Marktplaats data; eBay’s entry into classifieds

Back in 2004, while working for eBay in The Netherlands, I was able to work on an interesting project. The project was code named Manhattan, and resulted in the acquisition of the largest classifieds site in The Netherlands, Marktplaats.nl, and eBay's entry into the classifieds business. Just last week, eBay bought another classifieds site, in Belgium. The classifieds business has become an important source of revenue for eBay Inc. an important school for learnings and a breeding ground for top talent who would migrate to other countries while working for eBay.

The decision to buy Marktplaats for $300 million in 2004 was not a fast and easy decision. The team in The Netherlands conducted a lot of research on the Marktplaats customers, the business model, and the sites ability to grow traffic and monetization to grow revenue. One of the research questions the team tried to answer was the affinity of the Dutch consumers to use the classifieds site over the eBay auction model. It turned out from part of our research, that the Dutch consumer would be willing to drive from one end of the country to the other end just to save a couple of Euro's. And with the country of The Netherlands being rather small, you can drive from the upper Northern part to the most Southern tip in a couple of hours, no seller is to far to quickly pick up your new DVD or piece of clothing.

Tweet this post:

That was 2004, now 9 years later, more data has become available through Marktplaats itself in a cbs publication. (CBS is the official state bureau for statistics in The Netherlands. The study is in Dutch) I stumbled on this data through the Facebook postings of some of my old colleagues.
The research shows some interesting maps and how the population of The Netherlands is using Marktplaats.nl to buy and sell stuff.

There is one map which shows exactly more in depth results of the research we did in the run to the acquisition of Marktplaats; how far are buyers prepared to travel to pick up the item for sale, and maybe even hackle over the price to save a little. Check out the map below, which shows the distance between buyer and seller for all ads placed on Marktplaats in 2010.
Distance between buyer and seller in The Netherlands on Marktplaats.nl, the largest classifieds siteThe table below shows the distance between buyer and seller for each of the items in the different categories, where it shows that the distance becomes smaller the more expensive or heavier the item is. Makes sense, as some of the smaller items, which also do not cost that much, might be mailed to the buyer after the successful bargaining.

distance between buyer and seller per category on Marktplaats

Marktplaats turned out to be a very successful acquisition for eBay, not only for the direct revenue growth, but more for the learnings on the business model and its ability to grow into new verticals and grow revenue. The research we as a team conducted proved to be on the mark, where years later the market still shows the winning business model for The Netherlands is classifieds rather than the eBay classic auction model.

You made it this far, why not share this post on Twitter
, or post a comment using Facebook below

eBay’s Paid Search Study Alternatives Motives for Publishing Ineffectiveness

I couldn't hide from all the chatter around the white paper about eBay's paid search campaigns today. All kinds of people with little knowledge on internal optimization within eBay are jumping on this story to give their perspective. The overall message they are so loudly shouting into the world is how bad eBay really is in PPC advertising, trying to win a little bit more authority on the subject themselves in the hope to win new clients.

I won't go into the details of the article, as I've never worked on eBay's paid search campaigns. Reading the research white paper eBay published is interesting though, and I know that some of it holds up, other parts not so much.

Like this article so far, please share on Twitter

Skeptical on the eBay Report

I'm somewhat skeptical on the actual outcomes of the research in the eBay report.

First of all, every site is different. So what would work for eBay, might not work for anybody else!

Second, you could use Paid Search campaigns in so many other ways, cross channel objectives, which has not been taken into account in this study.

Third, although the study looks at conversion to sales ratio's of the campaigns, the real ROI is hardly discussed.

Fourth,You should look at the fully loaded ROI to understand the complete impact on eBay's eco-system. This does not only incorporates the eBay fees, but should also include the Paypal fees when an item is being paid through Paypal and advertising revenue. And with Paypal penetration rates at a high double number, and advertising on almost every pageview, the campaigns ROI is highly influenced by multiple revenue streams.

Fifth, if the research holds true, can somebody explain me what happened end of Q3, beginning of Q4 last year, where SEO traffic for eBay tanked, and suddenly Paid grew out of the normal trend:

eBay SEO visibility tanking end of Oct 2012

Where Paid Search Campaigns for eBay are ramping up immediately

These graphs are made with Searchmetrics

Sixth; conducting a test on non-branded keywords, like Cell-Phone might be giving interesting outcomes, if you keep the way Google looks at sites from a SEO perspective into account. Keyword Cell-Phone is a head term for electronics, where eBay is not represented well in SEO. From the human rater documents which leaked from Google, there might be more obvious reasons why eBay is not ranking well:

How to rate eBay landing pages?

Alternative Motives for eBay to Publish Paid Search effectiveness Study

What most people commenting fail to take notice off, what are the real motives of eBay to publish such research? You really think a large, for profit company would make their direct competition smarter just to get some feathers stuck somewhere..? Below here I describe possible motives for eBay to publish this paper. These are all hypothetical, and not based on what really happened as I don;t have the internal information what has happened. Read these more as thin foil hypothesis for my enjoyment, and for you to think further than just the optimization of Paid Search campaigns.

Large Companies F*** each other over

eBay has had a long history of working with Google on a Frenemies status of their relationship. eBay gets a tremendous amount of traffic through SEO, and have worked with Google on several fronts to make the Internet a better place. However, there are well documented instances which shows the underlying relationship, which is one of stiff competition.

  • Google Dark; Back in 007, eBay shut down all Paid Search at Google to run a test. The black out period gave some interesting insights in the effectiveness of Paid Search at Google.
  • eBay sued Google over the launch of Google Wallet and trade secrets

For obvious reasons, I can hardly go into too much detail here, but imagine the drop in eBay spend over 2012, while Google is ramping up their efforts in the payments industry AND making Google Shopping pay-for-play, after killing off the shopping comparison vertical completely!

Internal Politics Kill Real optimization

In a company like eBay, there is a power struggle at the top on a continuous basis. The one with the bigger budgets control their own destiny, and can become very successful empire builders. With a change in C-suit players, the power struggle can become more aggressive, where studies to proof ineffectiveness of campaigns could destroy the power base of the existing rulers.

You've got Lies, Damn Lies and Statistics...

It has happened that one group proofed with analytic proof, that the advertising on eBay was cannibalizing the  conversion rate of items on the site, and had an impact on site speed. 4 months later, the VP of the group moved to the business side, became responsible for the P&L, and with 3 weeks into the quarter and a $20 million hole in the budget, made the executive decision to move the ads back on the site. Just make a check on the eBay site, it's full with ads right now, even on the View Item page which had always been a sacred page for sellers!

This Paid Search campaign effectiveness study could very well be a power play from inside groups to take control of the advertising budgets. Objective: proofing incompetence to replace and take control

False Information for Competitors

What happens to the overall marketplace if the research is adopted by more large advertisers? Right, less demand for the impressions out there will bring CPC's down in the marketplace.

By spreading this information, under the umbrella of statistical research, could very well have an effect on the ROI of existing eBay campaigns.

Be Careful What You Wish For

So when I read an article like this, where the writer is challenging the CEO of eBay to give him a chance to proof what he could do for eBay with Paid Search, I cannot help myself to think: "Be careful what you wish for" as the truth of the matter might not be as cookie cutter clear as you think, and you might find yourself in a situation you are not set up to succeed in the first place, even if you are the best Paid Search consultant in the world, there might be forces out there which want you to fail at it! (furthermore I don't think the writer did himself a service to build his arguments on 6 year old screenshots...)

The people who have managed the eBay paid search campaigns are good friends of mine, which is why I won't go deep into the criticism if the campaigns have been working for eBay or not, just to stay objective here. What I can say, is that the majority of the people commenting on the quality of the creatives, the supposedly fail to optimize these campaigns on a ROI positive manner, might not be so right at all. They probably have never managed a budget of the size of the eBay Paid Search budget, while optimizing for 170 million keywords, while landing the budget within 1% of the amount, each and every quarter and showing incremental ROI improvements time and time again. The teams who manage these campaigns are very smart and creative in doing new tests, which you can hardly judge upon as an outsider!

Disclaimer: I worked for eBay for 9.5 years in different positions, last 6 years of my tenure focused on Global SEO for all eBay sites cross the world. I cannot disclose all information I have, and do not claim to be an expert in Paid Search! These view do not represent the views of my current employer, nor eBay's!

Breakdown of @HackerNews Visitors

Ever wondered who else read HackerNews? You might have made friendships, found new business partners or got great amounts of traffic from HackerNews. I was interested in what the average HackerNews reader runs on their computer, where they are living and how engaged they are when a HackerNews reader clicks on a link. To satisfy my interest, I pulled the numbers for Slashdot and did some analysis... Here it goes:

Since August 1st, in the last 6 months, more than 115,000 people came to visit Slashdot from HackerNews. Several stories featured and discussed on Slashdot made the homepage of HackerNews, of which the top 3 were:
1. Rob ‘Cmdr Taco’ Malda resigns from Slashdot (13,828 pageviews) (HN link)
2. Canonical drops CouchDB from Ubuntu One (5,185 pageviews) (HN link)
3. Sopa creator in tv/film/music industry pocket (4,839 pageviews) (HN link)

With 115,000 visits, I believe I have a big enough sample to analyze the HN users to find some commonalities and to profile the tech set up of their computers. For the analysis I used the Google Analytics data from Slashdot. For simplicity sake, I created an Advanced segment, which will generate all reports in fast-mode. This means the reports are based on Sampled data in GA! Learn more about fast mode in GA at the Google Support pages.

HackerNews visits to Slashdot for the last 6 months in 2011First, let's breakdown the traffic, these 115K visits, a little further.

  • 115,812 Visits
  • 63,956 Unique Visitors
  • 155,569 Pageviews
  • 1.34 page/visit
  • 2.51 Ave time on site
  • 75.22% bounce rate
  • 41.49% New visits
Compared to the average numbers of the Slashdot visits in that same period, HackerNews visitors were spending half the amount of time on the site, there were twice the percentage of new visitors and all these people were just passing through, bouncing back with a much higher rate than the average Slashdotter. I guess we can say that Slashdot has a very loyal, sticky audience, on average!

Where are HN users from?

Off course it would be interesting to see where these 115K HackerNews readers are living. In Google Analytics the location report is a great tool to see where your website visitors are coming from.

Which countries do these HackerNews visitors coming from?

The majority of HN visits to Slashdot came from people located in the USA. There is a large gap between the US and the #2.

World map of HackerNews visits

Which countries do HackerNews visitors come from?The top countries with the visits are:

  1. United States 70,524 visits
  2. Australia 10,099 visits
  3. India 4,839 visits
  4. United Kingdom 4,148 visits
  5. Canada 3,457 visits
  6. France 2,074 visits
  7. Netherlands 2,074 visits
  8. Germany 1,728 visits
  9. Norway 1,037 visits
  10. Argentina 691 visits
I would like to mention here, that the fast-access mode in GA is rounding up the total reported visits in some cases. This is causing a large number of countries to report same amount of visits. Number 10 to number 24 all report 691 visits in the last 6 months. As this number is so small, it's insignificant to the overall analysis. So if you live in Brazil, Italy, Israel or Pakistan, consider yourself lucky to share the #10 spot with Argentina!

Which Cities do these HackerNews readers call home?

The outcome of which city was sending the most visits from HackerNews to /. in the last 6 months was surprising to me. Based on the US sending the most visits from HN to Slashdot; I would have expected a city in the US to have the top spot on this list. Some would say this pends how GA slices the data, where the Bay Area is a good contender for the first spot based on the high concentration of high tech. However, would you have thought the #1 spot would be a shared nomination between an Australian & US city: Brisbane and San Antonio?
City map for HackerNews visits to Slashdot
Not as I expected, but the #1 spot for cities HN came from is shared between two cities. One is Brisbane, the other is San Antonio.
  1. HackerNews visitors city Brisbane 9,679 visits
  2. San Antonio 9,679 visits
  3. San Francisco 8,988 visits
  4. Redwood City 6,222 visits
  5. Oakland 5,531 visits
  6. Bangalore 4,148 visits
  7. New York 2,419 visits
  8. London 2,074 visits
  9. Bellevue 2,074 visits
  10. Paris 1,728 visits
Number 11 & 12 have the same number of visits as Paris. Both Los Angeles and Sydney were reporting 1,728 visits in the last months to Slashdot from HackerNews. If you would add up all the Bay Area cities, San Francisco; Redwood City and Oakland, the Bay Area would have been #1 with 20,741 visits! 

What browser is the favorite for HN users?

It has become clear from the visitor numbers I've analyzed, HackerNews readers use Internet Explorer very limited. The browser of choice for the HN visitor is Chrome. It's impressive how Chrome is seeing high growth in market share in the browser wars, especially in tech savvy communities. For Slashdot's HackerNews visits, the browser shares were dominated by Chrome to an extend I didn't expected.Browser of the HackerNews visitors to Slashdot

  1. Chrome 67,104 visits
  2. Firefox 21,434 visits
  3. Safari 21,088 visits
  4. Internet Explorer 2,765 visits
  5. Android Browser 1,728 visits
  6. Mozilla Compatible Agent 345 visits
  7. Opera 345 visits
You could say, based on these numbers, the role of Internet Explorer in the browser market is finished!

What Operating System is the choice for HN users?

The operating system of choice for the HackerNews visitor is not a landslide victory for Apple Mac OS, but there is a large gap before Windows appears on the #2. Linux takes a respectable third place, after which the mobile OS' are coming in on #4,5 and 6. Operating system of the HackerNews visitor

  1. Macintosh
  2. Windows
  3. Linux
  4. iPad
  5. iPhone
  6. Android

It would be interesting to see how the mobile operating systems take a larger share over time.

Of the Window users 84% is on Windows 7, while only 2% of HackerNews visitors rocking it on Windows Vista. Thirteen percent is still on Windows XP. Just 1% of Windows users are on Server 2003.

Screen resolution of the HN user?

And last, I wanted to see what kind of large screens HackerNews readers are using. The larger the screen resolution might indicate a very large screen!

  1. Screen resolution of the HackerNews visitors1440x900 20,051
  2. 1280x800 14,865
  3. 1366x768 14,174
  4. 1920x1080 13,828
  5. 1920x1200 13,136
  6. 1280x1024 8,297
  7. 1680x1050 7,951
  8. 1024x600 4,494
  9. 768x1024 4,494
  10. 1024x768 3,457

 

Engagement of the HackerNews visitors

As the image at the top shows, the Average HackerNews visitor to Slashdot was not very engaged. With a bounce rate of ~75%, and 1.4 pages per visit and an average of 2.4 minutes on the site. These engagement numbers are considerable lower than the average Slashdot visitor.

Conclusion

On my Chrome browser, I have HackerNews set as the homepage. Every time I start my browser I can read instantly what is happening in the technology industry. HackerNews can bring any website a large amount of traffic, but you should always pay attention to what kind of traffic you need to be successful. If you are running a media company, and get paid based on CPM's, HackerNews can be a great source of traffic.

Let me know if you would like to see these kind of analysis more often on different referrals.

Data: The data showcased in this post has been derived from the Slashdot Google Analytics for the period of August 1st - December 20th.

Disclaimer: The data presented in this analysis has been derived from Google Analytics on Slashdot. It represents a small subset of the total. The comments included herein are my own and don’t necessarily represent Geeknet’s opinions

Christmas Sweaters Hit The Big Times

Could not parse file.