Market shares have many virtues. They size the market, allowing firms and investors to estimate market potential, both total addressable market and total serviceable market. They create categories in which vendors might be leaders and show where vendors are placed in comparison to the wider market. Finally, they provide market momentum for large vendors and for vendors in growing markets. We’ll come back in our next transmission on how vendors can use market sizing, forecast and shares. This time, we’re exploring the powers and limitations of market share methodologies.
Technology vendors, start-ups and scaleups all can weaponise market shares. We spoke to some industry analysts that are experts in market shares to unpack the process of creating them and the limitations of the method. Here’s what you must understand about them.
What is a market share?
In the most basic of terms, a market share is a report of the size of market and what it is made up of. Using industry taxonomies, analysts calculate the revenue of market categories. A market share report will show the relative size of a vendor in a given market and establish the key market players. They are based on market sizing which themselves underpin market forecasts.
Vendors should be aware of the market share timeline –this is not a quick process. Market shares are published annually or quarterly, depending on the firm and the market they represent. According to Karsten Weide / IDC typically publish market shares during April and October each year. For them, this means the process starts all over again in October to prepare for the next report. Martin Hingley / IT Candor publishes calendar quarterly market shares as “quarterly reports became popular in Europe, however it leads to the difficulty of trying to publish before the next quarter begins.” Rachel Lashford / Canalys has found that clients are increasingly requesting more regular market share reports, the result is that “we are now delivering monthly data to more and more clients for some sectors and countries, following their requests, but will only do so where we can apply a rigorous methodology to the process.”
The market share methodology.
The majority of quantitative industry analyst firms will produce some kind of market share report. There will be slight differences between each firm, however, the underlying method is similar across the board. Gartner analyst Fabrizio Biscotti posted 9 steps that they take to create a market share. We spoke to more analysts to understand the process further and have simplified them into 3 main steps.
But, keep in mind that understanding the method involves understanding its limitations. Beyond being criticised as a representation of the past, there are limits with market share precision, granularity and objectivity. Bendle and Bagga at MIT suggest abandoning them as a metric in most cases. However, as long as you understand the limits you can still extract value from them.
STEP ONE: Segment the market.
Analysts create and define market segments, which together form a taxonomy, to create categories within which they can assess vendor’s positions. Bo Lykkegaard / IDC, explained that they use their annual taxonomy to do this: “It’s almost like a phone book and it’s categorised as logically and intuitively as possible. At IDC these taxonomies have three layers: primary markets, which are the macro category, secondary markets which are semi-aggregated and functional markets, which is the lowest level and is where the details lie. In some markets, such as security software and data management, we are now adding a fourth layer: a sub-functional segmentation, which is even more detailed.”
This is a tricky process, as they must strike the balance between accuracy and representation. Christophe Chalons / PAC describes how “we try to compare apples with apples, the idea is to be precise but still far reaching”. Ideally, there should be enough granularity to make meaningful markets, without creating so many markets that it becomes overly complex. As technology advances rapidly and new, innovative ideas are realised regularly the market becomes more and more difficult to categorise. However, Bo emphasised that: “For a taxonomy to be a useful tool there has to be some level of continuity, otherwise historical data would need to be restated with every change to provide historic consistency, otherwise comparison over time becomes impossible. For example, taxonomy changes must be ‘rolled back’ in time to ensure historic apples-to-apples comparison and most clients want a view of organic growth rates adjusted for M&A activities over time.”
LIMITATION: With market shares, the more granular you go the less accurate it gets.
Granular segmentation by functional capability is harder when a software category has a strong representation of local vendors. Bo Lykkegaard explained how in a more technical software category, like operating systems, it’s a very global market. This is in contrast to other software categories, e.g. payroll accounting applications, where there are many local vendors that analysts may not even know exist. Therefore, they can miss out on market share inclusion.
Granularity within products can certainly make market shares less accurate, but going granular by geography can degrade their confidence intervals. A worldwide market share is typically the most accurate as the majority of companies are legally obligated to report on their revenue. Since most companies operate globally, very few have even calculated their revenue by geography themselves. Even less so would share it with an analyst for their research. As a result, analysts are forced to base estimates on assumptions and other vendor information, such as number of employees in a country, sales office locations, number of local VARs and services partners, local customer reference and qualitative discussions by the vendor on country coverage. The more detailed the geographical coverage, the higher the reliance on assumptions and related vendor information. At some point, market shares become derived and strongly assumptions-based. Despite this, Rachel Lashford / Canalys maintains that “Depending on the sector, country data can be very accurate.” Sources used at Canalys also extend to distributor sales-out data, trade data and from third-party research companies.
STEP TWO. Decide between a top-down and bottom-up approach.
The approach defines the direction the method takes. A top-down method begins with an estimate of the market size, using the market-level, macro data to estimate size of the overall market before estimating the size of vendors within it. A bottom-up approach begins by listing as many active vendors as possible who operate in that market, calculating their revenue and adding it together at the end. Researchers speak to as many vendors as possible and survey their channel partners and buyers to estimate vendor size. Bo Lykkegaard estimates that “in software market shares, we reach out to hundreds of software vendors each semi-annual research cycle.”
The approach an analyst takes may be impacted by the firm they work for or the industry they analyse. Karsten Weide / IDC, believes that “a bottom-up approach works with real companies and so it’s more accurate. It’s closer to the truth by definition.” However, Martin Hingley / IT Candor prefers a top-down approach as “you have to try your best to think of the bigger picture outside of the market share. If IBM sells a service to Oracle then it has never been part of the market and it shouldn’t be part of the market share. Unless I say otherwise, my market shares are an estimate of what users have spent within that market.” This Orders v Sales-in vs Sales-out conundrum has led to inflated market shares from a number of suppliers when working with weak taxonomies. A solid methodology should always specify exactly what they are counting.
LIMITATION: Market shares’ precision varies, particularly between technology types and companies.
It isn’t truly possible to calculate the exact size of a market and so, analysts create calculations to approximate some numbers. Martin Hingley explained that when calculating an install base you need an algorithm for replacement rate. “If it’s shipped in year 1 then in year 2, 100% will likely still be there, in year 3, 98%, and so on.” Similarly, Karsten Weide explained that for a lot of privately held companies there is no revenue data available publicly. “In that case, we base our revenue estimates on the number of employees and the life stage of the given company. Some estimates will be too high, some too low, but overall, these inaccuracies will cancel each other out so that you can get an accurate overall market sizing.”
A market share may not always be a precise reflection of your business because of how the numbers are calculated. According to Christophe Chalons, “Vendors will count big deals as the biggest part for their business. Analysts try to understand the big contracts to adjust the figures. We can’t just take the numbers we get – we use them to build our opinion based on a wider understanding of the market.” That means that big deals will be broken down by their technology types. As a result, your own numbers may differ from those reported on by analysts.
STEP THREE: Analysts reach an internal consensus.
Research teams spend hours as market shares are a rigorous process. The final result is calculated based on inquiries, briefings, customer references, general PR stories, customer stories, vendor communications and any publicly available revenue information. Armed with this external guidance, analysts group together to reach an internal consensus on the estimated size of the market.
LIMITATION: Market share methodology relies on analyst assumptions.
At IDC, Karsten Weide explained that once you have calculated the revenues of the market that you have identified, you must estimate how much spending there is on those vendors that you do not track. This estimation of other vendors, i.e. the long tail of the market, is based upon qualitative and quantitative assessment of known, untracked vendors, as well as factors such as prevalence of local software vendors and fragmentation of market.
At Gartner and Canalys, estimates are embedded in the methodology. Steps 4 to 6 according to Fabrizio Biscotti rely on analyst estimates. Canalys specify that their findings are market shares estimates. What’s more, market shares don’t even provide a complete overview of the technology market when all combined. Martin Hingley points out that “One problem is that you can miss the whole market, because the market may not have a market share. For example, the internet is always missed. It’s clear that not everything can be assessed through market shares.”
In summary, a market share is created by a team of analysts, based on rigorous market research to estimate the amount of revenue within a given market. Although they are based on numbers, always remember that they have their limits.
Market shares are estimates.
Market shares are not an objective truth, but they are useful for vendors, investors and clients alike. In the technology industry, research analysts create them based on months of background research and their own assumptions of the market. Investors use them to identify growing market segments and thriving vendors within them. Clients use them to understand who the market players are. Vendors use them to get an understanding of the wider technology landscape.
Armed with this understanding of the methodology, watch out for our next blog post on how vendors can leverage quantitative research from industry analysts to drive the 4 impacts of AR. You will discover how much market shares drive public opinion, how they influence technology buyers and investors, where vendors can leverage them for go-to-market tactics and most importantly why they’re essential to inform your company’s strategic direction.
Many thanks to our contributors for their time and providing their insights for this Starsight Transmission:
- Rachel Lashford (@RachelLashford | LinkedIn) / Canalys
- Martin Hingley (@mHingley | LinkedIn) / IT Candor
- Christophe Chalons (@CChalons | LinkedIn) / PAC
- Bo Lykkegaard, (@BoLykkegaard | LinkedIn) / IDC
- Karsten Weide (@KarstenW | LinkedIn) / IDC
Gartner declined to comment or brief us for this blog post.