China to launch trial of 3G mobile phones

China Mobile (CHL) is going to start the TD-SCDMA commercial trial on 4/1/2008. The following is the news in English and in Chinese.

http://www.usatoday.com/tech/wireless/phones/2008-03-28-china-3g_N.htm?csp=34

http://tech.sina.com.cn/t/2008-03-28/11182106926.shtml

My last post on WVAS is here:

http://chinese-net-gaming-stock.blogspot.com/2007/04/rumored-china-mobiles-chl-wap-policy.html

The Chinese news from sina has a lot more information in them.

TD-SCDMA is a 3G cellular phone standard. The world only have three 3G standards, the WCDMA, the CDMA2000 and TD-SCDMA. WCDMA is the standard developed by Europe (its 2G equivalent is GSM). CDMA2000 is developed by American (its 2G equivalent is CDMA95). TD-SCDMA is developed by China.

It was a tremendous technical break through for China.

Chinese mobile operators don’t like TD-SCDMA. They have good reasons. A cellular phone standard is so complicated, it will be impossible for TD-SCDMA to be even close to the two existing standards in terms of performance and robustness.

Even if Albert Einstein and Thomas Edison are the one invented TD-SCDMA, it will be at least 10 years behind both the WCDMA and CDMA2000.

The only way TD-SCDMA can survive is for the Chinese government to dictate China’s mobile operator to use it.

But to conform to WTO charter, China has to give WCDMA and CDMA2000 level playing field. But if China follows the intend of WTO charter, TD-SCDMA won’t have any chance to survive.

But I guess lawyers have to have some use in them. I think China will do anything and everything to ensure TD-SCDMA will not just survive, but to thrive.

Expert had been predicting this date since 2005. It had been delayed so many times. The date is finally here. So this is a pretty big news.

I think there are several type of companies that will benefit from the wide-spread usage of TD-SCDMA.

First, equipment maker like SPRD will benefit for obvious reason. Second, I think WVAS companies such as KONG and SINA will benefit greatly. The WVAS companies that specialize in WAP will be the winner. KONG is betting it future survival on this date. SINA has been gearing all its web site to be compatible with WAP. For SINA, every story SINA generated have WAP counterparts. Both KONG and SINA had been throwing money into the drain for the past few years just to gear up for the arrival of 3G.

Of course, for the WVAS companies, the mobile operators such as CHL or CHU can throw monkey wrenches into this. But, unlike for SMS where KONG and SINA are just distributor and middlemen, now KONG and SINA will be content generators for the WAP. KONG and SINA will be in a stronger position. In addition, there is talks about China shaking up the mobile operators. That may prove positives to the WVAS companies since the mobile operators will be busying in re-structuring. Also, China will have a new ministry that will handle this function. The charter of the new super ministry can change also. That adds some uncertainties. The power of this industry is so lopsided tilt toward mobile operators such as CHL and CHU, any change to the existing structure can only help the WVAS companies such as KONG and SINA.

But according to sina link above, only 20,000 users will take part in the trial. There is no news of how long the trial will last. Therefore, it will not have any material contribution to all the companies that are discussed here. But in a year time, this could be an exciting growth driver for all the company mentioned here.

FMCN’s comments on China Internet advertising business in 2009

My last post on online advertising business can be found here:

http://chinese-net-gaming-stock.blogspot.com/2008/03/statistics-from-dcci-on-chinas-online.html

On 3/19/2008, FMCN had a earning conference. In this call conference, they talked about the overall internet advertising business in 2009.

Companies such Sina or Sohu might know a sector of China’s internet advertising business, but for the overall China internet advertising business, there is nobody better than FMCN. FMCN’s customers are companies such as Sina and Sohu. Therefore, it give them a bird’s eye view of the overall industry.

Given how well FMCN had been growing and how capable is Mr. Jason Jiang been piloting the company, there is nobody better than predicting the future direction of China’s internet advertising sector.

From the earning call transcript:

http://seekingalpha.com/article/69120-focus-media-q4-2007-earnings-call-transcript?source=yahoo

In 2009, he expect the overall advertising industry to grow a very strong 15%. He is confident FMCN will grow 50% in 2009. The following is the quote on the internet sector:

“…..for the Internet mobile advertising we believe actually during the Olympic period, the benefit of advertising in this media will be highlighted and we expect both Internet advertising and mobile handset advertising will grow very strongly in 2009 as these forms of media get recognized by the advertisers and it will enter a very strong growth period after the 2008 Olympics.”

He actually down play the effect of Olympics on FMCN. There won’t be any price increase specifically for the Olympics.

The above is exactly how I felt about the affect of Olympics. There will be some pick up for online advertising revenue prior and during the Olympics. But somehow, I felt the effect is being exaggerated.

But I think the real effect of the Olympics is to serve as the demonstration platform for all the new products (video, P2P, internet TV, blog, and other web 2.0 services) that companies such as Sina and Sohu had been working on. These new products are much more suitable for advertising than the old text-based products.

China Mobile taking action to stop the mobile spamming

My last post on China’s mobile industry can be found here:

http://chinese-net-gaming-stock.blogspot.com/2008/03/fmcn-apologized-for-mobile-spamming.html

From the FMCN mess, China Mobile also apologized.

http://it.sohu.com/20080320/n255805285.shtml

http://tech.sina.com.cn/t/2008-03-19/17482087916.shtml

In addition, it provides some additional information that I think is very interesting for the whole Mobile Value Added Service (MVAS) industry.

http://tech.sina.com.cn/t/2008-03-20/09342089505.shtml

I will translate in the following.

In Guangdong province, the number of customer complains had went down from 1378 cases in 2006 down to 1022 cases, or a decrease of 27%.

In the coming year, the government will emphasize on three areas for enforcement. First, IVR; second, SMS spamming; and third, un-solicited voice emails.

I think most of the SP had been nailed by SMS spamming. Therefore, I think companies like SINA, KONG, HRAY, and LTON had learned their lessons.

I am not familiar with illegal IVR and un-solicited voice emails and how much the above mentioned companies’ involvement in these two areas. Even though I don’t think the above four companies are involved, but it is something needs to keep track of.

FMCN apologized for mobile spamming

My last post on China’s mobile industry can be found here:

http://chinese-net-gaming-stock.blogspot.com/2007/04/rumored-china-mobiles-chl-wap-policy.html

http://it.sohu.com/20080320/n255804036.shtml

http://it.sohu.com/20080320/n255806286.shtml

http://english.people.com.cn/90001/90776/90882/6377632.html

There was a major scandal that involves FMCN. It was kind of confusing for a couple of days. But after the dust settles, I think I got a pretty good handle.

First some background information:

On 3/15/2008, China’s state-owned flag ship TV station, CCTV, had a show that documents that 7 MVAS (Mobile Value Added Services) companies had conducted SMS spamming. The SMS spamming involves sending unwanted SMS messages to un-suspecting customers.

There are two issues involved, first, was FMCN involved in actually doing SMS spamming? And if so, how did FMCN obtained the mobile phone number of these customers?

On 3/18/2008, China Mobile apologize and said it had block off these 7 companies.

On 3/19/2008, FMCN CEO apologize for doing SMS spamming. Also on 3/19/2008 (China time), during the earning call conference, FMCN announced the stoppage of all business activities involved in that part of wireless business.

On 3/72006, they purchased Dotad Wireless and convert into a subsidiary of FMCN and called it Focus Media wireless. It has three parts of business, one is to help companies send SMS messages to its members. The second is the new type of mobile business that FMCN talks extensively in it earning conference and they had already signed up big companies such as MSN and KONG. The third type of business is the SMS advertising business and SMS spamming will be within this business.

After the CCTV reporting, FMCN did an internal auditing and found that its subsidiary did committed SMS spamming. As a result, FMCN will completely get out of all SMS advertising business. The other two wireless business categories are not affected. Based on this, they predict they will make 54 to 55.8 million in 2008 (from 47m in 2007). The full affect of getting out of SMS advertising business will start to be felt in the 2^nd quarter of 2008.

In addition, they said all companies employees that caught doing this will be fired. They also explained that they obtained the users’ phone number by monitoring when these users browsing on their WAP site. That is legal. The outside Chinese legal experts quoted on various news reports all agreed that this is fine for FMCN to collect these user information because these are public domain information if users willing to give out their phone numbers on these WAP sites.

It took me a while to figure all these out. There are several conclusions to be made.

First, FMCN is just growing too fast. Jason Jiang is probably the most capable executive I had followed. But the pace he is buying companies is just way too fast. The company needs time to integrate. Promoting Dr. Tan Zhi to CEO is an excellent idea. Now it frees Jason for better monitoring of its other far flung subsidiaries. In the earning conference, FMCN talked about getting into the area of digital TV advertising business. I am sure they will probably buy some companies in that area. But I hope they would stop buying any more companies for couple of more years to allow time for better integration and better monitoring. As they focuses on integration and less on acquisition, that shall improve the margin for the next few years.

Second, the company made a correct choice. It came clean and basically just chopped off one of its arm. It is a clean break and it completely stopped the growing story dead on its track. By completely getting out of this business, it becomes immune to whatever China Mobile can do to it.

Third, the Wall Street’s reaction is completely irrational. Even after FMCN completely getting out of the SMS advertising business, it will make 900m to 930m (about 80% year over year) in 2008. It completely trashes the analysts’ estimate of 836m.

Now, lets make it even more extreme; lets assume the worst of the worst, lets assume they will completely getting out of all wireless business. Wireless is expected to be account for 6% of total business in 2008. Without wireless, FMCN expects to make 846 to 874m, or about 70% year over year.

As of now, FMCN share price is $36 and a trailing PE of 32 and forward PE of 17.
All that for an 80% growing company! Are you kidding me?

XYQ Server Update

My last update on XYQ can be found here:

http://chinese-net-gaming-stock.blogspot.com/2007/09/quick-update-on-xy3-and-xyq.html

NTES had been very good at publishing new Peak Concurrent User (PCU) records whenever one occurs. In 2006, XYQ had 9 PCU records. In 2007, it only has two. The first PCU record in 2007 occurs on 3/4/2007 and it had a PCU = 1500k. It takes another 9 months to have another record. On 12/2/2007 it had a PCU = 1520k. This new record is only 20k more than the one happened 9 months ago.

From the above data, I started to think that this game had peaked and we will start to see the long term slow decline for XYQ. But it looks like I was wrong.

On 3/2/2008, XYQ not just get a new record, they absolutely crashed the old record. Its new PCU record is an amazing 1660k (an increase of 9.2%)!

But is this new PCU indicates that there is something fundamentally changed? Let’s see how the server count says.

The following is the server count table:

Month	Saturday/Sunday	4Q2007	1Q2008
1	1
	2	1254	1385
	3	1318	1340
	4
2	1	1498	1469
	2
	3	1338	1397
	4	1383	1489
3	1	1519*	1561**
	2	1325	1443
Holiday	5	978	1117
	6	965	1268
Total		11578	12469

* On 12/2/2007, Official PCU = 1520k while my calculated PCU (based on my server count) = 1519k

** On 3/2/2008, Official PCU = 1660k while my calculated PCU (based on my server count) = 1561k

The first column is the month. There are three months in each quarter. In addition, there is a one week national holiday in both 4Q 2007 and 1Q 2008. I used 2 days out of the 7 day national holiday.

The second column is the order of weekends. There are usually 4 weekends per month. The number in the third and fourth column represents the PCU for that particular weekend.

From the last row, one can see ACU increased from 4Q07 to 1Q08 by (12469 – 11578)/11578 = +7.7%

Note that on 3/3/2008, I used my calculated PCU of 1561 instead of the real official PCU of 1660. If you want to use the real number of 1660, then the accumulated total will be 12566 instead of 12469. In that case, the increase from quarter to quarter will be (12566 – 11578)/11578 = +8.5%

The strength of the growth is very impressive.

Now, lets try to relate ACU with the revenue. Let me use the following table:

	4Q2006	1Q2007	2Q2007	3Q2007	4Q2007
PCU	1335	1500	1470	1440	1520
PCU %		12.3%	-2%	-2%	5.6%
ACU	444	510	491	517	515
ACU %		14.9%	-3.7%	5.3%	-0.3%
Revenue	40.5	45.3	46.2	48.8	50.7
Rev. %		11.9%	2%	5.6%	3.9%

The second row of the table is the Peak Concurrent User number. The third row is the quarter to quarter increase/decrease for the PCU. The fourth row is the Average Concurrent User. The fifth row is the QoQ increase/decrease for the ACU. The sixth row is the revenue from XYQ. The seventh row is the QoQ increase/decrease for the XYQ revenue.

For 1Q 2008, my server count suggested an increase of 7.7% for the ACU. My guess for the revenue will be an increase of 8%. This will give me a revenue of 50.7 * 1.08 = 54.8 millions for 1Q2008.

This result surprises me. An over-the-hill title is not supposed to have a QoQ ACU increase of 7.7% and PCU increase of 9.2%

It is possible that XYQ may still have one or two more years of growth to go. But whether this indicates a fundamental shift or just a blip, we will have to see how 2Q2008 come out.

Market in China keeps on surprising me. A six year old title like XY2 is still growing. A four years old title like XYQ explodes like a new title. Yet a new title like XY3, even with a pretty good head start, is showing signs of aging. If the 1Q2008 earning conference call confirmed my server count estimates, I would certainly eager to hear Ntes’s management’s explanations.

Finally, Ntes had stopped most of its marketing campaign for all its games during the first quarter. I just can’t say enough about how disappointed in Ntes’s management in pinching pennies. There are so many good titles out there in China. A new title like XY3 got to have marketing help. XY3 can still achieve very popular status if Ntes is willing to spend some money to market it.

XY2 and XY3 Server Status part2

Part 1 of this article is about XY2. I will devote part 2 for XY3. Part 1 of this article can be found here:

http://chinese-net-gaming-stock.blogspot.com/2008/03/xy2-and-xy3-server-status-part-1.html

On 11/7/2007, XY3 consolidated its servers from 140 servers to 60 servers (as of now, the server had increased to 68). It introducing an accuracy question regards the statistics from the server count. For one thing, it is not clear that they uses all brand new servers. If they use all brand new servers, then it becomes more likely that the new server will have more user capacities. In that case, they might change the player counts before declaring the server busy, full, etc.

Another factor that introduces inaccuracy is that the game is very new and there is very little data to base on. When Ntes say its server is busy, full, good , or excellent, how do that corresponds to the actually number of users on that server? There are a lot of calculations between server count that I observe every night to the generation of PCU estimate from the observations. From the limited data I collected, I am able to make some rough estimates for me to calculate PCU. But they are really very rough estimates. At this point, the confidences for these estimates are not high. It will take another one to two quarter before I will feel confident in my PCU estimate.

Something else to keep in mind is that I can no longer stay up everyday to do server count. I only do server count during weekends.

But since I am using the same data to compare 4^th quarter 2007 and 1^st quarter 2008, they are apple to apple comparison. Even though they are rough estimates, but they are the only thing I have to go by. Better or worse, this is all I have.

The following is the server count table:

Month	Saturday/Sunday	4Q2007	1Q2008
1	1
	2	121	127
	3	183	118
	4
2	1	128	121
	2
	3	113	116
	4	114	123
3	1	114	117
	2	108	120
Holiday	4	179	122
	5	166	124
Total		1226	1088

The first column is the month. There are three months in each quarter. In addition, there is a one week national holiday in both 4Q 2007 and 1Q 2008. I used 2 days out of the 7 day national holiday.

The second column is the order of weekends. There are usually 4 weekends per month. The number in the third and fourth column represents the PCU for that particular weekend.

From the last row, one can see ACU decreased from 4Q07 to 1Q08 by (1088 – 1226)/1226 = -11.3%

It is a little surprising to me that there is actually an increase this quarter. Ntes didn’t really add a lot of contents this quarter. The expansion pact won’t be ready until the 2Q. It is clear that XY2 is growing again.

Now, lets try to relate ACU with the revenue. Let me use the following table:

	4Q2007
ACU	53
ACU %
Revenue	4.9
Rev. %

The second row of the table is the Average Concurrent User estimate. The third row is the quarter to quarter increase/decrease for the ACU. The fourth row is the revenue from XY2. The fifth row is the QoQ increase/decrease for the XY3 revenue.

For 1Q 2008, my server count suggested a decrease of -11.3%. My guess for the revenue will be a decrease of -12%. This will give me a revenue of 4.9 * 0.88 = 4.3 millions for 1Q2008.

This result is very disappointing.

There are very little features added in XY2, yet its ACU increases. On the other hand, for XY3, Ntes added a pretty major feature at the beginning of March, yet its ACU decreases. XY2 is an old game, but it is growing, while XY3 is brand new and it is decreasing in popularity.

I think Ntes erred in two major areas. First, there is virtually no advertising for any of its games during the 1Q. They ought to start to advertise for XY3. Secondly, they shall not have announced that there will be an free-to-play version of XY3 during the 2^nd half of 2008.

When players want to invest their time and money in a game, they want to be sure that this game will be around for a long time. When you market a game, it tells the players that the company is willing to spend money for the game. By announcing another version of the same game, it definitely introduce uncertainty in players’ mind about whether the current version of game will survive.

In summary, I am very happy with XY2’s performance while very dissatisfied at XY3’s performance. Again, the problem is not the game, but the marketing of the game.

XY2 and XY3 Server Status part 1

My previous post on XY2 is here:

http://chinese-net-gaming-stock.blogspot.com/2008/01/xy2-xy3-server-status.html

On 1/1/2008, XY2 consolidated its servers from 275 servers to 128 servers (as of now, the server had increased to 136). It introducing an accuracy question regards the statistics from the server count. For one thing, it is not clear that they uses all brand new servers. If they use all brand new servers, then it becomes more likely that the new server will have more user capacities. In that case, they might change the player counts before declaring the server busy, full, etc.

Without any additional insight, I assume they keep the same servers. Something else to keep in mind is that I can no longer stay up everyday to do server count. I only do server count during weekends. For most games, it is fine because most players play during weekends. But for XY2, Thursday is also a busy day. Any statistics that won’t take into Thursday for XY2 will be incomplete.

But since I am using the same data to compare 4^th quarter 2007 and 1^st quarter 2008, they are apple to apple comparison. It might not be extremely accurate, but they are still accurately enough to be useful.

The following is the server count table:

Month	Saturday/Sunday	4Q2007	1Q2008
1	1
	2	168	319
	3	298	297
	4
2	1	255	248
	2
	3	315	286
	4	247	302
3	1	274	271
	2	254	322
Holiday	4	325	285
	5	318	257
Total		2454	2587

The first column is the month. There are three months in each quarter. In addition, there is a one week national holiday in both 4Q 2007 and 1Q 2008. I used 2 days out of the 7 day national holiday.

The second column is the order of weekends. There are usually 4 weekends per month. The number in the third and fourth column represents the PCU for that particular weekend.

From the last row, one can see ACU increased from 4Q07 to 1Q08 by (2587 – 2454)/2454 = 5.4%

It is a little surprising to me that there is actually an increase this quarter. Ntes didn’t really add a lot of contents this quarter. The expansion pact won’t be ready until the 2Q. It is clear that XY2 is growing again.

Now, lets try to relate ACU with the revenue. Let me use the following table:

	4Q2006	1Q2007	2Q2007	3Q2007	4Q2007
ACU	173	170	162	124	134
ACU %		-1.7%	-6.4%	-23.5%	8%
Revenue	16.2	15.4	15.6	11.9	13.2
Rev. %		-4.9%	1.3%	-23.7%	10.9%

The second row of the table is the Average Concurrent User estimate. The third row is the quarter to quarter increase/decrease for the ACU. The fourth row is the revenue from XY2. The fifth row is the QoQ increase/decrease for the XY2 revenue.

From the above table, with the exception of 2Q2007, the revenue follows ACU. For a paying game, ACU can be a reasonable estimator.

For 1Q 2008, my server count suggested an increase of 5.4%. My guess for the revenue will be an increase of 6%. This will give me a revenue of 13.2 * 1.06 = 14 millions for 1Q2008.

The second part of this article will be about XY3.

TLBB Status Update

My last TLBB specific post is here:

http://chinese-net-gaming-stock.blogspot.com/2008/02/tlbb-expands-oversea.html

The server counts for the TLBB can be found in the following table:

Month	Saturday	4Q2007	1Q2008
1	1
	2
	3	447	545
	4	450	496
2	1	481	461
	2
	3	502	499
	4	500	526
3	1	500	480
	2	510	482
Holiday	4	395	431
	5	401	436
Total		4186	4356

The first column is the month. There are three months in each quarter. Both 4Q2007 and 1Q2008 have a one-week long national holiday. I compare the holiday between the two quarters also.

The second column is the Saturday in each month. Most month has 4 Saturdays. I also compare the 4^th and 5^th day of the 1-week long national holiday. The third column and fourth column are the peak concurrent user number for that particular Saturday (or holiday).

The last row is the total (or the sum of all the rows).

You may ask why are so many days missing. Well, I can no longer stay up during the wee-hours every day to keep track of the server counts.

From the last row of the table, you can see the average concurrent user has increase by about (4356-4186)/4186 = 4%

But how can we use this estimate (+4% in ACU) to give us an estimate for the game revenue for Sohu? The answer is we need to do a lot of guess work.

Note that since I only have a few data point in calculating ACU, this ACU estimate won’t have any relevance when compare with ACU that were developed using more data points. But it is ok to use it for this particular case.

The following is a table for Giant Online’s (ZhengTu) game statistics

	3Q2006	4Q2006	1Q2007	2Q2007	3Q2007	4Q2007
APA	.698			1.248	1.318	1.405
APA (%)					5.6%	6.6%
ARPU	220		320	295	305	309
ACU	271			515	481	512
ACU (%)					-6.6%	6.4%
PCU	558			1072	880	983
PCU (%)					-18%	12%
Revenue	23.7			48.9	54.1	59.6
Rev (%)					10.6%	10.2%

The second row is APA or Active Paying Account in unit of millions. The third row is the quarter over quarter (QoQ) growth of the APA. The fourth row is ARPU, or Average revenue per active paying customers. The fifth row is ACU (Average Concurrent Users) in unit of thousands. The sixth row the QoQ growth of ACU. The seventh row is PCU (Peak Concurrent Users) in unit of thousands. The eighth row is the QoQ growth rate for PCU. The ninth row is the monthly revenue in millions. Since Giant Online only have one game of significance (ZhengTu), they also represent the revenue generated from this one game. The final row is the QoQ revenue growth.

One thing interesting can be found on 3Q2007 estimate. On 3Q2007, the ACU decreased by 6.6% and PCU decreased by 18%, yet the APA increased by 5.6% and the revenue increased by 10.6%.

This shows that both PCU and ACU are not very reliable estimate for free-to-play games. APA and ARPU are much more reliable. I think since ZhengTu is a relatively new game, both APA and ARPU tend to trend up from quarter to quarter. While increase in ACU tend to for-tell increase in APA, the reverse might not be true.

The following table has the statistics for TLBB.

	2Q2007	3Q2007	4Q2007
PCU	400k	400k	500k
PCU (%)			25%
APA	209k	690k	1100k
APA (%)			59%
ARPU	82	118	147
ARPU (%)			25%
Game Revenue	3.8	12.7	24
TLBB Revenue	2.3	10.9	22
TLBB Rev. (%)			102%

This table is pretty self explanatory. Compare this table with the table for ZhengTu, we can see several points. Compare TLBB’s PCU of 500k and ZhengTu’s PCU of 1072k, TLBB may still have a way to go. Not saying TLBB will eventually has as many users as ZhengTu, but considering how young TLBB is, the odds are TLBB’s PCU will significantly improve.

Compare TLBB’s APA of 1100k with ZhengTu’s APA of 1405k, you would think TLBB’s APA will grow, but not by a very big percentage from quarter to quarter.

Compare TLBB’s ARPU of 147 with ZhengTu’s ARPU of 309, I think TLBB’s ARPU can still have some distance to grow. In the earning conference, Sohu say that they predict the peak ARPU is 170. I think they are way too conservative.

I think in two to three quarters, the APA will probably be something like 1600k (or 45% increase) while ARPU be something like 190 (or 29% increase).

Now, given an ACU increase of 4% quarter to quarter, I estimate we shall have APA increase of 8%. For the ARPU, I estimate we might have 5% increase. Thus, the revenue from TLBB shall have an quarter over quarter increase of 1.08 * 1.05 = 1.134, or 13.4%. or the revenue from TLBB will be 22 * 1.134 = 24.95M

The following table is the revenue from Sohu’s other game, Blade Online

	2Q2007	3Q2007	4Q2007
Blade Revenue	1.5	1.8	2.0

Assuming Blade Online no longer grows, I predict that Sohu will get 2M from Blade Online in 1Q2008.

Combine the two, we have the total game revenue of 24.95+2 = 26.95M for 1Q2008. Compare to Sohu’s guidance of 25.5 to 26.5 millions, I think Sohu gave a pretty good guidance for the gaming portion of her revenue mix.