WeDiscover - homeTalk to us
Guides · 12 Jun 2023 ·

Branded Share of Search Reporting

Abstract cover: one ring around the WeDiscover O split into five segments of different sizes.

If you are familiar with marketing data, it is quite likely you will have heard of the term Share of Voice (SoV).

Although similar, this project focuses on something - termed by Les Binet - known as Share of Search (SoS). Mathematically speaking the distinction between these two terms can be summarised as per the following:

SoS=number of searches for your brandnumber of searches for competitors in your categorySoV=number of brand mentionstotal number of brand mentions for you & your competitors\begin{aligned}\textbf{SoS} &= \dfrac{\text{number of searches for your brand}}{\text{number of searches for competitors in your category}}\\[1.4em]\textbf{SoV} &= \dfrac{\text{number of brand mentions}}{\text{total number of brand mentions for you \& your competitors}}\end{aligned}

Research conducted by Les Binet & Co has shown that SoS is a good indicator of how market share will be distributed across competitors. The below is an example of the mobile phone industry, focusing on the brand LG:

Line chart of LG's 12-month share of searches and share of market, 2004 to 2013: searches peak at 8.8% in late 2008, market at 10.9% in early 2009, and both fall to about 1% by 2012.
Share of volume vs share of search: LG.

SoS shows how your brand is performing relative to competitors

In digital marketing, and specifically Paid Search, the ability to understand competition in a market is vital to curating the optimal marketing strategy.

When pairing SoS metrics with metrics available through platforms like Google Ads (e.g. Impression Share, Top of Page Rate) a complete view of the competitor landscape can be gained, as shown via the screenshot below of our custom-made Looker reporting for viewing auction insights.

Looker competitor auction insights dashboard: impression share by competitor over time and a table of auction metrics

SoS trends are indicative of business performance

Although there are instances where they can be incredibly valuable, in-platform metrics like Impression Share are often not particularly useful in isolation, especially when trying to understand the overall trend of a brand.

Looking at SoS over time enables you to understand how your business is trending, and to observe overall growth, relative to a defined set of competitors.

Here is an example from the sportswear industry, looking at market share of various similarly sized, defined competitors.

Line chart of sportswear share of search since 2004: Reebok falls from about 100% to 19%, while Gymshark rises from 2013 to 29% (peak 43%). Under Armour ends at 28%.
Share of search in sportswear: Gymshark and competitors.

Methodology

Theoretically, there are multiple approaches and sources of data we could have used in building a Share of Search report, including Google Analytics, Search Console & more.

However, we settled on a solution that leverages the following two tools:

  • Google Trends - Shows index of popularity of search queries on Google Search across various regions and languages.
  • Historic Keyword Planner Data (via Google Ads) - Enables you to see monthly search volumes for a certain keyword/search term over time.

Read on for our take on how we approached SoS reporting, or check out our video instead.

Google Trends

Understanding the Source

In Google Trends there are 2 distinct ways of searching for a brand - using Terms or Topics. Fundamentally, these can be summarised as follows:

  • Terms - Get matches for the terms in your chosen language. Depending on setup, this can be broken down into sub-categories as follows:
  • Simple search for the term - e.g. if you search John Lewis you can get matches for “John Lewis furniture”, “shop John Lewis” or even “Lewis John”.
  • Phrase search - By wrapping the search term in quotes “John Lewis” you limit the searches to phrases containing those two words in that order e.g. “John Lewis”, “shop John Lewis” etc - this works in a similar fashion to Exact match keywords.
  • Topics - a group of terms that share the same concept in any language.
    • For example, if you search London, and choose the corresponding topic, your search includes results for topics like “Capital of the UK” and “Londres,” which is “London” in Spanish.

In the context of SoS both of these can be used, though Phrase Searches are most appropriate in the majority of instances (exceptions are brands with broad names that can also mean something else e.g. Coral, Next, Thread).

Source Limitations

Although a powerful source, Google Trends has some limitations in relation to data usability & quality: because the data is indexed and sampled, understanding the total volume becomes almost impossible.

Actual search volume is of paramount importance for this project. Using Google’s Keyword Planner is an opportunity to un-index this data.

There are some considerations we need to note:

  1. As Google Trends data is sampled, what is the sampling error in this data?
  2. Is Historic Keyword Planner data an accurate representation of actual search volume? Does un-indexing the Google Trends in this nature provide an accurate representation?
Diagram of a sample drawn from a target population, beside a Google Trends line chart of indexed search interest

The LHS above shows an example of sampling a population, while the RHS shows an example of indexed data retrieved from Google Trends.

Data Validation: Google Trends Sampling Error

To understand the sampling error in Google Trends data, we set up an automated extraction from the Google Trends front end for 2 brands of differing sizes in terms of total search volume.

Using this data we then compared the lower & upper boundaries of the data relative to the average, to determine the overall sampling error. From this, it can be concluded that for brands with sufficient search volume - > 20,000 searches per month - the sampling error is minimal; however, for smaller brands the error incrementally increases. As such, this approach is not applicable for all brands.

This example shows interest for a low-volume brand with ~5,000 monthly searches, and its associated error.

Line chart of average weekly search interest (re-indexed) over weeks 2 to 40, with upper and lower bounds: it ranges from about 58 to a peak of 173 in week 16 and ends at 74.
Average weekly search interest with upper and lower bounds (re-indexed).

While the example below shows a higher-volume brand with ~300,000 monthly searches, and its smaller error bounds.

Line chart of average weekly search interest (re-indexed) over weeks 2 to 40, with tight bounds: it peaks at 446 in week 12 and 396 in week 15, and ends at 228.
Average weekly search interest with upper and lower bounds (re-indexed).

Keyword Planner

Before we can start to investigate whether or not we can use Keyword Planner to un-index Google Trends data, we first need to understand the data available.

As mentioned, Keyword Planner enables you to see monthly search volumes for a certain keyword/search term over time (up to last 4 years). The issue with this data is that it is bucketed i.e. Google does not provide the exact search volume for the term but instead a rounded figure for the searches that month. The size of these buckets is dependent on the size of the search query but generally the bucket increments are around 18% of the next increment’s value.

Fundamentally, this issue adds in-built error into the system that cannot be avoided. However, this data source enables us to at least get some kind of nominal understanding of the market volume for a brand/market.

To validate the accuracy of un-indexing Google Trends data against Keyword Planner volumes, our best point of reference is Search Console data for branded terms.

When we compared Keyword Planner volumes vs Search Console for two of our clients in a given month we observed the following:


Keyword Planner

Search Console

Difference

Brand One

5,400

5,900

9.3%

Brand Two

301,000

290,570

-3.5%

Coupling this with Google Trends data we then gained a weekly/monthly view of search interest (we avoided daily due to short-term volatility) for the different brands. Combining multiple brands within a market we then got a view of relative search volume (and market competitiveness).

Case Studies

We explore the SoS logic with two industry examples here.

Example 1: John Lewis - Established Brand

Comparing John Lewis to the following brands we consider its share of market. Note that the brands listed are acting as an example and are not an extensive view of the market - this list could be extended for future reference.

➡️ Currys | PC World

➡️ House of Fraser

➡️ Selfridges

➡️ Marks & Spencer

As shown in the chart, we see John Lewis retain a stable share of the market relative to the selected competitors since 2004 (the earliest date Google Trends data is available from).

Line chart of department-store share of search since 2004: John Lewis stays between 29% and 43%, ending at 33%. Currys ends at 29%, M&S 28%, Selfridges 6%, House of Fraser 4%.
Share of search among UK department stores: John Lewis and competitors.

Example 2: Gymshark - Recent Unicorn Brand

Comparing Gymshark to the brands below, we consider its share of market. Note that again competitors selected are used for example purposes and are those of a similar size to Gymshark. Larger brands, such as Nike, are not included, as their overall search volume dwarfs that of our core brand, and makes it difficult to identify insights.

➡️ Lululemon

➡️ BoohooMAN

➡️ Reebok

➡️ Under Armour

As shown in the chart, unlike John Lewis, Gymshark has seen a substantial rise in its SoS relative to its competitors.

Line chart of sportswear share of search since 2004: Reebok falls from about 100% to 19%, while Gymshark rises from 2013 to 29% (peak 43%). Under Armour ends at 28%.
Share of search in sportswear: Gymshark and competitors.

Comparing Share of Market

Comparing the share of the market across the two brands we see from the first example that the overall retail market has seen large stability (with the exception of House of Fraser seeing a substantial drop-off), whereas, in the Gym/Fitness market shown in example 2, older brands like Reebok have seen their SoS reduced whilst brands like Gymshark have seen stark increases.

John Lewis share of market over time:

Stacked area chart of share among John Lewis, M&S, Currys, House of Fraser and Selfridges since 2004, broadly stable

Gymshark share of market over time:

Stacked area chart of sportswear brand share since 2004, with Gymshark's share growing rapidly from 2015

Comparing Total Market Size

Comparing total market size we see stark differences in a more stable retail market versus a growing fitness market.

John Lewis & competitors’ total search volume over time:

Line chart of total retail search volume (thousands) since 2004 with bounds: December peaks rise from 49,000 to 128,000 in 2013, then fall back to about 54,000 by 2022.
Total retail market search volume, with upper and lower bounds.

Gymshark & competitors’ total search volume over time:

Line chart of gym and fitness search volume (thousands) since 2004 with bounds: about 300 to 800 until 2014, then rising to 1,500 to 3,000 from 2019, peaking at 3,800 in late 2020.
Total gym and fitness market search volume, with upper and lower bounds.

Summary

In marketing, one of the greatest challenges for brands looking to grow is understanding the competition within a market & where your company sits.

Share of Search acts as a point of reference for overall search performance and a leading indicator for more difficult to monitor measures like Share of Voice. As such, by monitoring and understanding this measure, businesses can determine the required aggressiveness of their search marketing to drive sustained business growth.

We’d love to hear your thoughts and experiences with share of search reporting - get in touch with us here.

Related services

Related posts

Partner with us

Tell us what you want your marketing to do next.

Talk to us

Join our team

Join specialists in media, data, engineering and creative, and learn from all of them.

See open roles