12. When does the 1st day of school really start?
September
4th, 2014
New York City, the largest
school district in the US,
will have 1,000,000
students back in school
August
25th, 2014
All 5,000,000 of Texas’s
K-12 Public School
students go back
August
1st, 2014
All 180,000 of Hawaii’s
K-12 Public School
students go back
#HeroConf | @jmgagnon | @bingads
13. #HeroConf | @jmgagnon | @bingads
When parents will shop for back-to-school
Use staggered start dates to hit
the 1 week to 1 month sweet
spot. Two months for New York
means, July 9, whereas it means
June 5 for Hawaii.
22.4% will shop
as early as June.
<6% will shop the
‘week of’ or ‘after’
school starts. Keep
budgets low during
this time.
2 Months prior
1 Months prior
1-2 Weeks prior
Week of
After
22.3%
47.8%
24.0%
2.7%
3.2%
22. #HeroConf | @jmgagnon | @bingads
Chegg’s bid-up strategy by week
8/11 – 8/17 9/8 – 9/149/1 – 9/78/25 – 8/318/18– 8/24
Group A
(8/18 start)
Group B
(8/25 start)
Group C
(9/1 start)
15%
20%
15%
10%
20%
15%
10%
20%
10%
Chegg April 2015
23. #HeroConf | @jmgagnon | @bingads
Chegg makes the grade
Cost per order
decreased 30%
$325K saved YoY
30% 325k
Chegg April 2015
24. #HeroConf | @jmgagnon | @bingads
Agenda
Chegg: A case study
Voice search
Research:
Brand-term bidding
Conversion path
Remarketing theory
25. #HeroConf | @jmgagnon | @bingads
Siri Cortana Google Now
Let’s make Voice search is part of the PPC conversation
26. #HeroConf | @jmgagnon | @bingads
Voice search is now part of the PPC conversation
US smartphone users who use mobile personal assistants
38%
39%
59%
71%18-29
30-43
44-53
54+
Thrive Analytics, “Is the Personal Assistant the Successor to Search?” October 2014
27. #HeroConf | @jmgagnon | @bingads
How is voice search being used?
43%
38%
31%
Call
someone
Ask for
directions
Help with
homework
Teens
40% 39%
31%
Ask for
directions
Dictate
texts
Call
someone
Adults
Northstar Mobile Voice Study for Google October 2014
28. #HeroConf | @jmgagnon | @bingads
Are we all confused?
What does the fox say?
Weather
What time is it?
Where am I?
Cortana natural language conversational queries, voice
Microsoft internal data April 2015
29. #HeroConf | @jmgagnon | @bingads
Voice search is already adding to text search
24%
76%
Text
Voice
Microsoft internal data April 2015
30. #HeroConf | @jmgagnon | @bingads#HeroConf | @jmgagnon | @bingads
How is voice search different from text search?
More
Mobile
More
Local
More
Conversational
31. #HeroConf | @jmgagnon | @bingads
Question phrases are more likely to be voice search
Who What When Where Why How Total
Growth in Question Phrases Year over Year
Search Engine Watch, Jason Tabeling, “How Will Voice Search
Impact A Search Marketer’s World?” December 2014
32. #HeroConf | @jmgagnon | @bingads
Question phrases = voice search degree of intent
Where
When
How
What/whoInterested
Ready to act
33. #HeroConf | @jmgagnon | @bingads
Does successful query length differ for voice search?
Microsoft internal data April 2015
1 Words 2 Words 3 Words 4 Words 5 Words 6 Words 7 Words 8 Words 9 Words 10+
Words
Text
Longer searches: More words!
34. #HeroConf | @jmgagnon | @bingads
Search queries from Speech have more words than from text
Microsoft internal data April 2015
1 Words 2 Words 3 Words 4 Words 5 Words 6 Words 7 Words 8 Words 9 Words 10+
Words
Speech Text
Text searches more
concentrated around 1-3 words
Speech searches
longer through the tail
35. #HeroConf | @jmgagnon | @bingads#HeroConf | @jmgagnon | @bingads
More clicks via personal assistant go to brand names
30
30 of the top clicks went
to brand names
Microsoft internal data April 2015
36. #HeroConf | @jmgagnon | @bingads
Take action: Adapt PPC campaigns for voice search
1. Use question words in your keywords to drive to
intent, and negative them if they don’t apply to
your offering
2. Hit the sweet spot of 3 words in your keyword sets
to capture more voice searches
3. Bid on your own brand terms
37. #HeroConf | @jmgagnon | @bingads
Agenda
Chegg: A case study
Voice search
Research:
Brand-term bidding
Conversion path
Remarketing theory
38. #HeroConf | @jmgagnon | @bingads
Study: Brand term bidding
research from Bing Ads
Should SEM campaigns bid on their
own brand terms?
39. #HeroConf | @jmgagnon | @bingads
Methodology of the study
January – March 2014
impressions
50M
Financial services vertical
#HeroConf | @jmgagnon | @bingads
40. #HeroConf | @jmgagnon | @bingads
Brand term bidding + SEO delivers more clicks
56%
Organic only
Microsoft internal data April 2015
41. #HeroConf | @jmgagnon | @bingads
32% more clicks with paid + organic
56%
Organic onlyPaid in ML1 + organic
88%
Microsoft internal data April 2015
42. #HeroConf | @jmgagnon | @bingads
32% more clicks with paid + organic
56%
Organic onlyPaid in ML1 + organic
88%
Microsoft internal data April 2015
43. #HeroConf | @jmgagnon | @bingads
32% more clicks with paid + organic
56%
Organic onlyPaid in ML1 + organic
88%
32% gain
Microsoft internal data April 2015
44. #HeroConf | @jmgagnon | @bingads
But do we cannibalize SEO?
56%
Organic onlyPaid in ML1 + organic
38%
Microsoft internal data April 2015
45. #HeroConf | @jmgagnon | @bingads
Some paid clicks you may have received organically
56%
Organic onlyPaid in ML1 + organic
38%
18%
overlap
Microsoft internal data April 2015
46. #HeroConf | @jmgagnon | @bingads
But in general, many more clicks
56%
Organic onlyPaid in ML1 + organic
38%
50%
18%
overlap
32% gain
Microsoft internal data April 2015
47. #HeroConf | @jmgagnon | @bingads
Bidding on your own brand terms = fewer clicks to competitors
Organic onlyPaid in ML1 + organic
Competitor’s share: 7%
Competitor’s share: 18%
Microsoft internal data April 2015
48. #HeroConf | @jmgagnon | @bingads
Navigational brand terms vs. search intent brand terms
Facebook
No Brand Ad
AT&T Wireless
No Brand Ad
AT&T
No Brand Ad
Victoria’s Secret
No Brand Ad80%
60%
40%
20%
0%
100%
Brand Organic
Brand Ad
Competitor Organic
Competitor Ad
Microsoft internal data April 2015
49. #HeroConf | @jmgagnon | @bingads
Brand term bidding has huge benefits
Keep competition under
control – keep 2x the clicks instead
of letting your competitor get them
More traffic –
36% more
Test offers and landing pages.
No control over organic landing page
means no way to test meaningfully.
#HeroConf | @jmgagnon | @bingads
50. #HeroConf | @jmgagnon | @bingads
Does a PC search stay on a PC?
No click query
Clicked query
Engaged clicked query
Converted query
Microsoft internal data April 2015
handyman little elm tx
51. #HeroConf | @jmgagnon | @bingads
The conversion path is long and winding
A Nikon digital camera search
nikon d3200
bundle
nikon d3200
bundle
nikon d3200
bundle deals
nikon
d5000
nikon d5100 black
friday deals
nikon d5100
black friday deals
nikon d5100
dslr camera
nikon d5100 dslr
camera nn2 lens
bundle
nikon d5100
dslr camera
nn2 lens
bundle deals
nikon d5300
deals usa
nikon
d7100
nikon d5100
body only
deals
11-21-14
18:29 10:58 11:17 11:18 14:26 14:27 10:11 10:13 10:1820:5612:07 10:18
11-27-14 11-28-14
No click query
Clicked query
Good clicked query
Converted query
Microsoft internal data April 2015
BLACK FRIDAYTHANKSGIVING
57. #HeroConf | @jmgagnon | @bingads
Key Takeaways:
Chegg: Data hiding in plain sight. Use it
to your advantage across channels.
Voice: As Voice search grows use clues
from voice search for smarter targeting
Research: Entering a more complex
world full of data. Creative solutions like
offsite tagging are ripe.
Read this article: http://www.nytimes.com/2015/03/13/upshot/free-throw-distraction-the-best-fans-in-the-ncaa.html?_r=0&abt=0002&abg=0
Anybody here get into March Madness last month? Do we have basketball fans here? Well, I love watching the games and because I’m a total nerd I was watching the free-throw distraction thing and I started wondering if it was effective. I like stats like that. I want data. So I went looking and it turns out I’m not the only one who wants to know things like this. The New York Times did a fantastic article about the free-throw distractions and whether or not they’re effective.
Could also make a gif out of the first bit of this video: http://www.bing.com/videos/search?q=free+throw+distraction+gif&FORM=VIRE11#view=detail&mid=B5E746DDD9BAA9427CEAB5E746DDD9BAA9427CEA
The success of all those waving arms and big faces on sticks is nominal – the average team changes the opposing team’s free-throw percent by only a few percentage points.
Unless you’re Arizona State. They win close to 2 points in every game as a result of their distraction shenanigans. You see that asterisk next to their name?
If you can’t read that from where you’re sitting, I’ll tell you what it says. “Since implementing the Curtain of Distraction.” This story just got more interesting. The Curtain of Distraction? How can a curtain be responsible for such a big difference in success?
This is what I found. This is the Curtain of Distraction. You’ll note that you can’t see what’s behind it. And that is the whole point.
http://www.nytimes.com/2015/02/14/upshot/how-arizona-state-reinvented-free-throw-distraction.html?_r=1&abt=0002&abg=1
The Curtain of Distraction works! But I started thinking – the success can’t possibly be because of the distractions. The distractions are not what catches the shooters off guard – everybody knows the distraction is there! Visiting teams know exactly what to expect. They could certainly prepare themselves for the coming distraction – and these players are pros. They know how to ignore a distraction. I’d have to suggest that the reason for the Curtain’s success isn’t the distraction itself – it’s the unexpected.
Nobody knows what’s going to pop out of the curtain, and that element of the unexpected is intriguing. One time it might be Elvis signing a song. Another time it’s a shark. This unexpectedness draws the shooter to glance at what’s going on. And usually, it makes him smile. Arizona State holds the keys to the unexpected, and that’s how they win.
If you hold the keys to the unexpected, you can make it work for you. As marketers, we get to dig into context and take advantage of the unexpected as in the case of the contact lenses or the dental care.
The teams that play against Arizona State don’t get a chance to hold the keys to the unexpected, and that is how Arizona State gets the advantage.
Today we’re going to talk about how one advertiser, Chegg, saved 30% on their search budget by using an unexpected strategy – one that I talked about here a year ago. We’ll talk about how voice search is going to (and may already be) impact your paid search campaigns, and we’ll touch on some recent research we’ve done at Bing, the results of which may surprise you.
I’m John Gagnon, a true search nerd from Bing. I’m happy to be here today talking to you about the future as we know it, or, put another way, winning with the unexpected.
A year ago, a couple of digital marketing guys from Chegg sat in this audience and listened to me talk about a strategic way to plan your campaigns, using PowerMap to visualize timing as it rolls out across the country. Chegg is, so far, completely unique in that it’s working to capture the entire student lifecycle. They have programs that provide scholarship and college information to high school seniors, they do text-book rentals for colleges, and tutoring, and then they connect graduates with internship opportunities. They even offer classes for continuing education, for people who already have a degree or want to keep learning.
You might remember the PowerMap demo from last year. We looked at the question of school start dates – because they’re different everywhere. And if you’re running a search campaign for back-to-school, you can save a lot of money by targeting by geography and date.
We combined the school start dates with what we know about when folks will start shopping for back-to-school.
We tried to visualize this with a basic Excel chart.
We tried to visualize this with a basic Excel chart.
And we even combined schools and dates and school district population into one chart.
But until we were able to visualize it with PowerMap, we couldn’t make sense of the data in order to make a plan.
We pulled all the data into Excel….
And used PowerMap to visualize the first day of school across the country.
The guys from Chegg had the same reaction you’re having: “How can I use this awesomeness?” They went back to their office and this is what they did: First they pulled the top 2000 schools who drive the most revenue in text-book rentals. Then they put those schools into 3 buckets by start-date. <click> August 18, August 25 and September 1.
This is an easy way of looking at the bid schedule. The week before school starts, bids increase by 15% because students already know what their classes are and want to get ahead of the game. The week that school starts, bids increase by 20% - now things are on fire. The week after, bids stay up by 10% because students are still getting their books as the semester continues.
They created a bid shift schedule that would give their text-book ads more visibility at critical times across the calendar. Chegg repeated this rolling bid-up strategy for their second big rush of the year, when winter term starts around the first of January.
Chegg says, “I wish we could do this again!” And that’s one of the sad things about winning this way – an unexpected victory doesn’t come along every day. Otherwise it wouldn’t be unexpected!
Now we’re going to talk about voice search, and the impact this will or won’t have on our work as paid search marketers.
First let’s get a good view of the voice search landscape. Who’s using it? As expected, the younger generations have embraced voice search more than others, but isn’t it a little surprising that voice search is actually being used by a pretty large percentage across demographics. I would’ve expected it to be much less than this.
http://www.emarketer.com/Article/Personal-Assistants-Mobile-Users-Service/1011449/1
Apparently, we’re lost at all times, no matter our age. Asking for directions is one of the top uses for voice search. This is supported, hilariously, by our data from Cortana queries….
…The top voice query for Cortana is “Where am I?”
Now that we understand a bit about the voice search landscape, let’s talk about how voice search differs from text search.
Voice search is more mobile, which is a strong signal right there. We already know that our mobile audience is in a very different mid-set than our desktop searchers. Voice search is more local and more conversational. What does it mean to be more conversational? Here’s one idea…
The use of question words – who, what, when, where, why and how make a search more conversational. If you’re typing in a query, you might type, “Trevor Noah”, but if you were speaking, you’d certainly say, “Who is Trevor Noah?” (He’s the guy who’s going to replace Jon Stewart on The Daily Show.) Jason Tabeling, at Rosetta, published a small study in Search Engine Watch in December that gives us a great view into how to think about voice search. He looked at the increase in these question words in queries year over year, and it maps with the growth in voice search. Why do you think “what” isn’t seeing as big of an increase? [get audience to participate?] This is because “what” is about a noun – and we generally already know “what” things are around us.
http://searchenginewatch.com/sew/how-to/2383498/how-will-voice-search-impact-a-search-marketers-world
As search marketers, we look at this and we want to understand how can this impact our digital campaigns? How can we use this to get ahead? What if we broke down the question words by their degree of intent? If someone is asking what or who, how likely are they to be buying something, or close to buying something? Not very. It’s only as we get to the words “when” and “where” that we really drill into action. Someone who searches for “Home Depot” is in a very different place than someone who queries“Where is Home Depot.” One might be looking for price comparisons or a phone number – but “where is Home Depot” is pretty clear in its intent – and would be worth a higher bid.
The second way that voice search differs from text search is in query length. Because Bing is a data powerhouse, we pulled data for query length for Cortana searches and compared that with query length for general text searches. What you can see is that the successful voice searches, the ones that get the most volume, impressions and clicks, are the ones with 3 words in the keyword or query. For text-based searches, the successful query length is just two words long. This was a little unexpected – I thought the query length for voice searches would be significantly longer than for text-based searches.
The second way that voice search differs from text search is in query length. Because Bing is a data powerhouse, we pulled data for query length for Cortana searches and compared that with query length for general text searches. What you can see is that the successful voice searches, the ones that get the most volume, impressions and clicks, are the ones with 3 words in the keyword or query. For text-based searches, the successful query length is just two words long. This was a little unexpected – I thought the query length for voice searches would be significantly longer than for text-based searches.
Finally, another way that voice searches differ from text searches: searchers using personal assistants click on brand names. This adds to an already-compelling argument that we have for encouraging brand-term bidding in paid search.
Let’s talk about remarketing in paid search. Many of you know that Bing is rolling out remarketing right now, and this is a great time to explore some remarketing theory, some strategy that can help you think a little differently about remarketing in paid search.
All Brand Terms Are Not Created Equal
We know that advertisers benefit when they invest in their own brand terms, but how much they benefit varies by type of search query. We can think about search terms on a scale from the most "navigational" (for example "facebook.com," "ebay.com"), where users are looking for a specific destination, to more general "search intent" searches, where a specific brand is top of mind (for example "Xerox," "Victoria’s Secret"), but a user may really be looking for a product type or category. In the former case, competitors get few clicks no matter if the brand owner advertises or not. In the latter case, competitors can potentially capture a huge proportion of another advertiser’s brand clicks. But for most advertisers, even navigational searches can send some potential customers to your competitors, and many brands should consider investing in these brand terms as well, especially if the lifetime value of a customer is high.
In the above chart (showing normalized click rates), a navigational query like "Facebook" generates few clicks for competitors regardless of a brand owners ads. The brand term "AT&T Wireless" leans more toward the "navigational" side of the spectrum than "AT&T," meaning that most people searching for the former are looking to log into their wireless account, while many people searching for the latter are looking to sign up for a service or make a purchase. And if the brand owner is not advertising in this case, they lose more than half the total clicks, and competitors capture more than one-third of total clicks. The more extreme example is "Victoria’s Secret," where the brand owner loses almost two-thirds of their clicks if they do not advertise, and competitors capture more than half the total clicks on the page.
Imagine a home improvement store with owned pixels [not sure what the right language is for this] across other sites like a do-it-yourself idea center and a paint store. Suddenly, the combinations for action have exploded. If you’re that home improvement store, you’re putting these qualified visitors into a list that adjusts the copy for painting tips and paint colors. You could also EXCLUDE those visitors for an ad that focuses on kitchen appliances, saving money by not showing ads that your audience isn’t interested in.
This kind of remarketing strategy is just an idea right now – but somebody who has the means to develop these relationships with other companies is going to significantly shift the way remarketing in paid search works.
Nobody knows what’s going to pop out of the curtain, and that element of the unexpected is intriguing. One time it might be Elvis signing a song. Another time it’s a shark. This unexpectedness draws the shooter to glance at what’s going on. And usually, it makes him smile. Arizona State holds the keys to the unexpected, and that’s how they win.