, Claude and similar assistants recommend for a shopper query. Track it next to ad share of voice, never inside it, because different levers move each one.
• Then read the pattern. Impression share, click share and purchase share together tell you whether the fix is bids, creative, or the offer.

Which Amazon reports carry share of voice, and what they leave out

Four different surfaces carry a piece of the picture. None of them names a competitor.

Search Query Performance gives you funnel share, not ad share

The Search Query Performance report in Brand Analytics (Seller Central, under Brands, then Brand Analytics, then Search Analytics) is the closest thing Amazon gives you to a market-share view of a query. For each search query it reports search query volume plus total count and brand count at four stages: impressions, clicks, cart adds and purchases, with your brand share at each stage. You can toggle between brand view and ASIN view, and set the reporting range to weekly, monthly or quarterly. We have a fuller walkthrough in the Search Query Performance report guide.

Two properties matter for competitive reading. The counts combine organic and paid, so a share drop here does not tell you which of the two moved. And the report only contains queries your brand already appeared on, which means the terms where you have zero presence, the ones a competitor may be quietly building, do not show up at all.

Search term impression share and rank show the ad auction

The search term impression share report for Sponsored Products is the one that measures you against the field. Amazon's own description: a Sponsored Products impression share of 20% for a search term means you won 20% of all Sponsored Products ad impressions for that term on that date. The companion metric, search term impression rank, gives your ordinal position among advertisers on the term. Rank 3 means you took the third most Sponsored Products impressions that day.

Rank is the more actionable of the two and the more misleading. It tells you the order and nothing about the spread. Moving from rank 4 to rank 6 can mean two competitors gained a point each, or it can mean one advertiser tripled a budget and swallowed a third of the term.

Top-of-search impression share measures your eligibility, not the field

This is where most weekly reviews go wrong. Amazon defines top-of-search impression share as the percentage of top-of-search impressions your campaign received out of the total top-of-search impressions it was eligible to serve on, available for Sponsored Products and Sponsored Brands with a 90-day lookback. The denominator is your own eligibility, not the whole auction.

A falling top-of-search impression share with a flat search term impression share is almost never a competitor. It is a budget exhausting mid-afternoon, a placement modifier someone dialled back, a bid sitting under the top-of-search floor, or an ASIN that went out of stock. Fix pacing and stock before you touch bids.

The Brand Impression Share report covers your branded terms

In the ads console under Insights and Planning, the Brand Impression Share report tracks top-of-search impression share on your brand terms over time, including a Lost Top-of-Search Sales metric that estimates the sales sitting behind visibility you did not win. Amazon's reserve share of voice guide points advertisers at the same report. This is the surface to use when the question is defensive rather than expansionary, and it pairs with the decisions in our post on brand defense.

Amazon also now sells a way to buy the answer directly. Sponsored Brands reserve share of voice is a goal-based campaign type that reserves exact-match branded keywords for the Sponsored Brands top-of-search placement at a fixed upfront CPM rather than through the auction. It is limited to terms Amazon validates as yours, which means it solves branded defense and does nothing for generic head terms.

Nothing in Amazon's reporting names the competitor

This is the gap that sends operators to third-party tools, so be precise about it. Amazon tells you the size of the gap. It does not tell you who is on the other side of it. Search term impression share gives you a percentage and a rank with no roster attached. Search Query Performance gives you a brand share with no breakdown of the remainder. Identity has to come from observing the search results page, which is a sampled read rather than a census, and every tool that offers it is inferring from scrapes.

Read the denominator before you read the number

Every share number is a division: your count on top, a larger count underneath. That larger count is the denominator, and it is the half nobody checks. The interesting question in any share move is which side of the division changed, and that single habit resolves most of the arguments a weekly review produces.

Query volume is the first candidate. Search Query Performance reports search query volume alongside the counts, so you can see when the term itself got bigger. If volume doubles going into a seasonal peak and your brand count holds, your share halves while your absolute presence is unchanged. Nothing competitive happened. Something arithmetic did.

Your own catalog is the second. Brand share aggregates across every ASIN you own on that query. Add two variations and brand count rises without anyone losing anything. Delist a slow mover, suppress a listing over a compliance flag, or let a child ASIN go out of stock, and brand share drops with no auction event behind it.

The third is the one worth naming out loud, because it is where the two most-quoted metrics disagree. Search term impression share and top-of-search impression share sound like siblings and answer opposite questions. One is your slice of everybody's impressions on a term. The other is your slice of the impressions you personally qualified for. When they move together, the market moved. When they diverge, you moved.

Separate a real shift from noise before you spend against it

Write these rules down before the week you need them. A threshold invented in the middle of a bad Monday is a threshold bent to fit the mood of the room.

Set a volume floor. On a query with a few hundred searches a week, a point of share is a handful of impressions and a single shopper's session can move it. Decide the minimum query volume that makes a term eligible for a share decision at all, and let everything below the floor sit in a watchlist rather than a work queue.

Use both an absolute and a relative threshold. Forty percent down to thirty-six is four points and ten percent relative. Four percent down to two is two points and half your presence. The second is the larger event and the smaller-looking number. A rule that only reads point movement will chase the wrong one every time.

Require a second week. Search Query Performance is weekly at its finest grain, which is a reasonable confirmation window. One week in a direction is a data point. Two consecutive weeks in the same direction on the same term is a signal. The exception is a rank collapse of several places at once on a head term, which is worth looking at immediately even before it confirms.

Check your own change history first. If share fell on every term in the same week, the cause is almost certainly inside your account: a budget cap, a bulk bid change, a campaign someone paused, a suppression event. Competitors do not coordinate.

The decision rules: what to change when share of voice moves

The pattern across impression share, click share and purchase share is diagnostic. Each combination points at a different part of the funnel, and each one has a change that helps and a change that wastes the week.

Defend or concede is a margin call

The hardest row in that table is the new entrant. Someone shows up on your best term, takes eight points of impression share, and the reflex is to bid until they leave. Sometimes that is right. It is right when the term converts above your account average, when your contribution margin on the ASIN behind it has room, and when losing the term threatens the query's purchase history rather than just its traffic.

It is wrong when the term is broad, converts below average, and was already being subsidised by the rest of the account. Conceding a term you were losing money on is a decision, and it should be written down as one, with the reason and the date, so the next person to review it does not quietly reverse it.

If you want the bid mechanics behind that call, the Amazon impression curve post covers how impressions respond to bid movement across the idle, growth and top-of-search zones, which is what determines whether defending a term costs a little or a lot.

What to watch weekly beyond the share of voice numbers

Share metrics tell you the score. Competitor ad intelligence tells you what changed on the field. Keep this list short enough that it survives a busy week.

• Who holds Sponsored Brands top-of-search on your top 20 head terms. The banner slot changes hands slowly, so when it changes it means something.
• New ASINs appearing in the first row of Sponsored Products results. Entrants show up here before they show up in your share numbers.
• Sponsored Brands creative changes. A competitor moving from a collections ad to video on a specific term is a spend commitment, not a design refresh.
• Pressure on your branded queries. Who is bidding on your brand name, and whether the Lost Top-of-Search Sales figure on those terms is trending.
• ASIN targeting on your own detail pages. Product-page ad slots on your listings are competitor buys against your traffic, and they never appear in any search-term report.
• Terms where you have no presence at all. Search Query Performance will not surface these, because it only contains queries your brand appeared on. This gap needs either a rank tracker or a periodic manual sweep of category head terms.

Two things stay out of this list on purpose. Competitor price moves belong to a separate loop, for the reason given at the top. And competitor keyword monitoring is worth running as a diff rather than a dashboard: what changed since last week, on which terms, is a five-line answer. The full standings table is a report nobody reads twice. For a starting point on the wider competitive picture, our Amazon competitor analysis guide covers the non-ad surfaces, and the free competitor research tool will get you a first pass on who you are up against.

AI share of voice is a separate shelf: which brands the assistants recommend

Everything above measures Amazon's own surfaces. Shoppers now also ask an assistant what to buy before they ever reach a search results page, and the brands named in that answer are a different set from the brands holding your top-of-search slots. Our guide to Rufus and Sponsored Products covers the conversational shopping layer inside Amazon, and our breakdown of the Amazon Ads and ChatGPT pilot covers where paid placement is starting to show up outside it. This section covers the organic side: whether the assistants recommend you at all.

Ad share of voice and AI share of voice answer different questions

Ad share of voice divides your presence by everyone's presence on a query. AI share of voice counts how often your brand appears in the recommendations an assistant gives for a set of shopper questions, against the other brands named alongside you. The two can point in opposite directions on the same term. A brand can hold a strong share of impressions on "protein powder for weight loss" and never be mentioned when someone asks an assistant the same question, and the reverse happens too.

How to run an AI share of voice scan

In Qore, the AI Share of Voice tool runs the scan for you. You enter the shopper queries and your brand name, and it reports which brands assistants such as ChatGPT and Claude recommend in their answers, with your brand's standing against the rest. Use the same steps whether you use the tool or build the check by hand.

1. Start from queries you already track. Take the three to ten head terms from your Search Query Performance review and rewrite each the way a shopper would ask an assistant: "what is the best collagen powder for skin" rather than "collagen powder".
2. Add the questions with no ad equivalent. Comparison and problem queries ("X vs Y", "what should I buy if I have Z") rarely map to a keyword you bid on, and they are where assistants make their most consequential recommendations.
3. Set your brand in the scan and repeat it on a schedule. One answer is a sample. Rerun the same query set weekly or monthly and read the movement, exactly as you read ad share of voice.
4. Record who is named, in what position, and with what stated reason. The stated reason (price, ingredient, reviews, brand reputation) is the part you can act on.

Read the two shares together before you change anything

Put ad share of voice and AI share of voice for the same term side by side and the pattern tells you where to work.

• High ad share, low AI share. You are buying visibility the assistants do not back up. Look at the listing content, review themes and product claims the assistant is quoting for the brands it does name, before adding spend.
• Low ad share, high AI share. The assistants already recommend you. Defend the branded terms and do not bid into a head term to fix a gap the recommendation is covering.
• Low on both. This is a category-entry problem. It is a listing, reviews and content project, and ads will not close it alone.
• High on both. Protect it. Note the date and the conditions so the next reviewer knows what was working.

The limits of an AI share of voice number

Assistants do not publish how they choose which brands to name, and their answers change with phrasing, session and model updates, so treat any single scan as directional. The scan shows who is recommended and how the answer frames them. It does not show clicks, purchases or revenue, and it cannot tell you whether a recommendation converted. It also leaves your Amazon share of voice numbers untouched: a better AI standing will not show up in Search Query Performance on its own. What it adds is competitive intelligence for the content side, meaning listings, blog and SEO work, because it shows how an assistant positions you against the brands you compete with.

Where the weekly share of voice review breaks down

None of this is hard. It is repetitive and it is judgment-heavy at the same time, which is the combination that defeats a process.

Run the honest version of the arithmetic. Thirty head terms, five share metrics each, three sanity checks per movement, two brands, every Monday. A careful analyst gets through it. The work is not what fails. The standard is: the person who builds the review in week one applies a five-point threshold and a two-week confirmation, and the person covering in week six eyeballs it. By week ten nobody can say whether the erosion on your third head term is real or whether the review changed underneath it, and the answer to "why did share drop" becomes an opinion rather than a finding.

Exporting to a general-purpose AI tool helps with the reading and not with the standard. It will summarize the file quickly and it will summarize it slightly differently next week, which is the specific thing you cannot have when the output is a trend.

Where Qore fits

Qore is where you write that Monday review down once and have it run the same way every week: the volume floor, the two-week confirmation, the defend-or-concede prompt, the six diagnostic patterns from the decision table, applied identically on run one and on run forty. It is an approval log with an automation component. A scheduled skill reads the share reports, applies the rules you wrote, and proposes the change, with the approval recorded against the run that produced it so you can argue with the logic rather than guess at it.

How much of that needs your sign-off is a setting you tune. A new skill starts with every change gated, which is right while you are still learning whether its calls match yours. Share of voice is a good candidate for loosening it later, because the diagnostic patterns are stable: a top-of-search impression share drop with flat search term impression share is a pacing fix, every time, and once you have watched a skill make that call correctly for a month there is little value in queueing it. Tighten the rules until the mechanical cases run themselves and the defend-or-concede calls, which turn on margin rather than on a pattern, still come to a person.

Qore reads and analyzes any Amazon account and makes the change you approved regardless of what runs the bids. It sits alongside whatever bid manager you already use, and it works just as well if you run no automation platform at all. If a third-party tool later overwrites a change you approved in Qore, that is a constraint of that tool's own optimization loop, and Qore's change history will show you exactly when it happened. Qore is in open beta, and you can sign up here.

If you would rather the bid and placement changes run continuously instead of one approval at a time, that is Qinetix, our ads automation platform: per-campaign optimization logic, bidding zones with an operator-set floor and ceiling, and a change log that lets next month's review see what last month's review did. Ads and pricing run as parallel mechanisms there with shared visibility, which is why the price side of competitive intelligence stays its own loop rather than getting folded into this one.

Visibility into competitors is not a strategy. A share report can tell you a gap exists and how big it is. It cannot tell you whether the gap is worth closing, because it does not know your contribution margin on the SKU behind the term, and it will never tell you why the other advertiser moved. Someone still has to make the call. The value of writing the review down is that the call gets made against the same standard every time, rather than that it gets made for you.

Start with three terms, not thirty

Pick your three highest-volume head terms. For each one, pull the last eight weeks of brand share of impressions, clicks and purchases from Search Query Performance, and the last eight weeks of search term impression share and rank from the ads console Report Center. If you have never exported either of those before, both are self-serve: Brand Analytics sits under the Brands menu in Seller Central, and the search term impression share report is a checkbox on the Sponsored Products report builder. Put the two exports side by side and write down what your volume floor, your absolute threshold and your relative threshold are going to be. That is a two-hour exercise and it will settle at least one argument you have been having for a month.

Then run it weekly, in the same order, with the same rules, and change one thing per confirmed signal. Amazon share of voice is a measurement of a market you share with people who did not consult you. The part you control is what you do on the Monday after the number moves.

The bottom line for ecommerce teams

• Check the denominator before the number. A share drop when query volume doubled is arithmetic, not competition.
• Know which impression-share metric you are reading. Search term impression share measures you against the field. Top-of-search impression share measures you against your own eligibility. Confusing them is the most expensive mistake in the weekly review.
• Set a volume floor, an absolute threshold and a relative threshold in advance. Four percent down to two is half your presence and looks smaller than forty down to thirty-six.
• Require two consecutive weeks before acting, with one exception: a multi-place rank collapse on a head term is worth looking at the day you see it.
• Share falling on every term at once is your account, not the market. Open the change history and listing status before building a competitive narrative.
• Defend or concede is a margin decision. Write down the reason and the date so the next reviewer does not quietly reverse it.
• Measure AI share of voice separately. Which brands assistants recommend is a different shelf from ad share of voice. Read them side by side and never average them.
• Nothing Amazon reports names the competitor. Identity comes from scraped, personalized pages, so read it as a trend rather than a measurement.