Share of voice is your brand's portion of the attention in a market: your number divided by the total for you and your competitors, times 100. In search that number is clicks, on social mentions or engagement, and in AI answers brand mentions across a prompt set.
This guide runs one hypothetical competitor set through all three channels so you can see the arithmetic, then shows the free and the paid way to collect each number. Every figure in the worked example is invented for illustration.
Key takeaways
- One formula, three inputs. Your metric divided by the total for you plus competitors, times 100. Search uses estimated clicks, social uses mentions or engagement, AI answers use brand mentions across a fixed prompt set.
- The denominator is a choice. The competitor set and the keyword or prompt list decide the result. Freeze them before you compare one month with the next.
- Do not average channels. A blended number hides that you can lead in AI answers and trail on social.
- A good number is relative to market share. Above it points to growth, below it points to decline.
- You can measure it free. About three hours a month by hand, or a few minutes a month with the tools below.
Video: the channels and the formula in four minutes
The first two minutes are a clean summary of which metric to use on each channel, which is the decision most SOV reports get wrong.
Three things worth taking from it:
- Each channel has its own metric. Mentions, tags and comments for social; estimated traffic for a keyword set in search; impression share for paid search; article mentions for PR.
- The formula does not change. Your brand's number divided by the total for your brand plus competitors, multiplied by 100.
- A drop in paid impression share is a signal. The video points to bidding or targeting problems as the usual cause.
Where we differ: the second half is a pitch for ClickUp as the place to collect the data, and the video does not treat AI answers as a channel. This guide adds that channel and a worked example.
What is share of voice, and what is the formula for each channel?
Share of voice is the portion of the attention in a defined market that goes to your brand rather than to the competitors you measure against. The formula is the same everywhere:
Share of voice = your metric ÷ (your metric + every competitor's metric) × 100
The term started in advertising. Nielsen still defines it as a brand's media spending expressed as a share of all media spending in the category, for one market and one channel (Nielsen, 2025). Digital channels swapped spend for whatever the channel counts.
| Channel | What you count | Formula | Free source of the number |
|---|---|---|---|
| Organic search | Estimated clicks across a fixed list of non-branded keywords | Your estimated clicks ÷ total estimated clicks × 100 | Manual rank checks plus keyword volumes |
| Social conversation | Posts and comments that mention each brand | Your mentions ÷ total mentions × 100 | Native search on each platform |
| Social publishing | Engagement on each brand's own posts | Your engagement ÷ total engagement × 100 | Counting likes, comments and shares by hand |
| AI answers | Brand mentions across a fixed set of prompts | Your mentions ÷ total brand mentions × 100 | Running the prompts by hand |
| Paid search | Impressions you received vs. impressions you were eligible for | Google Ads impression share | Your own Google Ads account |
Paid search is the odd one out. Google defines impression share as the impressions your ad received divided by the impressions it was eligible to receive (Google Ads Help), so the denominator is the auction, not a competitor set you chose.
How do you calculate share of voice? A worked example
Here is one hypothetical market: your brand and three competitors, measured on the same 30 days in three channels. All numbers are invented to show the arithmetic.
| Brand (hypothetical) | Search: est. clicks, 20 keywords | Search SOV (%) | Social: mentions | Social SOV (%) | AI: mentions in 40 answers | AI SOV (%) |
|---|---|---|---|---|---|---|
| You | 1,800 | 18 | 480 | 15 | 10 | 20 |
| Competitor B | 4,200 | 42 | 1,120 | 35 | 20 | 40 |
| Competitor C | 2,600 | 26 | 960 | 30 | 15 | 30 |
| Competitor D | 1,400 | 14 | 640 | 20 | 5 | 10 |
| Total | 10,000 | 100 | 3,200 | 100 | 50 | 100 |
Search. You rank for some of the 20 keywords, and your estimated clicks add up to 1,800 of the 10,000 the four brands share. 1,800 ÷ 10,000 × 100 = 18, so you hold 18 of every 100 estimated clicks.
Social. In the same 30 days, 480 of the 3,200 posts and comments that name any of the four brands name you. 480 ÷ 3,200 × 100 = 15.
AI answers. Twenty prompts run in two engines give 40 answers, and those answers name one of the four brands 50 times in total. You are named 10 times: 10 ÷ 50 × 100 = 20.
Two readings matter more than the numbers. First, you are relatively strongest in AI answers and weakest on social, so a single blended score would hide where to work. Second, Competitor B leads every channel, which makes it the benchmark to study, not the average.
There is a second AI number worth logging. Share of answers counts answers rather than mentions: you appear in 10 of 40 answers, so 25 of every 100 answers name you. Share of voice tells you how you compare with rivals inside answers. Share of answers tells you how often you are in the conversation at all. Our AI visibility tracking guide explains when to use each.
How to measure share of voice in search, social and AI answers
Each channel has a free method that costs time and a paid one that costs money. The free method is enough for one brand and a monthly report. The paid one earns its price when you track several brands or need weekly numbers.
Organic search
The free method: list 20 to 50 non-branded keywords, check each brand's position for each one, and turn positions into estimated clicks by multiplying the keyword's monthly volume by the click-through rate you expect at that position. The keyword volume checker gives you the volumes.
Google Search Console is free and accurate, but it shows only your own clicks and impressions, never a competitor's, so it cannot produce a share on its own.
The paid method is a rank tracker that does the multiplication. Ahrefs describes its Rank Tracker visibility metric as the percentage of all clicks from your tracked keywords that land on each website, and recommends leaving branded keywords out (Ahrefs).
Semrush includes Position Tracking from its entry SEO plan and lists a dedicated SEO share of voice report on its Advanced plan.
Current plan prices are in our Semrush pricing breakdown.
To see which keywords carry a competitor's traffic before you build the list, run its domain through the competitor keyword finder.
Social
Social has two numbers, and most reports mix them up. Conversation share counts what other people post about each brand. Publishing share counts the engagement each brand earns on its own posts.
For conversation share, the free method is native search on each platform for each brand name and handle over the same 30 days, counted in a sheet. It works for small brands and breaks once mentions pass a few hundred a month.
The paid method is a listening tool, which counts mentions across platforms and adds sentiment. Brandwatch is built for enterprise teams that need long archives.
Sprout Social bundles listening with publishing and reporting, which suits teams that already schedule posts there.
Brand24 is the fastest to set up for one brand.
Our social media monitoring tools guide compares the category.
For publishing share, count likes, comments and shares on each brand's last 30 posts per platform. Pengu Insights Starter, at $39 a month, benchmarks competitors' Instagram, TikTok and LinkedIn accounts by engagement rate, which is the same comparison without the counting.
Paid social has a proxy: the number of active ads each brand runs in the Meta Ad Library. Spend is not shown for most commercial ads, so active ads per brand is the closest public number.
Our Facebook Ad Library guide shows how to read it.
AI answers
The free method is a fixed prompt set: 20 questions your buyers ask, without your brand name, run once a month in ChatGPT and in Google, logging every brand named. It takes about 45 minutes.
The paid method is an AI visibility tracker, and several report this share against competitors directly. Otterly.AI Lite, for one, costs $29 a month for 15 prompts on the Otterly.AI pricing page, checked 1 November 2026. Pengu Insights Pro, at $69 a month, tracks ChatGPT and Google AI Overviews, and its Prompt Explorer asks ChatGPT, Claude, Gemini and Perplexity.
Our comparison of the best AI visibility tools lists engines, metered units and prices for nine trackers.
How we checked the tools and prices
We checked the tool descriptions and prices in this section on the vendors' own pages on 1 November 2026. Pengu Insights prices come from our own app on the same day. Semrush and Ahrefs plan prices are not quoted here because they change by plan and billing period; the linked pricing breakdowns carry them with a date.
How to measure SOV step by step
Plan an afternoon for the first run and about two hours for each monthly repeat.
- Freeze the competitor set (15 minutes). Pick three to five competitors you actually lose deals to and write them down. This list is the denominator, and changing it changes every number.
- Pick one metric per channel (10 minutes). Estimated clicks for search, mentions or engagement for social, brand mentions for AI answers. Write the definition next to the list.
- Build the keyword list and the prompt list (40 minutes). 20 to 50 non-branded keywords and 20 prompts. Leave out branded terms, or the brand with the most searches wins by default.
- Collect the search data (45 minutes). Positions for every brand on every keyword, converted to estimated clicks, or the share from your rank tracker.
- Collect the social data (30 minutes). Mentions per brand over the same 30 days, or engagement on each brand's posts. Not both in one number.
- Run the prompt set (45 minutes). One fresh chat per prompt, in ChatGPT and in Google, with every brand named logged in order.
- Calculate and log (15 minutes). One row per channel per month in a sheet, with the date and the list version.
- Repeat on the same day next month (2 hours). Same competitors, same lists. A change in the method is a new baseline, not a trend.
The normalization rules that keep this comparable are the same ones we use for any competitive benchmarking: same period, same definitions, same set.
What is a good share of voice?
A good share of voice is one above your market share. Nielsen's summary of the Binet and Field analysis of 171 campaigns from 1980 to 2010 found that market share typically gained 0.5% for every 10% of excess share of voice, with effects that vary by channel, category and brand size (Nielsen, 2025).
Excess share of voice (eSOV) is the gap between the two: your share minus your market share. Take the hypothetical brand above and suppose its market share were 12. Its search share of 18 would be 6 points of excess, a sign of room to grow, while its social share of 15 would be only 3 points above.
Fixed benchmarks such as "25 is strong" do not survive a change of competitor set. If you add a large competitor, everyone's share drops without anything happening in the market. Compare yourself with your own market share and with your own last month.
Common mistakes
- Changing the competitor set. Adding or dropping one brand moves every share. Keep the set for a year, and start a new baseline when you change it.
- Counting branded keywords in search. The brand with the most searches for its own name wins by default. Ahrefs recommends leaving branded keywords out for this reason.
- Mixing mentions and engagement. Conversation share and publishing share answer different questions. Report them as two numbers.
- Treating all mentions as good. A product recall raises your share. Read the number next to sentiment for social and AI answers.
- Calling one channel the whole picture. A search number describes search only. Ahrefs makes the same point: one channel does not equal overall share of voice or market share.
- Comparing AI numbers from different tools. Each tracker runs prompts differently, so the same set gives different results. Compare a tool with itself.
Frequently asked questions
What is share of voice in marketing?
The portion of the attention in a market that your brand holds compared with a set of competitors. Marketers measure it per channel: clicks in search, mentions or engagement on social, impression share in paid search, brand mentions in AI answers.
What does share of voice measure?
Relative visibility, not reach or sales. It shows how much of the clicks, mentions or answers in a defined space went to you rather than to your competitor set.
How do you calculate SOV?
Your number divided by the total for you plus your competitors, times 100. Keep the competitor set and the keyword or prompt list the same every month.
What is a good share of voice?
One above your market share. Nielsen's summary of the Binet and Field research links eSOV to market share growth over time; there is no universal fixed benchmark.
What is the difference between share of voice and market share?
Market share is your portion of sales. Share of voice is your portion of attention. The gap between them, eSOV, tends to show up before sales move.
What to do next
Write down your competitor set and your 20 keywords today. Those two lists are most of the work, and they turn every later number into a trend.
When you want the inputs without the counting, Pengu Insights collects competitor keywords and traffic, social engagement benchmarks and the Meta Ad Library on Starter, and adds ChatGPT and AI Overview tracking on Pro. Compare the plans on our pricing page.
Or start on the Free plan, with 80 credits at sign-up, in Pengu Insights and run your first competitor through the search and social side.
In short: one formula, one frozen competitor set, one metric per channel, and a monthly comparison with yourself.
Written by
Founder of Pengu Insights. Competitive intelligence practitioner for DTC brands and marketing agencies.
