How to determine share of voice (and what it actually tells you)

TL;DR: Share of voice (SOV) measures the percentage of total category visibility your brand owns versus competitors. The core formula is: (your brand mentions or impressions / total category mentions or impressions) × 100. Calculate it separately for paid search, organic search, social media, and AI assistant citations, because those channels diverge sharply and a blended number hides what's actually happening.

What is share of voice and why does the definition keep shifting?

Share of voice started as an advertising metric in the 1970s, rooted in a simple idea: the brand spending the largest fraction of category ad dollars usually wins the largest fraction of category sales. The original SOV equation compared your media spend to total category spend. That held up reasonably well when television and print absorbed most budgets.

The definition cracked when digital search arrived. Impressions and clicks became easier to measure than dollars, so practitioners started defining SOV as the share of total search impressions a brand captured for its relevant keyword set. Then social listening platforms defined SOV as share of brand mentions across forums, reviews, and social posts. Then AI assistants arrived and introduced a fourth variant: the share of relevant AI-generated responses that mention your brand by name.

All four definitions are legitimate. The problem is that most dashboards report one number without flagging which variant it is. A brand can have 40% paid SOV, 12% organic SOV, 28% social SOV, and 6% AI SOV simultaneously. Those are not interchangeable. If someone pitches you a single "share of voice" figure with no channel qualifier, ask which one before making any decisions.

This article covers how to calculate each variant, what the number actually predicts, and where most measurement efforts go wrong.

What is the share of voice formula?

The base formula is the same regardless of channel:

SOV = (Your brand metric / Total category metric) × 100

What changes is what you put in as the metric. Here is how that breaks down by channel:

| Channel | Numerator | Denominator | |---|---|---| | Paid search | Your ad impressions for target keywords | All advertisers' impressions for same keywords | | Organic search | Pages where your brand ranks / visibility score | Sum of visibility scores for all tracked competitors | | Social / PR | Brand mentions of your brand | Total brand mentions across all tracked brands | | AI assistants | AI responses citing your brand | Total AI responses sampled for category queries |

For paid search, Google Ads calls this metric "Search Impression Share" and defines it as impressions received divided by the estimated number of impressions you were eligible to receive [1]. That eligibility-based denominator is different from a raw total, so Google's native number is not directly comparable to a manually calculated competitive SOV unless you pull impression share data from every competitor, which you can't do natively in Google Ads. You'd need an auction insights report or a third-party tool for that.

For organic search, most tools (Semrush, Ahrefs, Sistrix) calculate a visibility score based on keyword rankings weighted by estimated click-through rate at each position. Your SOV is your visibility score divided by the sum across all tracked competitors. The denominator is not the entire internet. It's your defined competitive set, which makes competitor selection the single most consequential methodological choice you'll make.

For social and PR, the denominator is bounded by what your listening tool indexes. Sprout Social, Brandwatch, and Meltwater each crawl different source sets, so the same brand on the same day will show different SOV numbers across platforms. Agree on one tool and stick with it. Switching mid-year breaks your trend line.

For AI assistants, the methodology is newer and less standardized. The practical approach is to assemble a list of queries a buyer in your category would ask an AI assistant, run them repeatedly across ChatGPT, Claude, Gemini, and Perplexity, and count citation rates by brand. BrightEdge published an analysis in 2024 finding that AI Overviews appeared on roughly 84% of informational search queries in some categories [2], which signals how fast AI surfaces are becoming part of the SOV calculation for most brands.

How do you define your competitive set for the denominator?

This is where most SOV calculations break down quietly. A too-narrow competitive set flatters your number. A too-broad one buries it. Neither is honest.

Start with direct competitors: brands the same buyer would seriously consider instead of yours. Then add category challengers: brands growing fast enough to matter in 18 months even if they're small today. Then add indirect substitutes: alternatives that solve the same problem differently. Stop there. If your denominator includes 40 brands, your SOV will look small and become hard to move meaningfully.

For most mid-market B2B companies, five to eight competitors in the denominator is the right range. Consumer brands with heavily fragmented categories can reasonably go up to twelve before the metric loses actionability.

Review the competitive set quarterly. Brands get acquired, pivot, or collapse. A competitor that commanded 20% SOV and then exited the market inflates everyone else's number mechanically. That's not a strategy win. It's an artifact.

One practical test: if a sales team can't name a brand you've included, it probably doesn't belong in your SOV denominator. Conversely, if reps keep losing deals to a brand not in your set, add it.

Excess share of voice and market share growth

| | | |---|---| | Small-share brands (ESOV +10 pts) | 0.7% | | Mid-share brands (ESOV +10 pts) | 0.5% | | Large-share brands (ESOV +10 pts) | 0.3% |

Source: IPA Databank analysis (IPA, The Long and the Short of It)

How do you calculate share of voice for organic search?

Organic search SOV is the most commonly tracked version for brands without large paid budgets. Here's a step-by-step approach:

1. Define your keyword universe. Start with the keywords that actually drive qualified traffic, not vanity terms. For most B2B brands this is 50 to 300 keywords; for large e-commerce it can be thousands.

2. Pull ranking data for your brand and all competitors across those keywords. Tools like Semrush, Ahrefs, or Moz do this at scale.

3. Weight each keyword by search volume and by the estimated click-through rate at the ranking position. A rank-1 position on a 10,000-volume keyword is worth far more than rank-3 on a 500-volume keyword.

4. Sum each brand's weighted visibility scores and divide each by the total to get SOV percentages.

Semrush's own methodology page explains their Visibility metric as an estimation of how often a domain appears in the top 100 search results, weighted by position-based CTR assumptions [3]. The exact CTR curve they use is proprietary, but the principle is public.

A practical benchmark: in competitive B2B SaaS categories, the market leader typically holds 20 to 35% organic SOV among a set of five to eight tracked competitors. The second-place brand is usually in the 15 to 25% range. If you're tracking your set honestly and you're below 10%, you have a content or authority gap more than an SEO gap.

AI is changing this calculation materially. Google's AI Overviews now appear above traditional organic results for a large share of informational queries [2]. If the AI Overview cites a competitor but not you, you can rank third organically and still lose the visibility. That's why you now need to track AI citation share alongside traditional organic SOV. See the section below on AI share of voice for how to do that.

How do you measure share of voice for paid search?

Google Ads exposes Impression Share (IS) natively in campaign and keyword reports [1]. This gives you your own paid SOV directly. What it doesn't give you is competitors' raw impression counts.

For competitive paid SOV, you have two approaches. First, use Google's Auction Insights report, which shows impression share, overlap rate, outranking share, and top-of-page rate for all advertisers bidding on the same auctions as you. It doesn't name every competitor in the market, only those who overlap with your campaigns. If a competitor doesn't bid on your keywords, they won't appear.

Second, use a third-party tool (SpyFu, iSpionage, Semrush Advertising) that estimates competitors' paid volume by crawling search result pages at scale. These are estimates with meaningful error bars, not exact figures. Treat them as directional rather than precise.

For calculating your paid SOV from Auction Insights: take your impression share, find competitors' impression shares, and express each as a percentage of the sum. If your IS is 45%, Competitor A is 30%, and Competitor B is 25%, your paid SOV in that auction set is 45%.

Target impression share (the goal of bidding to show at the top of page X% of the time) is a useful benchmark. Google's own help documentation notes that top-of-page impression share is the impressions received in the top positions above organic results, divided by the estimated impressions available [1]. Knowing that number tells you how often you're actually winning the most valuable paid real estate.

How do you track share of voice on social media?

Social SOV is the share of online brand mentions your brand owns within a defined category conversation. The numerator is mentions of your brand; the denominator is mentions of all tracked brands combined.

The core challenge is source coverage. Twitter/X, Reddit, Facebook, Instagram, TikTok, LinkedIn, review sites, and news outlets each have different access rules. No listening tool indexes all of them equally. Brandwatch has broad academic and forum coverage; Sprout Social is stronger on managed social; Meltwater adds news wire. The right tool depends on where your category conversation actually happens.

For B2B software, LinkedIn and Reddit threads are often more material than Twitter. For consumer packaged goods, TikTok mentions may outweigh all others combined. Map where your buyers actually talk before choosing a tool, not the other way around.

Once you have a tool, the setup is: define brand mention queries for each competitor (including misspellings and abbreviations), set the same date range and geographies, and export total mention counts. Your SOV = your mentions / sum of all mentions × 100.

Volume alone misses sentiment. A brand with 40% SOV that is predominantly negative mentions is in a worse position than a brand with 20% SOV and strongly positive sentiment. Run sentiment analysis alongside volume and report both. Some practitioners prefer "positive share of voice" where the denominator is only positive mentions across the set, which gives a cleaner signal for brand health.

The benchmark that holds up across multiple industry studies: brands with SOV significantly above their market share tend to grow toward that SOV level over time. The IPA (Institute of Practitioners in Advertising) analyzed data from 886 campaigns and found that brands with SOV above their market share (excess share of voice, or ESOV) grew at an average rate of 0.5 percentage points of market share per 10 points of ESOV [4]. That's not a law, but it's the best empirical estimate available for the SOV-to-growth relationship.

How do you measure share of voice in AI search and assistant responses?

AI share of voice is the newest and least standardized variant, but it's becoming the one that matters most for informational and consideration-stage queries. The methodology requires more manual setup than traditional channels, but it's not complicated.

Step 1: Build a query list. Write out 30 to 100 questions a buyer in your category would actually ask ChatGPT, Gemini, Claude, or Perplexity. Include category-level questions ("what's the best [category] for [use case]"), comparison questions ("[your brand] vs [competitor]"), and problem-led questions ("how do I [problem your product solves]"). This query list is your measurement universe.

Step 2: Run queries and record outputs. You need to run each query multiple times and across multiple AI platforms, because outputs vary by run. Record which brands are mentioned by name, whether they're cited positively or neutrally, and whether a source link is included.

Step 3: Calculate citation rate. For each brand: AI SOV = (number of responses citing your brand / total responses sampled) × 100. Do this per platform and in aggregate.

Step 4: Repeat on a regular cadence. AI model updates change citation patterns. What GPT-4 cited in January may differ from GPT-4o's outputs in June. Monthly measurement is the minimum; weekly is better for fast-moving categories.

Seer Interactive published analysis in 2024 showing that AI Overviews in Google disproportionately cited sources that ranked in the top 10 organic results for the same query, but with meaningful exceptions where sources ranking outside the top 10 still received AI citations if they had strong entity authority [5]. That finding matters: traditional organic SOV and AI SOV are correlated but not the same metric.

For teams that want a structured way to run this at scale rather than manually, tools built specifically for AI visibility tracking make the query automation and citation logging much faster. Spawned's AI visibility audit is one approach if you want a starting baseline without building the infrastructure yourself. ai-search-visibility-metrics-kpis has more on which metrics to track alongside citation rate.

The IPA's long-run dataset and the Seer finding together point to the same practical implication: brands with genuine content authority in a category show up more in both traditional and AI-driven SOV. The mechanism differs (ranking algorithms vs. model training and retrieval), but the input, real expertise expressed in indexed content, is the same.

What tools do professionals use to measure share of voice?

There is no single tool that covers all four SOV variants well. Here is an honest breakdown:

| Channel | Primary tools | Limitations | |---|---|---| | Paid search | Google Ads Auction Insights, SpyFu, Semrush Advertising | Auction Insights only shows overlapping bidders; third-party estimates are approximations | | Organic search | Semrush, Ahrefs, Sistrix, Moz | Keyword universe must be defined manually; CTR curves are proprietary | | Social / PR | Brandwatch, Sprout Social, Meltwater, Mention | Coverage varies by platform; Reddit/TikTok indexing is incomplete in most tools | | AI assistants | Manual query sampling, BrightEdge, Semrush AI, Authoritas, Spawned | Category is nascent; no tool has complete coverage of all AI outputs |

For organic search, Semrush and Ahrefs are the industry defaults. Sistrix is particularly strong for European markets. Moz is a reasonable budget option but its index is smaller.

For social, Brandwatch is the enterprise standard with the broadest source coverage. Sprout Social is better for teams that also manage social publishing and want SOV reporting inside the same workflow. Mention is a lower-cost option adequate for smaller competitive sets.

For AI assistant SOV, the tooling is genuinely early. Most enterprises are still running this manually or with custom scripts. BrightEdge added AI-specific tracking in 2024 [2]. Semrush launched AI-related features in their content marketing toolkit. Authoritas has AI Overview tracking. Expect this space to consolidate significantly over the next 18 months.

See ai-seo-tools for a fuller breakdown of AI-specific measurement tools and what they actually track.

How often should you measure share of voice?

The right cadence depends on how fast your category moves and what you're using the data for.

For strategic planning, monthly SOV snapshots are usually sufficient. A monthly trend line catches meaningful shifts without creating noise from day-to-day ranking fluctuations.

For campaign-level optimization, weekly measurement lets you tie SOV changes to specific actions (a content push, a PR hit, a competitor's product launch). Daily measurement is rarely worth the overhead unless you're running a major campaign or your category has genuinely high news velocity.

For AI assistant SOV specifically, the cadence should be at minimum monthly, because model updates from OpenAI, Google, Anthropic, and Perplexity happen frequently enough to cause citation pattern shifts that aren't tied to anything your brand did. Measuring quarterly will cause you to miss those shifts and misattribute the change.

One practical rule: pick a cadence you'll actually maintain. A monthly measurement that runs consistently for 12 months is worth more than a weekly measurement that gets skipped whenever the team is busy. SOV is a trend metric. Single data points have limited value. The trend line is everything.

What does share of voice actually predict about business outcomes?

SOV is a leading indicator, not a lagging one. That's its value. Revenue shows you what happened; SOV shows you what's likely to happen.

The strongest empirical link between SOV and business outcomes comes from the IPA's Databank, which aggregates effectiveness data from hundreds of advertising case studies across decades. Their analysis confirmed the ESOV principle: "Brands that maintain excess share of voice grow market share over time, at a rate of approximately 0.5 percentage points of market share per 10 percentage points of ESOV" [4]. This relationship held more strongly for brands with smaller market shares than for dominant category leaders.

For digital-native businesses, a 2021 analysis in the Journal of Marketing Research found that organic search visibility changes led revenue changes by approximately six to nine months in tracked e-commerce categories [6]. That lag matters for planning: improving your organic SOV today may not show in revenue for two quarters.

AI assistant SOV is too new for longitudinal outcome data. The reasonable hypothesis, supported by the Seer Interactive analysis [5], is that brands cited in AI responses during consideration-stage queries will see higher branded search volume and higher direct traffic over time, because AI recommendations function similarly to word-of-mouth from a trusted source. Nobody has a multi-year dataset on this yet. The honest answer: we think it predicts business outcomes similarly to organic SOV, but the causal evidence will take another two or three years to accumulate.

What SOV doesn't predict well: it won't tell you anything about conversion rates or revenue per customer. A brand can dominate SOV and still underperform on revenue if its messaging lands broadly but its product converts poorly. Use SOV alongside conversion and retention metrics, not instead of them.

What are the most common mistakes when calculating share of voice?

These are the errors that reliably produce misleading numbers in practice.

Mixing channel definitions. Reporting a blended SOV that combines paid, organic, and social into one number. This sounds thorough and is actually meaningless. Each channel has different cost structures, different audience behaviors, and different points of influence. Separating them is non-negotiable.

Too-narrow competitive set. Including only two or three direct competitors makes your SOV look healthy when you're actually being outflanked by brands you haven't tracked. Review the set quarterly and add fast-growing challengers early, not after they've already taken share.

Not controlling for query scope. For organic search SOV, if you add 50 new keywords to your tracked set, your competitors' SOV will change even if nothing actually changed. Treat your keyword universe as a fixed denominator for any trend comparison, and document when you change it.

Confusing sentiment with volume. A spike in brand mentions often looks like a SOV win. If those mentions are complaints or negative press, it's the opposite of a win. Always run sentiment alongside volume.

Ignoring AI surfaces. Many SOV reports still don't include AI assistant citations, AI Overviews, or AI-generated summaries in social platforms. Given that BrightEdge found AI Overviews appearing for a high share of informational queries [2], omitting this channel means your SOV measurement is now structurally incomplete for the fastest-growing part of search.

Static snapshots without trend lines. A SOV number at a single point in time tells you almost nothing. You need at least six months of consistent measurement before the trend becomes interpretable. Start measuring before you feel like you need to.

How do you improve your share of voice once you've measured it?

The levers differ by channel, but a few principles apply everywhere.

For paid search SOV, the fastest lever is budget and bid strategy. If your impression share is limited by budget (Google tells you this in the "lost IS (budget)" column), increasing spend directly increases SOV. If it's limited by rank, improving Quality Score through better ad-to-landing-page relevance is more efficient than just raising bids.

For organic search SOV, the inputs are content depth, site authority, and technical health. The brand that owns the most genuinely useful content across a keyword category tends to accumulate organic SOV over time. The lag between content investment and ranking improvement is real. Expect three to six months before new content materially moves your visibility score.

For social SOV, earned media (press coverage, influencer partnerships, community participation) moves the needle faster than owned posting alone. If a competitor is dominating social SOV with product launches or PR hits, trying to outpost them on volume is inefficient. Being more newsworthy is more effective.

For AI assistant SOV, the inputs that matter are entity authority (how clearly AI models understand who you are and what you do), citation in high-authority sources (the sources AI models retrieve from), and structured information about your brand available in places like Wikipedia, Wikidata, and well-cited industry publications. The generative-engine-optimization article covers the specific tactics for improving AI citation rates in detail.

One thing that moves all four channels at once: genuine category leadership expressed in content. Brands that publish original research, own data, or produce the most cited resources in their category tend to accrue SOV across paid (because their brand terms get searched more), organic (because external sites link to their research), social (because people share their data), and AI (because models retrieve well-cited sources). That's not a quick fix, but it's durable in a way that individual channel optimizations are not.

For brands starting this measurement process, Spawned's AI visibility audit can give you a baseline AI SOV score across the major assistant platforms before you build out a full internal measurement system. It won't replace ongoing tracking, but it tells you where you stand today. You can also explore ai-visibility-tool options to understand the tool landscape.

Sources

1. Google Ads Help – About impression share
2. BrightEdge – AI Search Research and Generative AI Trends
3. Semrush – Visibility metric methodology
4. IPA (Institute of Practitioners in Advertising) – The Long and the Short of It / Databank
5. Seer Interactive – AI Overviews SEO Analysis 2024
6. Journal of Marketing Research – Organic Search and Revenue Lag Analysis
7. Google Search Central – How Google Search works
8. Brandwatch – Social listening methodology
9. Ahrefs – Domain Rating and organic visibility methodology
10. Perplexity AI – About Perplexity