Tracking webOS Use
May 13th, 2010, 6:15pm by JakeSo for a couple months I’ve been tracking the growth in webOS using statistics from StatCounter. While far from being a perfect metric, it does provide insight into overall trends. Since webOS first showed up in July, share of the mobile browser market in the U.S. has climbed from 0.5% to 1.4%. But that doesn’t tell the whole story, as the mobile market has grown tremendously since then. So I took it to the next level and created a spreadsheet to track usage. The in-flight WiFi is flaky with Google, so I can’t share the link or update the usage to include the first few weeks.
StatCounter also tracks overall mobile vs. desktop share, so I’ve incorporated that. Then factored in the overall upward growth in internet usage by assuming a constant growth rate, increasing ~8%/year. Multiply the three together, and there’s a reasonable estimate of webOS browser use. Now, there are some bumps & wiggles that are obviously not real. I have done a couple things to smooth the data: 1) used an exponentially weighted average w/ a factor of 0.05 for the mobile share; 2) ditto for the webOS share with a factor of 0.2. For any non-engineers that stumble upon this, the exponentially weighted moving average with a low factor (0.05) places less weight on the previous day (5%) and a large weight on the previous day’s average (95%). Lower number leads to smoother curve that’s less responsive.
The trends since mid-August are clear– a very slow rise for 8 months during which use climbed 50%. Then, another 50% rise in the last 5-6 weeks. That coincides with the Verizon marketing for the Pre Plus & Pixi Plus. As Kelvin theorized, those increased sales may be the reason for Palm’s latest earnings warning– Verizon is doubtful to have ordered any more Pre’s, but for each one they sell Palm is probably kicking in a $50-100 kickback. That would be reflected as a decrease in revenue for this quarter. And Palm timed the warning to coincide with the HP buyout to deflect possible criticism for selling “low.” So while sell-in may meet Palm’s earlier guidance, the revenue warning could in fact indicate that sell-through has picked up. Conversely, it could also mean that sell-in & sell-through continue to lag and StatCounter’s stats are bogus.
May 13th, 2010 at 11:13 pm
Verizon has been aggressively marketing Pre/Pixi Plus since Feb 1, with more primetime spots than Palm has ever seen. Palm has been actively working to educate Verizon staff since their warning late Feb (Project Jumpstart). If your analysis is to be believed, none of these initiatives have had any effect whatsoever. What does seem to have an effect was the huge price drop ($150 -> $50) and the inclusion of the mobile hotspot, as of 4/1 (I think).
Regardless, it looks like your analysis indicates that total installed base has increased 70% since the end of Q3 (about 1.4 million webOS users at the time). If Palm really does sell-through (to end users) about a million phones, they would be completely out from the inventory overhang that got them in this mess.
Of course, they just dug themselves into the same hole with an AT&T launch at the original Verizon pricing. What are they thinking?
May 19th, 2010 at 11:09 pm
So the recent move continues… we’ll find out soon enough whether these browser stats are the result of increased sell through or something else. The graph is now updated, going back to the initial inclusion of webOS in Statcounter.
May 20th, 2010 at 10:42 am
Well, the trend seems pretty definitive. I take it that webOS browser share drives most of the recent spike, rather than the mobile market growth?
May 20th, 2010 at 11:43 am
Yeah, after being stuck at 1% mobile market share for over 8 months (August 2009- March 2010), webOS share looks to have grown to 1.5% in the last 6 weeks. I don’t know what else would be the reason besides lots of Pre sales. Well, I guess the other possibility is that a lot of Pre sites started using StatCounter in the past month.
May 20th, 2010 at 1:30 pm
Given the VZ price cuts/ promotions and this evidence, it seems like 2X sell-through is in the bag (from Q3’s numbers). Mission accomplished I guess. Wonder if VZ will bother ordering any more, once their inventory is depleted. They seem pretty committed to this Android thing.
May 21st, 2010 at 3:54 pm
Saw a PrePlus for the first time at Starbucks! User seemed pretty proficient with it.
May 27th, 2010 at 11:39 am
Admob’s April metrics are out and don’t seem to bear out these market share gains (Pre goes from 2.4% of US “share” in March to 2.1% share in April).
http://metrics.admob.com/wp-content/uploads/2010/05/AdMob-Mobile-Metrics-Apr-10.pdf
May 27th, 2010 at 12:03 pm
We’ll soon discover who has the better proxy. StatCounter shows an increase of 1.05% to 1.16% from March to April, which includes all mobile devices (not just smartphones). One interesting bit is that total ad requests dropped in the US. Also, I was surprised to see the Pre show up in the UK report, as reports have always indicated slow sales there.
June 1st, 2010 at 2:40 am
So what’s your guess for sell-through this quarter? I’ll make my guess based on StatCounter, but I’ll need the numbers for the past few quarters to work out the model. You’re going to find those for me, OK?
June 1st, 2010 at 3:20 pm
You just want webOS sell-through numbers for the last few quarters? I would guess the following:
Q1 (end Aug 09): 550k
Q2 (end Nov 09): 478k
Q3 (end Feb 10): 423k
Q3 (end May 10): ??
June 1st, 2010 at 4:44 pm
OK, putting total faith in StatCounter we see a rise from 136 to 240 in the May quarter. That’s 10.7k sales for every point (1.5M/136). So 104 x 10.7 = 1.1 million phones. Problem is, the model is far from proven… it would predict sales of 952 in Q1, 160 in Q2 and 342k in Q3. The overestimate in Q1 (and corresponding underestimate in Q2) could be ascribed to the early adopter effect. Even so, there’s a 20% error in Q3 numbers.
1.1 million seems high. But relying solely on global stats, I think a range of 850K-1.1 million would be expected. FWIW, this chart shows some strangeness last summer/fall, when mobile use was certainly NOT declining. Best guess is that many mobile oriented sites stopped using StatCounter. Dunno. The big step in webOS share seems suspicious too, no way was it adopted that quickly. But that only accounts for about 1/4 of the quarterly gains, suggesting that 800,000 might be a better estimate.
http://spreadsheets.google.com/oimg?key=0AsZ890mI5DLHdHBJT3hNb250c001Uk9JMEtQS1hkdHc&oid=5&v=1275423512981
June 1st, 2010 at 6:01 pm
“StatCounter shows an increase of 1.05% to 1.16% from March to April, which includes all mobile devices (not just smartphones).”
Looking at the stat counter mobile OS chart, I see the end of April share to be 1.34%, not 1.16%. This is a much bigger discrepancy with the Admob report. Hmm.
Regardless, I’m going to guess 850K for Q4 (doubling the Q3 number). Assuming nearly all of those are Verizon, that would put about 2/3 of WebOS users on Sprint, 1/3 on Verizon. I wonder how many of those Sprint users will transition to the EVO? As of today, I have an upgrade eligibility burning a hole in my pocket :).
June 1st, 2010 at 7:26 pm
I was looking at the average for the entire month.
OK, I’ll make my official prediction. I’ll go ahead use the 1.11 million, subtract the 80K Q3 underestimate and call it 1030K for Q4. I contemplated estimating 851K, but that doesn’t seem very sporty.
June 15th, 2010 at 7:27 pm
Chugging along, browser use up another 16% in the 5 weeks since the OP. We’ll find out if these numbers are meaningful in a few days.
June 20th, 2010 at 1:50 am
Another few data points: Millenium Media is an Admob like mobile ad network that publishes their ad impression analytics (and are thus probably verboten now on the iPhone). Regardless, their reports fromMarch, April, and May indicate that Palm’s ad share (all US cell phones) has moved from 1.36% (in Feb) to 1.24% to 1.39% to 1.55% (in May). I find their results questionable in other areas, though– they also have the Pixi as the #8 US handset with 1.39% in May (Pre not in the top 20, but it would seem to be < 0.16%). Even weirder is the Nexus 1 at #3 (and no Droid to be seen). Most estimates have the Nexus 1 at maybe 200K total sales, while the Droid must be close to 4 million by now.
July 13th, 2010 at 8:17 pm
Looks like browser use is holding steady the last month, no big boost from the AT&T launch. Some slowdown may be attributed to people switching to Android or iOS after the EVO & iPhone 4 launches and putting their Pres away… but I would bet that’s a small number. We’ll see what happens in the next month.
July 13th, 2010 at 8:29 pm
FWIW, webOS use is about 60% of Android use from 1 year ago. In the meantime, while webOS use has tripled, Android use has risen by a factor of 7.
July 13th, 2010 at 11:56 pm
Does that seem right to you? Seems like Android’s installed base must have grown more than that, at least globally.
Anyway, what do you make of the latest Comscore numbers that suggest that the “Palm” platform had 2.3 million users as of May 31, ranking them 5th in the US? Seems more than could be accounted for by WebOS, but less than WebOS PalmOS. Is it possible that there are still more PalmOS users than WebOS users?
July 14th, 2010 at 12:04 am
Well, there were 1.4 million sold through February. Add in your 850, and that’s close to 2.3 million. Figure 90% in the US, so there numbers would seem a touch high for webOS, but not unreasonable. Must still be another 1-2 million Centros & Treos being used, wouldn’t shock me if the installed base was higher than webOS.
July 14th, 2010 at 12:18 am
FWIW, webOS use comes out to 265 units. For simplicity, we’ll assume that each “unit” equates to 10,000 phone sales in the US, and reduce it by 20% at the end. Android was at 450 units, or 4.5 million phones x 0.8. They now stand at about 3200 unites, or 25.6 million phones. That’s 22 million sales in the past year… which seems high? Might not scale well across platforms. How many phones does Google activate each day?
July 14th, 2010 at 2:38 am
Also, Apple is at 7500 units, up about 3000 units in the past year, barely outpacing Android for change in US browser share.
July 15th, 2010 at 8:59 am
I guess we’ll never know, but after talking to Palm, my feeling was that sell-through wasn’t as high as we’d hoped. Better, but not where it needs to be.
Couple responses:
* Would not have guessed there were 4.5m Android users in 6/09. 2 million, maybe.
* Both Apple and Android selling 22 million in a year sounds a little high, but probably close. Android is currently selling 5m a month (globally).
The Kin behind-the-scenes stories are fascinating to me. I’m shocked that both Microsoft and Verizon would “go through the motions” just to fulfill some 3 year-old contract/agreement. Makes me wonder if Verizon and Palm really did just sign a guranteed deal at CES 2009.
July 15th, 2010 at 11:12 am
Sad thing about Palm is it seems clear that they “bet the company” on having such a compelling product that consumers would still be beating a path to their door in January 2010, a year after the premiere. Yet they had hedged their bets, afraid carriers wouldn’t commit after their debut in January 2009. Clearly, they bet wrong & looking back it makes no sense. So confident to give be on only Sprint for a year after debut, yet lacking confidence to make that debut w/o soul-sucking contract in place.
Did you see the ChangeWave report, which shows Palm’s desirability falling off a very short cliff? http://www.engadget.com/2010/07/15/apple-and-htc-lead-charge-as-smartphone-market-looks-set-to-grow/ Anyone that wanted a Pre has already got one, and now Palm is left selling phones to smartphone newcomers– at least that’s what I gather from it. My guess is sales will sputter again soon without a new model, and if Palm doesn’t have a product to launch by early fall, it’s clear that the business would’ve gone broke w/o HP.
July 15th, 2010 at 1:10 pm
Not really surprised by Changewaves’ poll. Tough to get excited about WebOS as a platform with the complete lack of any visible activity in the past 5 months. It seems like they gave up after things didn’t go their way at Verizon.
My friend who was at Google got a free Pre at some developer event, but he called it unsable (mostly hardware related). He’s still kind of biased towards the Nexus 1, though.
August 15th, 2010 at 2:00 am
Surprisingly, my proxy for webOS installed base shows another uptick, 16% since June 1. Using my old guesstimate of 10,700 sales for each browser “point,” that yields an estimate of 428,000. Extrapolate that to the end of the quarter, we get 513,000 additional webOS devices. Take into account a 5% monthly attrition rate and assuming a wildly optimistic installed base of 2.4 million at the start of the quarter, and that adds another 360,000. That would give total sales of nearly 800,000.
Which has gotta be way high. Might just be that StatCounter’s tools are being used on more mobile websites, as the jump in usage seems outsized even given the thriving smartphone market.
http://gs.statcounter.com/#mobile_vs_desktop-US-daily-20100603-20100814
Now, I’ve seen other guesstimates of webOS installed base in the US of 2 million, which is probably 80% of all webOS devices. In contrast, my numbers suggest a total installed base of 2.9 million, about 2.4 million in the US. If that’s the case, I’d say Palm will sell closer to 650K this quarter, worldwide. Any other good proxies out there to judge sales the past 6 months?
August 15th, 2010 at 9:38 am
My guess is that cumulative webOS shipments to carriers was around 2.7M as of May 10, so that has to be a ceiling to the numbers entering the quarter. I haven’t heard of sell outs at either Verizon or Sprint. I can’t imagine any drivers to sales. In fact everything I’ve seen would lead to sales tapering off: high profile competitor launches, poor market share numbers, discontinuations at Best Buy and RadioShack, low rankings on Amazon best sellers. Heck, my Verizon store no longer even has any webOS phones on display, although they will bring out a box from the back on request. I agree with Precentral’s recent “article” suggesting total installed base at around 2m (10 days of Android sales). That seems more consistent with the sketchy evidence around.
August 15th, 2010 at 7:29 pm
I also realized that 5% attrition is too aggressive, that’s more appropriate for the 12-24 month folks. 2% attrition is probably more accurate, or 144,000 for the quarter. That’s total sales of 600,000 & a total installed base of 2.9 million (2.4 million in the US) as of 8/31/2010. That’s about 10% higher than the other estimates accounting that the other data is dated. All in all, I’d estimate total sales of 500,000-600,000 Q2 sales.
August 16th, 2010 at 2:09 am
Anectdotally, it seems like Pre has fallen off the radar screen at the retail and consumer level, so I’m pretty skeptical of year-over-year growth in webOS sales. Last year in Q2, I knew of 4 people who had bought Pre’s. This year, none.
August 16th, 2010 at 2:42 am
Oh, we’re still in Q1. Of course, you at least bought 1 Pixi! I was digging around for Pandora download stats, all I could find was the last month. Would be interesting to see how the rate has varied over the last 15 months. Agree with the lower profile of the Pre, but man, they are still making headway in the StatCounter numbers. Could be an artifact, but I’m not sure if there’s a better proxy measure of installed base.
August 16th, 2010 at 4:19 pm
Precentral reports today that both AT&T and Verizon are sold out online, lending credence to your “pie-in-the-sky” numbers. Total carrier sell outs would mean that end user sales would approach the theoretical maximum of 2.7 million (cumulative sell-in as of May). And we have no idea how many more got shipped in the current quarter.
August 16th, 2010 at 5:46 pm
We really should’ve published our work on PreCentral. Want to try turning our various guesses into a working article?
August 18th, 2010 at 2:06 pm
Millenial Media just sent out their report for July, and some of the irregularities (mentioned above) are cleaned up without explanation (Nexus One and Pixi is no longer highly ranked in the top mobile phones, as one would expect). Anyway, Palm is back to 1.38%, so like they are on statcounter, they are maintaining share in presumeably a growing market.
September 11th, 2010 at 3:28 am
Per http://apps.webosroundup.com/ , there have been 1.6 million Facebook installs. At a 48% install rate, that works out to … 3.3 million. That compares w/ my earlier estimate of total sales of 3.05 million as of 9/1. Just another data point.
September 12th, 2010 at 8:08 pm
test
September 12th, 2010 at 8:09 pm
That’s 1.6 million downloads, but it includes updates and re-installs.
October 5th, 2010 at 8:18 pm
Looks like webOS has dropped since August 15, and is pretty much flat lined since the HP acquisition completed on July 1. Almost surprised Pre sales help as well as they did so long. I have to imagine that Palm has several new devices planned to launch soon… surely they didn’t expect the Pre & Pixi to sell for 18 months nearly unchanged?
My take a long time ago was that webOS had to be better than Android for Palm to survive. That’s about where they were into the Fall of 2009, but since CES 2010 the platform has largely stagnated and the hardware is obviously dated. Sure, lots of great new apps & I’m still happy with the Pre. But… well, let’s hope the next round of gear at least matches the best of what HTC, Motorola & Samsung have to offer.
October 6th, 2010 at 4:10 pm
I think this is going to have to be the last time I participate in this thread. The lack of a scroll bar makes it excruciating to scroll to the bottom of this huge comment list on the Pre browser.
I have to assume that if todays’ rumors are true (no new webOS phones until 2011), that a whole generation of hardware must have been canceled along the way. Maybe it wasn’t state of the art compared to what the other OEMs were rolling out. At least as of last month, Palm was still sticking to a 2010 rollout of webOS 2.0, so FWIW, it means that our Pre’s will get it. I’d actually be satisfied with that, since Sprint hasn’t come out with a better alternative yet, except that I’m not sure my phone will hold up much longer. At this point, I’d hate to invest my upgrade or any cash at all on a current gen Pre. I’d be interested in trying out the Droid Pro if it were on Sprint. That’s going to answer the question of whether Android on a Pre would have been more successful than webOS on a Pre.
October 12th, 2010 at 12:48 pm
OK, I’m a masochist, extending this thread.
Don’t know where these guys get their data, but they say Palm had sell-in of 500K in the 2nd quarter (CY). Remember, that mostly corresponds to the quarter that was never reported, which Palm had pre-announced revenue of $100m. 500K would mean ASP of $200, which seems low, but possible, I suppose (combination of steep discounts to AT&T and rebates to VZ).
November 1st, 2010 at 12:37 am
Still flat lining. At this point, I’d wager that webOS device sales are just enough to balance the lost/ broken/ unused phones that have gone out of service.
November 1st, 2010 at 12:48 am
So with a blended attrition of 3%/month & an installed US base of 2.4 million, for the 3 months ending 11/30, I’d estimate total US sales of… 216,000. Maybe 250,000 worldwide. Ooph. Add another 100,000 if they actually launch on Verizon on Veteran’s Day.
November 1st, 2010 at 1:45 am
Sounds plausible. Odd that there’s no sense of urgency out of the HP camp.
Just for fun, let’s review the recent earnings reports. HTC announced shipments of 7 million (albeit with an ASP of $342), Moto shipped 3.8 million smartphones, Samsung shipped 7 million Galaxy S’s. Some of those HTC’s were WP7 handsets, not Android. Gonna be tough on Microsoft, let alone HP.
November 1st, 2010 at 2:30 am
HP isn’t worried because they figure one quarter is just that. I’ll post some thoughts in a new thread.
December 6th, 2010 at 12:35 am
In news shocking to no one, webOS installed base continues to slip. What’s interesting is to look at webOS share of the mobile market & mobile share of the overall market since August 2009. Mobile browsing has essentially tripled in a year.
https://spreadsheets.google.com/ccc?key=0AsZ890mI5DLHdHBJT3hNb250c001Uk9JMEtQS1hkdHc&hl=en&authkey=CK3Xi7cG#gid=2
December 6th, 2010 at 1:44 pm
That’s pretty crazy, the explosion of the mobile web. Wonder how much the iPad skews these numbers? its gotta be considered mobile (vs. desktop), and mobile share has doubled since its introduction, in no small part because b/c the average iPad user surfs the web >5X more than the average smartphone user, no doubt. The iPad could account for 12% of mobile web usage (3X webOS in units, 5X more in usage/unit = 15 * webOS’s mobile percentage). Not really relevant to the issue of Palm’s declining install base, but interesting nonetheless.
December 6th, 2010 at 2:35 pm
iPad really shouldn’t count as “mobile” anymore than netbooks. Alas, it does.
So yeah, the smartphone tsunami was real. Too bad Palm got sucked under the wave.
December 6th, 2010 at 6:20 pm
But I thought the iPad was just a big iPod touch? Certainly iPods are mobile. Besides, the only netbook Apple makes is the Air.
December 6th, 2010 at 7:06 pm
Statcounter divides web usage between desktop and mobile, which I take to mean “desktop” class browsers vs. “mobile” browsers. I’m not sure that’s a useful distinction, but there’s no ambiguity where an iPad should go vs. a netbook. Here’s a case study– I’m posting right now from my laptop, but in a cafe, but I’m using my Pre for internet connectivity b/c I’m paranoid about firesheep. Desktop, for sure, right, but no less mobile than the 10 iPad users I see sipping lattes around me (slight exaggeration).
When webOS starts appearing on printers, I imagine it will feature some variation of webOS’s neutered mobile browser, but calling it “mobile” isn’t very accurate. But what else can Statcounter do? Reading browser user agents is the extent of their powers.
December 7th, 2010 at 4:43 am
“HP isn’t worried because they figure one quarter is just that”
I’ve seen a lot of comments to this effect, based on the idea that it’s still early in the smartphone wars. It’s not really that early anymore. Yes, in terms of unit share (currently somewhere between 20-30%), there’s still a lot of room for smartphones to grow, but I bet the revenue share of smartphones is close to 70% in a market where the blended ASP is close to $100 and HTC reports an ASP 3.4 times that. Smartphones will trickle downmarket, but the the current players seem pretty entrenched in the high end of the market. No space for new entrants there unless they can go toe-to-toe with the best.
December 7th, 2010 at 8:34 pm
Ha, like clockwork, Jon Rubinstein just made the smartphones have “room to grow” argument. Well, if it helps you sleep at night, Ruby…
December 7th, 2010 at 9:41 pm
Too much is made of timing, but man, Palm has been perpetually 6 months late. If Palm Pre launched January 2009, with Verizon & AT&T getting the Pre Plus in September, and a next-gen phone in July 2010, we might be in a different place.