Typically, you will find that revenue per employee in a given industry tends to be fairly similar.  For instance, here’s a sample of some of the bigger enterprise software companies:

There’s certainly some variation, and actually this sample has a fairly high standard deviation of about $48k per employee due to the small sample size (and due to Oracle).  So I don’t want to put this out there as a science.  However, estimating revenue for smaller, private companies that do not disclose financial figures using this type of analysis is certainly better than nothing.  For a professional market sizing exercise, you should try to get a bigger sample – ideally at least 30 companies with known revenue per employee figures.  From there, you can use the average (or a weighted average) to estimate revenue for smaller competitors.

There are some caveats to consider, as always.  This approach does not work well for early stage companies funded by outside investors, such as venture-funded startups.  These firms can afford to build up their organizations prior to achieving the revenue and cash flow required of bigger companies, and in extreme cases they might not have much revenue at all.  More generally, you might need to discount the revenue per employee average for smaller companies (i.e., under 100 employees), since they may not be as efficient at generating revenue as some of their larger competitors.  Finally, be very careful when applying these kinds of figures to conglomerates.  General Electric’s finance division will have very different dynamics than their jet engine division.  If your industry includes conglomerates, you may want to pursue this analysis based on individual business units rather than on entire companies.

For big public companies, employee numbers should be easy to get.  They are always included in 10-K and 10-Q filings, and they will often be listed somewhere on the company’s site as well.  For a quick and dirty version, you can often get relatively recent employee figures from Wikipedia.  The picture becomes murkier for private companies.  You should check the company site and so some quick web searches.  There are also a few other standard sources to consider:

• LinkedIn – LinkedIn now provides company pages which often list employee numbers (although these can be overstated for small companies seeking to look more established).  Searching on the company name and counting people results returned works even if there is no company page.  The number of employees who show up LinkedIn searches is of course never the full number, and you’ll need to multiply out based on the industry to get an estimate of company employees.  So a management consulting company with 30 LinkedIn members might have 40 employees, while a small manufacturer with 30 profiles might actually have 300 employees.  Nevertheless, it’s one more data point to consider.
• Hoovers – Each company profile includes an estimate of employees.  Note that their employee numbers are sometimes way off.
• Also consider looking at individual resumes, if you have access to job sites like Monster or CareerBuilder.  People often list the number of subordinates they have managed, and this information can help you arrive at an overall estimate.

The truly old-school way of estimating employees used to be actually driving out to one or more of the company’s locations and counting the number of cars in the parking lot.  Management consultants used to end up spending a good bit of time on seemingly wasteful tasks like these.  Obviously, even basic information becomes a lot more valuable when it’s scarce.  These days, the contemporary analogue is satellite photos, available on any of the major mapping services like Google Maps.  Yet another example, if only a minor one, of how much easier technology has made our lives.

First, let’s clearly define the market.  We’ll include traditional games like chess and checkers, mainstream games (e.g., Monopoly), and specialty games.  Role-playing games will be excluded.  Traditional playing cards will also be excluded, as will newer card games like Magic.  To start with, we’ll look at the US in depth.  An international estimate will be harder since it would be difficult to gauge cultural attitudes towards board games and pricing differences.

Public data

Let’s see what numbers are already available on the web (US market unless otherwise noted), starting with a quick search on “board games” “market size”.

• Various figures attributed to NPD Group, including $802 million in 2006 and $870 million in 2002
  • Sources mention that Hasbro has 51% of the market.  SEC filings for Hasbro and Mattel will also be important sources to look into.
• According to the Toy Industry Association, the “games/puzzles” category totaled $2.4 billion in 2009
  • This seems a bit high compared to the specific board game numbers since puzzles and card games are unlikely to make up 2/3 of this category.

As you can see, published figures tend to vary even among reputable sources, but $800 million will be a good sanity check against our analysis of the US market.

Top down sizing

To start with, let’s think about basic customer segments.  The biggest would be families with kids, and the other important one would be adults (think of having friends over to play Pictionary).  Note that households are probably the best unit of analysis here rather than individuals because most people buy board games for the household.  Based on Census data, there are 112.4 million households in the US, of which 34.9 million have children.

From there, my quick back-of-the-envelope sizing is going to rely heavily on some assumptions.  For a real analysis, you would want to verify these with consumer research, talking to people in the industry, or simply talking to your friends to get a minimal sample.

You can see my sizing spreadsheet here, and I’ll go through the overall approach below:

• Start with households.
• Break into families and non-families because they will have different purchasing habits.
  • To be more sophisticated, you could also break non-families into casual buyers and game enthusiasts, who would purchase much more.
• Narrow down to the percent of families that would own games.  Some households simply won’t have any due to low income or other pastimes.
• Look at games owned and game lifetime – this is a way to back into annual game purchases.  Simply assuming a number of games purchased per year is more prone to being wildly off.
• Based on those figures, get annual games purchased.
• I estimated cost per unit pretty loosely based on looking up game prices on Amazon in each category.  In reality, there would be a range of prices to consider, especially for games like chess and backgammon where some sets can be very expensive, but for simplicity I ignored that aspect.
• Multiply annual purchases by average price to get total annual revenue.

This approach comes out to $930 million or so, which is roughly in line with the published figures.  Not bad, although I want to reiterate that a market sizing done for a business plan or consulting project needs much more verification than what I’ve gone through.

Bottom up sizing

Ok, let’s take a look at a bottom up sizing now.  We’ll start listing significant firms in the industry and see what we can make of overall market revenue based on those.

This is where you can see the potential pain of bottom up sizing comes in.  If you’re doing it using competitors, you have to find all the relevant companies, break out product line and geographic revenue appropriately, and then determine how much of the market may be unaccounted for based on entities you missed.

In this case, I just did a short version, where I estimated US (technically North America) revenue for Hasbro, Mattel, Ravensburger, and a couple of other companies.  There are many other board game companies out there, perhaps hundreds, although I’ve only listed a few.  As a result, I applied a 30% fudge factor for the rest of the market.  If I were being more diligent, I would want to explicitly account for each company with some kind of estimate and try to avoid a fudge factor of over about 10%.

As it turns out, the bottom up sizing comes out to $870 million – at the manufacturer.  Then we need to add in retail channel margin to make this estimate match up with the top-down one, which is based on consumer spending.  With a retail markup of 50%, the total estimate comes out to $1,305.  The markup is another area where you would want to do more research in a real market sizing.  Some sales would be direct from the manufacturer with little to no markup, some channels would have negotiated different markups, and so on.

The bottom up estimate ends up being significantly higher than the top-down one.  You should go into any market sizing exercise expecting your numbers not to match up correctly.  It’s almost inevitable that you will have to work through some issues with how different sources define the market, the accuracy of your top down estimates, and so on.  So don’t be shocked if the numbers don’t work out at first – it happens to everyone.  In this case, the biggest culprits are probably incorrect assumptions about purchasing frequency and neglecting to include additional market segments (maybe schools are significant board game buyers, for example).

Also be wary of fiddling with your assumptions based on trying to make the numbers come out looking “right.”  My view is that everyone does it to some extent, either intentionally or unintentionally, but be skeptical in deciding whether your numbers pass your own smell test.

Overall estimate

Averaging the two different approaches, I come up with an overall market size figure of $1,116 million for the US board game market.  This result is higher than published sources, but I’m inclined to accept it as a first cut.  Note that if you were much more confident in one analysis than the other, you would weight that result more heavily in coming to an overall estimate.

Next steps

If I were doing this analysis for work purposes, I would try harder to size different segments (e.g., families, adult casual, adult enthusiast, schools, etc.), both for greater accuracy and because those segments would be more relevant to business strategy than the overall market.  It’s likely that no product would play equally well in each segment, so you would want to think about positioning, distribution channels, promotion, and so on separately for each segment.  For a simple family game, you might ignore the adult enthusiast segment entirely because they would be looking for something different.  To beat a dead horse, you would also want to validate all of these crazy assumptions with actual research.  Of course, a real market sizing would often take several weeks instead of an afternoon.

Is this a helpful example?  What would you do differently if you were sizing the market?

US Census – Statistics of US Businesses

The Statistics of US Businesses is a fairly exhaustive look at firms, locations, employment, and payroll by industry.  The data tend to be released with a 3-4 year lag (e.g., 2006 data became available in 2009), but for more industries this is not a big problem since the data will not have changed too drastically in that time.

NAICS (North American Industry Classification System) codes are used to categorize firms by industry.  The codes can be a bit confusing, but there is a reference.  The NAICS codes form a hierarchy, so for example Professional, Scientific, and Technical Services (54) includes Management Consulting Services (5416) and Scientific Research and Development Services (5417) among others, with more detailed levels below those.

One challenge using this data is that NAICS codes may not correspond directly to the industry you’re analyzing, in which case you will need to do some creative slicing and dicing.

There are also great geographical breakdowns available by state, metropolitan statistical area (MSA), or county.

While there is no time series data directly available, you can construct your own time series from the historical data.

US Census – Population

Everyone is familiar with the herculean efforts of the US Census to measure the population every 10 years.  This process results in detailed demographic information that can be quite useful for business analysis.  In addition, the Census provides annual estimates based on smaller population samples every year to bridge the gaps between the more accurate official counts every ten years.

There are a multitude of different ways to access this information, which can actually be quite confusing.  The best place to start is by looking at breakdowns by specific topics, which include variables like age, education, income, and occupation.

Bureau of Labor Statistics

The BLS, which is part of the Department of Labor, is the agency in charge of providing information on labor economics and statistics.  As the description suggests, the agency is a favorite of economists and public policy analysts, but it can also be a valuable market sizing resource, especially for consumer spending and time use.

The Consumer Expenditure Survey is the premier BLS resource in my opinion.  It tracks consumer spending by income distribution, other demographic information, and spending category.  It’s great information, but I must admit that formulating the right queries can be a chore.  I’m not an expert, but I’ll provide some examples in a later post.

The American Time Use Survey can also be useful for thinking about how people divide their time among various activities.  Unfortunately, the data is available either in summary PDF tables or in raw data files which work best with statistical software like SPSS or SAS, so it can be a bit difficult to manipulate as desired.

These sources offer a huge amount of data but as a result can be a bit painful to use at first.  I’ll discuss some more detailed examples shortly, but in the meantime, I’d be interested in your comments as to which of these sources you find useful and how you prefer to access them.

Like forecasts, market size estimates should most often be considered directional (consulting speak for “in the right ballpark”).  You should also keep this point in mind when consuming market size information – the market size numbers you read in the news or market research reports are probably only accurate within 10-20% in many cases.  One of the best ways to ensure that you get at least a sensible result is to triangulate between different approaches: top-down and bottom-up estimates.

Top-down sizing

As the name suggests, you start from the top here: high-level numbers like the US population (about 305 million currently), the number of businesses in the country (about 6 million),  healthcare spending (perhaps $2.5 trillion for 2009), or the number of management consultants in the US (estimates vary, but 500,000 is a reasonable number).

Then you work down from that overall number to figure out what’s relevant to your company.  Let’s say you want to launch a premium online job board like TheLadders.com, where job seekers pay a $50 membership fee to post their resume and search for openings.  Starting with the total population, the table below shows how you might drill down to an actual market size in terms of revenue.  Note that I just used assumptions for most of these numbers.

Sample Top-Down Market Sizing

The hardest part is usually finding the data to populate calculations like these.  How do you figure out what percentage of the job seeker market is willing to pay for a subscription site?  Well, the ideal approach would be a big consumer phone survey, but that type of research is expensive enough that it’s typically restricted to Fortune 500 budgets.  If you’re not in that league, you could do some secondary research on companies like TheLadders.com and see if any of it mentions those types of statistics.  Firms tend to keep those close to the vest, so you’re unlikely to find anything, especially if they’re smaller companies and/or privately owned.  You might have slightly better luck with sources on publicly traded companies, like say Monster.com’s investor conference calls.  You would probably have more success calling people in the online jobs industry and asking them – primary research (called primary because you’re going straight to the source of the information, not to a document that someone else prepared like a market research report). I’ll go into primary research in more detail in later posts.  If that’s not up your alley, the last resort is to “estimate,” or guess.  Most consulting firms go the primary research route (or create some kind of crazy Excel model that bakes in all sorts of assumptions and then gets footnoted as “ConsultingCo analysis”), while many startups use estimates and then try to validate those in the market once they launch.  If you use customer development, that’s a form of primary research.

There are also lots of packaged market research reports out there from various sources.  Be cautious both in purchasing and in using these reports.  They’re often quite expensive, their estimates aren’t necessarily reliable, and it’s amazing how often the  report that looks perfect for you turns out to define the market slightly differently than you do, often making it useless.  I’ve seen clients buy $4,000 reports and get absolutely no value out of them more than a few times.  These same caveats go for the various IT research firms like Gartner, IDC, and so on.  Now that I’ve warned you to take all of these reports with a big grain of salt, I should mention that FirstResearch.com has some more moderately priced reports.  Even better, the free previews show you a bit of their market size results.  I can’t vouch for accuracy, of course.  MarketResearch.com also aggregates a broad range of reports in one place, but these tend to be more expensive.

Bottom-up sizing

This approach works by looking at individual customers or competitors and rolling them up to get the overall size of the market.

For a great example of building up an estimate from individual customers, check out this presentation by Aaron Patzer from Mint starting at about the 12 minute mark.  It’s based on a lead generation model (where financial companies like credit card providers pay per customer that Mint sends them), so it’s slightly more complex than a straightforward consumer or B2B market.  Watch it twice!

Rolling up the revenue from individual companies to look at the overall market can be a great approach when there is a relatively small number of public companies in the market.  For the US car market, you could pretty easily look up US revenue in the filings of GM, Ford, Toyota, and a handful of others.  Add those up, round up by a few percent for smaller companies you didn’t look at directly (Tesla, Lamborghini, etc.), and you have a rough market size.

In smaller and more fragmented markets, adding up individual competitors quickly becomes painful, but it can still be your best option.  I once sized an obscure category of diagnostic supplies by individually estimating revenue for over 50 companies and then totaling them up for the overall market.  I’m sure the resulting number was not accurate to within a million dollars, but it was good enough for our purposes.  The added benefit is that this type of estimate can be easier to defend to clients or investors because it relies on a broader, more easily confirmed set of data points.  Bottom-up estimates can still be gamed by fiddling with the individual components, however.

The addressable market

Even when doing a bottom-up analysis, you have to be careful not to overstate the boundaries of the relevant market.  Usually, you won’t be looking at the overall car market or anything else that monolithic because no product competes across an entire industry.  For instance, you might be launching a new sub-compact.  Then you’ll need to cut down your market size based on the percentage of sales that are sub-compacts.  Or you might be opening one new dealership, meaning that you have to take into account the geographic radius from which you’ll get customers.  The same geographic cut would apply to any retail business.  Even for some business to business examples, your product will become less competitive past a certain distance because of shipping costs.  For heavy, low-value products like paper or cement, your addressable market may only be within a couple of hundred miles of your plant.  For products like pharmaceuticals or software, those considerations don’t come into play.  More generally, there are a variety of other factors to take into account in defining your addressable market, from regulatory restrictions to distribution channels and customer segments.

The bottom line here is that you need to be a little critical of your own estimate and make sure that you’re sizing the market that you or your client can actually address, not a general market that looks impressive but isn’t really relevant.  Don’t be like all those dot-com entrepreneurs who said, “The XYZ market is $100 billion globally, and even if we only get 1% market share…”

Triangulating

As you can tell by now, both top-down and bottom-up approaches involve a good bit of uncertainty.  Time and data permitting, you should triangulate among a top-down analysis, a roll-up of individual customer spending, and a roll-up of competitor revenue.  In an ideal world, these numbers should all be in the same ballpark, but if there is a discrepancy, you can troubleshoot your analysis or do a weighted average of the estimates based on your most reliable information.

The resulting number won’t be perfect, but you’ll have a good analysis to walk clients or investors through to get their buy-in.

Market trends and growth rates

For many mature industries, markets are growing at around the same rate as the overall economy, so thinking about market growth rates may not be particularly important.  Industries in growth or decline stages are very different.  Growing markets are much more attractive, size being equal, because brand new customers are entering the market.  It’s much easier to acquire a customer who’s not already buying from your competitor.

The extreme case here is new markets that don’t currently exist – obviously difficult to size.  In these markets, just try to make good assumptions, consider various scenarios for your key assumptions, and put a plan in place to test those assumptions as soon as possible.

Also consider whether your company will be generating revenue differently than current industry incumbents.  For example, the office applications market dominated by Microsoft Office might be a $10 billion market, but if you’re coming in at 1/10 the price of Office, you should really consider it a $1 billion market when developing your growth plans.  Disrupting an existing market is a great competitive advantage, but it also shrinks the available pie.  Another great example is Craigslist, which is turning the formerly $30 billion classified advertising market into a $100 million market.

Projecting market growth is another quixotic exercise that I’ll cover in another post.

Related posts

I’ll be working on a series on market sizing.  Here are a few of the other posts you might want to check out.

• Information Sources for Market Sizing (covers the US Census and Bureau of Labor Statistics, two great data sources)
• A Heavy-Duty Market Sizing Example (goes through a simplified version of a full market sizing)
• Using Employee Numbers to Estimate Revenue (discusses another approach to estimating revenue for companies that don’t disclose it)

Other resources

Finally, here are a couple of other good reads on market sizing.  Let me know if you have others to recommend.

• Market Sizing Comments from a VC Perspective
• The Definitive Guide to iPhone App Market Sizing
• Startups and Angels: Bottom Up Market Sizing
• Investor Pitch: Bottoms Up Market Sizing

This post has turned out to be rather long, but I hope you find it helpful.  I would be very interested to hear what aspects you would like more information on, or what you do differently.