Competitor Benchmarking Without a Big Team
Competitor benchmarking is putting the same measured numbers side by side for your brand and two or three rivals, so the gap between you is a figure rather than a feeling. It is the numbers layer of competitor analysis: it tells you how far behind or ahead you are on each measure, and it leaves the why to the analysis that sits on top.
You do not need a team to do it. You need two or three rivals, seven rows, an afternoon, and the discipline to write down where each number came from. Everything in those rows is public. The reason most small teams never benchmark is not that the data is hidden. It is that nobody told them which rows matter and which ones are estimates dressed as counts.
What is competitor benchmarking?
Competitor benchmarking is the measured comparison of your brand against chosen rivals on a fixed set of metrics, taken on a stated date, from sources you can name. The output is a grid of numbers with a source under every cell. Competitive benchmarking means the same thing, and the two terms get used interchangeably.
What makes it benchmarking rather than a list of stats is the word same. The same metrics, for every brand, from the same source, on the same day. A rival’s traffic figure from one tool next to your own from another is not a benchmark, because the two tools model traffic differently and the gap between them is partly the gap between the tools. Fix the instrument first, then read the numbers.
What is the difference between benchmarking and competitor analysis?
Benchmarking measures the gap. Analysis explains it and decides what to do about it. A benchmark that says a rival has ten times your referring domains is a fact. Working out whether that is because they have been publishing for a decade, or because they bought links, or because they are in the news, is analysis.
You need both, in that order. Analysis without a benchmark is a story with no numbers in it, which is most competitor slides. A benchmark without analysis is a dashboard, which is the thing that leaves the thinking to you. The definition of the wider exercise, and the types it comes in, are in what competitor analysis is and the types. This article is only about the numbers.
Which metrics should you benchmark?
Seven rows: organic keywords, referring domains, estimated organic traffic, share of voice, Google AI Overview mentions, ChatGPT mentions, and whether each brand is advertising right now. Every one of them is available for any brand from public data, and the table says where each comes from and how much to trust it.
| Row | What it tells you | Where it comes from | Count or estimate |
|---|---|---|---|
| Organic keywords | How many search terms each brand is found for, and which | Any keyword research tool’s domain view, free tiers included | Count, at the time of the pull |
| Referring domains | How many different sites link to each brand, which predicts how easily its new pages rank | Any backlink tool’s domain view | Count |
| Organic traffic a month | How many visits search sends each brand | The same tools | Estimate, always. Modelled from rankings, never counted |
| Share of voice | Each brand’s slice of the visibility across the set you chose | Derived from the rows above, across every brand in the set | Derived, and only as honest as the set |
| Google AI Overview mentions | How often a brand is named in Google’s AI answers in your market | Counted by hand from a fixed question list, or from a tool that tracks it | Count, at the time of the pull |
| ChatGPT mentions | How often a brand is named in assistant answers | The same method. Note which market the answers were measured in | Count, market-specific |
| Live advertising | Whether each brand is running ads now, and what they say | The public ad libraries the platforms publish | Observation, not a number |
The column to respect is the last one. Traffic is the row people quote first and it is the only pure estimate in the grid. If you quote a rival’s monthly visits as a fact in a meeting, you have quoted a model’s output as a count, and anyone with the same tool can show you a different number.
Two rows people want that are not in the table. Price benchmarking belongs only where you can source the price, which for most services and many products you cannot. I come back to that below. And paid search benchmarking, what a rival is bidding on, is partly visible through the ad libraries and partly a modelled guess, so label it accordingly if you add it.
How do you benchmark with no team and no tool budget?
Pick two or three rivals, open a spreadsheet with the seven rows above down the side, and fill one column per brand from free tiers and public sources in a single sitting. Write the source and the date next to every number as you go, not afterwards.
The single sitting matters. A benchmark assembled over three weeks is three different snapshots stitched together, and the gaps in it are partly time. Block an afternoon, pull every row for every brand on the same day, and you have an instrument you can repeat.
Most keyword and backlink tools have a free tier that allows a few domain lookups a day, which is enough for three rivals and yourself. The AI rows you can count by hand: write ten to twenty questions a buyer in your category would ask, put each one to the assistants, and tick which brands are named. The method for that is in how to run an AI visibility audit in an afternoon. The advertising row is a visit to each platform’s public ad library with the brand name typed in.
Choosing the rivals is the step that decides whether the rest is worth doing. The rule is the same as for any competitor work: the two or three names your buyer actually weighs you against, not the biggest names in the sector. I have set that out in how to run a competitor analysis before a pitch.
How do you read a benchmark honestly?
Read the counts as counts, the estimates as estimates, and the shares as a function of the set you chose. Then look for the row where the overall leader loses, because that is usually the finding that changes what you do.
A report my tool produced on 6 August 2026 makes the point. It benchmarked Maggi against Indomie and Nongshim in Malaysia. Maggi led on every row: 781 referring domains against 24 and 34, about 2,000 organic keywords against 44 and 65, an estimated 122K visits a month against 4.4K and 4K, and 80.5% share of voice against 8.1% and 11.4%. A benchmark read quickly says Maggi wins and there is nothing to do.
Read slowly, it says two other things. The share of voice figure came with a caveat the report stated itself: it was measured across each brand’s whole site, not only this category, so it reads as relative search presence and not as market share. And underneath the overall lead, the keyword row showed that Maggi’s terms were recipe and ingredient searches, while both rivals owned the noodle-purchase terms, the ones typed by someone about to buy. The leader on every row was absent from the purchase moment. That is the finding, and a benchmark that stops at the totals never surfaces it. The report is AI created, and that reading is my own judgement applied to its numbers, which is how every one of them should be used.
Should you benchmark prices?
Only when you can source the price for every brand in the set, and you should write down the source. Where you cannot, leave the row empty and say so, because a price gap built on guesses produces the most confident and most wrong recommendations in the whole exercise.
The same 6 August report listed, among the things it was deliberately not recommending, a price promotion to match the rivals’ marketplace placements. Its reason was that no sourced pricing or margin data existed for any of the three brands in that market, so nothing in the evidence showed that the leader competed on price. An empty row with a reason is more useful than a filled row with a guess, because the empty row tells you what you still need to find out.
How often should you repeat it?
Monthly, on roughly the same date, with the same rivals, the same rows and the same sources. One reading is a snapshot. Two a month apart are the first thing that looks like a direction, and repeating it is what turns benchmarking into competitor monitoring.
Keep every reading rather than overwriting it. The interesting question in six months will not be where you are but what moved and when, and you cannot reconstruct that from the latest grid. I hold myself to this in public: I measured my own AI visibility at zero on 7 August 2026 and committed to re-measuring it monthly with the same searches, whatever the number does. The instrument has to stay fixed or the trend means nothing.
The faster way to build the grid
The faster way is to let a focused tool pull every row for every brand on the same day, label each number as a count or an estimate, and hand you the grid with the reading already started. By hand the grid takes an afternoon. The reading is the part worth your time, and it is the part that needs you.
Vantage benchmarks your site against up to three competitors in about thirty minutes, including the two AI answer rows and a check on who is advertising, and it labels every figure as measured or estimated. The findings are AI created, so apply your own judgement before you act on them, which the labels are there to help you do. The first analysis is free, which is enough to see your own grid before you decide whether to keep running it.
Fifteen years in growth: VP Digital Lead at Vodafone Idea, digital lead for Nestlé Indonesia at dentsu, and nine years at Performics on brands like Citibank and Taj Hotels. Now co-founder at ShopLoco and Adroit Digital. He writes here and builds tools in the Lab.