Comms chips

Chips on Their Shoulders

Whenever our market research division, Heavy Reading, publishes a market perception survey, some of the firms that come out of it with egg on their faces elect to attack the methodology. Rather than examine the findings in detail to see whether and where their marketing plans might be falling down, they often claim that Heavy Reading must have surveyed morons.

I'm expecting this kind of backlash from the latest study - on communications chips – which we announced yesterday (see Survey Rates Chip Suppliers). As usual, some vendors come out of it better than others. But one of the most striking findings is that a lot of the respondents are remarkably unaware of the players in many segments of this marketplace.

Here are some examples:

  • 109 of the 449 respondents to our survey indicated that they were familiar with the market for Ethernet MAC chips. They were shown a list of 12 companies and asked which ones they recognized as suppliers. On average, they could identify only four of them.

  • 84 respondents said they were familiar with makers of Sonet/SDH data transceivers, but more than 20 percent of those respondents could not identify which of the 20 vendors in that segment offers the best prices.

So what does this mean? Were we indeed surveying morons?

In responding to the doubters, the first point to make is that this is a market perception survey, not a customer survey. For this survey, we set out to measure how people who work for equipment manufacturers view semiconductor suppliers in general. Vendors often fail to recognize this and get all hot under the collar because our results don't tally with feedback from their customers. Well, most of the respondents aren't their customers.

The second point is that we do screen the people who take Heavy Reading surveys. First, we send invitations only to those potential respondents who are directly relevant to the subject at hand. In the case of the communications chips survey, we sent invitations only to employees of systems vendors, original equipment manufacturers, and systems integrators. And we screen responses as they come in to weed out folks who don't fit the bill: For the comm chips survey, 945 people volunteered to answer questions; only 449 made the cut. Those 449 represent more than 200 equipment makers, OEMs, and integrators, with more than 40 percent of the base coming from the 25 biggest equipment makers in the world.

The third point is that having such a large number of responses makes the survey much more valuable. It means the results are more dependable. It also means results in most cases can be sliced and diced in all sorts of ways to provide different perspectives, without undermining the validity of the survey because the samples get too small. Having 449 responses makes it possible to see whether vendors are viewed differently in North America than they are in Europe, for instance. It's also possible to see whether corporate managers hold different views from engineering or sales and marketing staff.

The flip side of having a large number of responses is that it's impractical to check the detailed credentials of every survey respondent. In fact, it's pretty difficult to do that anyhow, unless the sample is very small - and then you run into the issue of whether it's valid to draw generalized conclusions from small sample sizes.

The real reason why our respondents failed to recognize a lot of the players in the marketplace probably comes down to a couple of issues.

First, nobody has ever conducted such a comprehensive survey as the one just published by Heavy Reading, so it's been tough to get an overall picture of who does what.

Second, the comm chips market is complicated. Each product category (and there are 31 in the report) constitutes a separate market. Although the big vendors play in many (and in some cases most) of these market segments, they're leaders in some and laggards in others.

Some examples of this are conspicuous in the report. In Ethernet access chips, for instance, Broadcom Corp. (Nasdaq: BRCM) is seen as the market leader, while players such as Agere Systems Inc. (NYSE: AGR.A), PMC-Sierra Inc. (Nasdaq: PMCS), and Vitesse Semiconductor Corp. (Nasdaq: VTSS) are viewed as also-rans. The opposite is the case in the Sonet/SDH transport chip market, in which PMC-Sierra, Agere, and Vitesse are top dogs, and Broadcom is a turkey.

So what's the bottom line? Might I respectfully suggest that the purchasers of comm chips have as much to learn from this report as the chip vendors themselves?

Respondents not only named the suppliers they recognized but also said who they considered to be the market leaders in terms of price, product performance, product quality and reliability, and service and support. With feedback from hundreds of their peers on these aspects, buyers might be encouraged to start shopping around for better deals than the ones they've got right now.

Perhaps that prospect might encourage chip vendors to take a closer interest in how their prospective customers view their products.

— Peter Heywood, Founding Editor, Light Reading

Heavy Reading's 99-page 2004 Communications Chips Market Perception Study costs $3,750 - which includes access to a searchable online database for slicing and dicing results by demographic, geography, and organization type. For more details, click here.

PO 12/5/2012 | 2:39:58 AM
re: Chips on Their Shoulders I appreciated your attempt to preemptively address many of the common attacks on such perception reports as this one; however, I have one comment I would like to reiterate.

It would be statistically useful to include a "self-rating" of familiarity ("How familiar are you with X"?) rather than just a "binary" selection ("Are you familiar with X?"). Then, a statistical test could be performed against that response: do those who indicate more familiarity reflect that in their other answers?

A control group (i.e. a set of respondents who are NOT familiar with X) should also be included in the overall analysis, to determine whether familiarity with the industry does correlate with familiarity with the players in that industry.

With such an enhanced methodology, much stronger conclusions are available, that even the "most familiar" individuals display only limited knowledge. Or, perhaps that actual familiarity is very narrowly concentrated in a few individuals. Or whatever else the data may indicate.

This should also help demonstrate that the true target of the survey (i.e. knowledgable insiders) have indeed been surveyed.

While the conclusions being drawn appear to be valid, it seems that for little additional effort conclusions regarding the correlation of familiarity could be made would be much more valuable.
fw23 12/5/2012 | 2:39:45 AM
re: Chips on Their Shoulders
I'm glad you guys were pro-active this time
before all the IDIOTS who don't understand
Science and research start coming out of the
walls to pump up their companies.

The whole chip industry needs to do a whole
lot of soul-searching as a result of this
tremendously enlightening report. I watched
with dismay at the NEGATIVE reaction by
companies like juniper and their paid stooges
the last time HEAVY READING brought some
LIGHT to BEAR on the dark corners of our crazy

So lets focus on real solutions this time
rather than trying to blame and kill the
messenger like last time.

The first obvious thing is that chip companies
need to get serious about branding and image.
The days of build-it-and-they-will-come went
the way of Kevin Costner's movie career years
ago. You need to get out in front of the
public if your one of these companies. And
the best way to do that is television.

Yes, the nay-sayers in the peanut gallery
will say its too expensive, but if people
selling mattresses and waffel irons can
buy TV time, so can people building
electronic do-dads.

So Bravo for taking the time to explain the
SCIENCE behind this survey to the complainers.

Sign In