The main point of the Microsoft event I was at yesterday was around improving travel advertising. We started with a bunch of numbers about how things compare from adCenter to the other players in this space:
Across the board, Nielsen Netratings have found that adCenter convert at a rate 81% higher than the average UK searcher. In particular, they see a conversion rate nearly twice as high as on Google: 5.8% vs 3%.
In the travel market specifically, these numbers change to: 9.4% for Live vs 6.3% for Google.
In addition to higher conversion rates, they see greater engagement (I didn’t want to tell them Avinash’s view that engagement isn’t really that great a metric):
Average time spent on site = 4 minutes for Live vs. < 2.5 for Google. In travel, 11% more time is spent on the destination site vs. the rest of the market.
Apparently, not only do Live searchers engage more with your website and convert better, they also spend more. This was supported by one particular data point - that the average Windows Live customer spent £2k or more online in the past 6 months. Unfortunately, the other side of this coin wasn’t shared - we didn’t get data on average spend of other searchers.
Doing it better
After the ’why use adCenter’ pitch, we got into quite a lot more detail about how to make the campaign more effective:
The account hierarchy is highly important. The advice given is to set up the account as follows:
- Campaigns for main geo areas (e.g. country-level), (splitting generic and brand keywords)
- Ad groups for more targeted geo areas (e.g. city-level)
- Keywords within ad groups very closely targeted
The primary reason for structuring the account in this way is to enable highly-targeted adverts. In addition, it allows better reporting and trend analysis - it is important to be able to split out the effects at highly-targeted levels.
Finally, demographic targeting is both applied and reported at the ad group level. In the same way as allowing detailed general reporting, this means that we can actually extract useful and usable demographic trends / differences between ad groups. By collecting this data about very targeted groups of keywords, it becomes actionable and any differences that we do spot can be targeted in the adverts.
In general strategies for big events should be set up around 90 days before the event - whether this involves having different keywords in the account (e.g. for Christmas) or incrementally bidding more for certain keywords at different times of year. As a guideline:
- Christmas and New Year keywords should be in by September
- January to March is the important time for winter skiing keywords, which means they need to be in place by October
- December needs January-peak keywords in (e.g. package holidays)
- June needs keywords for city breaks (and the associated flights / hotels / car hire) as these are primarily taken in October
Some of the smaller events have less of a lead time:
- January is the peak time for Valentine’s keywords
- March for is the busiest time for booking Easter breaks
Here we should be thinking about the search funnel. The typical example given was:
Information gathering phase:
Generic / seasonal / cost searches (e.g. hotel, easter flight, cheap car hire)
Refinement and filtering:
Destination / modifiers (e.g. flight to athens, business class flight)
Often branded searches
No surprise here. Write great ads.
Although you can have up to 20 ads in an ad group, they recommend having 3-5 ads per ad group to enable auto-optimisation and also allow you to test multiple changes to your creatives at once without diluting the impressions too much (which means it takes too long to gather statistical data).
Some specific examples of creatives that work particularly well in the travel industry are:
- Include your brand in the title of the advert (if you have a distinctive / popular brand name)
- Use reassuring phrases such as: the specialists, experts because many people are nervous about booking travel
- If you are offering budget pricing, end with do you want to save on X?
- Alternatively, try ending the creative with: save now!
The main advice given to improve conversions was:
- Deep linking - help people avoid getting lost by sending them to the exact right place on your website as far as possible (or pre-populate search fields if this isn’t possible)
- Report on demographics - then use this data to bid more for your perfect demographic target
- Day-parting - bid more for favoured times of day / day of week
- Geo-targeting - either only targeting people in certain geographies or bidding more for searchers in those geographies
One question that wasn’t answered particularly well for me was how to analyse the effects of both day-parting and demographic-targeting. It is easy enough to analyse when search volumes are high (and you can probably even tie this in to CTR) but it seems to me to be hard to analyse this in terms of conversion rates or on-site behaviour - at least until we get Gatineau to play with!
I am going to write more at some point about the difficulty of tracking the effects of day-parting, geo-targeting (and this also applies to demographic-targeting now this is offered by adCenter) - but that’s a post for another day. If anyone has any great tips, drop me a line and I will include it in the write-up of the current state of play of tracking this kind of thing.