Enterprise Metasearch Optimization

We recently wrote and released an ebook covering everything digital executives need to know about running high performance metasearch campaigns. This process gave us a framework for quantifying the metasearch opportunity, and we have been testing with clients and friends in the industry. We describe it as:

Revenue = f(Demand, Core Data, Distribution, Optimization)

What we’re trying to describe here is that total revenue is a function of market demand, core data, distribution, and optimization. We’re really excited about this framework. We’re not done building on it, though. In fact, I asked a good friend of mine – who is also a huge asset to the company due to his discernment – what he thought of this. His response?

“How tied are you to this? It’s not a mathematical equation in the strictest sense…”

This is meant to be directional and to help understand both constraints and opportunities. My good friend is right for now, but our intent is to refine this into a full model. Everything we’ve accomplished so far has been due to chipping away at broad, difficult to solve problems until we uncover a major success or breakthrough.

Even though it’s not strictly mathematical, it has already proven helpful when analyzing current campaigns and for future planning, so we wanted to share it here. The following content is a small excerpt from our new ebook Metasearch at Scale to explain more about the framework.

Demand

Think of demand at the hotel level as that total potential volume for a hotel or set of hotels. While demand can be affected by advertising, branding, and marketing activities, within the context of metasearch it is usually an upper bound of the total revenue you can achieve. This value has virtual constraints, such as the number of users searching on metasearch engines, and physical constraints, such as the location and number of the hotels in your campaigns. The total addressable demand for an advertiser is affected by pricing, inventory, and reputation.

Reputation can have a much greater impact on demand than some advertisers want to admit. If the total demand for a market is 1,000 rooms per night, but one of your properties has a few bad reviews, the total addressable demand can be much lower. There are some guests that simply will not consider your property or properties if their minimum rating and/or density of reviews is not met.

Core Data

Screen Shot 2014-11-11 at 4.52.20 PMBasic hotel information – hotel name, address, phone number, amenities, latitude, longitude, etc. – is usually not at the top of the marketing executive’s mind, and for fair reasons. Most advertisers think their core data is good, or at least good enough. This thinking can cause a lot of money to be left on the table. At Koddi, it hasn’t been uncommon to find data issues with 10-20% of a chain’s properties when onboarding a new client. Bad data can be a blind spot for advertisers because it often leads to silent failures that can add up to significant shortfalls.

Consider the formula shown in the introduction to this section. The potential revenue that you can drive from metasearch is, to this point, a function of the total demand across your property set and your core data. How so? Metasearch engines can only show (good) ads when they can match your property to the their own data set. This match is often done through a clustering of data points. Bad or missing core data can cause your ads not to be seen, or even worse, to be seen for the wrong property.

Core Data also isn’t always a 1 or 0 multiplier; amenities are a good example of where impact can be more nuanced. If your data is not listing “Free Breakfast” as one of your amenities when it is in fact offered, then you are probably not missing out on a significant number of searches. However, missing multiple amenities (or having incorrect ones listed) across hundreds of properties and multiple metasearch engines could cause a measurable effect on volume and performance.

Distribution

The impact that Distribution has on your total metasearch revenue is determined by two sets of factors: accuracy and speed. Accuracy is affected by how well rate, availability, and room data match up to the canonical data source and the overall completeness of data.

Speed is the amount of time it takes for the metasearch engine to consume and reflect the data from its canonical source. Speed is a relative concept here; what’s “fast” for one supplier may be irrelevant to another if that supplier updates rates more or moves inventory more or less quickly.

Measuring distribution efficacy is important to advertisers, and there are a lot of different ways to go about it. Many advertisers crawl publisher partners to pull to parity, and use this data for competitive research. There is a free method to acquiring this data if you participate in Google Hotel Ads, covered in the section on Management Patterns.

Optimization

The final element is also the one that responds the most dramatically and immediately to change and effort. Optimization represents all of the activities that go into the day-to-day management of a campaign. This can include bidding, reporting, analysis, testing, copywriting, and placement negotiation.

There are three major types of cost models for metasearch campaigns:

  1. cost-per-click or CPC, which is self-explanatory;
  2. cost-per-action, commission, or CPA, where a fee or % of booking value is charged; and
  3. hybrid, which is a combination of the two.

These cost models can individually become more complex. In Google’s Hotel Ads, CPCs can be adjusted by many different factors including the cost of the hotel room, the placement of the ad within the search results, the length of the stay, the location of the user, and the type of device a user is using. TripAdvisor and Trivago CPCs vary primarily by the source of traffic, but can also become more complicated by their Bucket system, in which a bid is made by selecting a Bucket that corresponds to a CPC value.

While these options do add complexity, they are also extremely valuable to advertisers. Each attribute described above has a significant impact on the value of the click. The ability to bid by those attributes allows an advertiser to match the value of the click with the cost per click. These bidding mechanisms give advertisers the ability to target users more granularly than they otherwise could.

We’ve included reporting here as a specific consideration because it is often not planned for. Without the right automation in place, reporting can take a significant amount of time. Some teams have dedicated analysts that in actuality do very little analysis, and mostly compile reports from disparate data sources. Reporting quality is made up of the accuracy of data, its granularity, and its efficiency. Reporting that comes in any later than the next day is a liability, because it can delay responding to any issues with campaigns.

Many metasearch engines don’t provide reporting data in as timely a fashion as digital executives are used to, sometimes taking days or even weeks to reconcile cost data at the most granular level. The means and methods for delivery are also dated: in an industry that heavily relies on APIs, only two major metasearch providers (Google and TripAdvisor) provide programmatic access to reporting data.

Optimization also affects the total revenue for a metasearch campaign by impacting the positions and ranks in which ads appear. As a general rule, high rankings drive more traffic. Ranking is typically a factor of the offer – the price for a room relative to competing advertisers – and the bid. On Google, some preference is given to the best offer but it is possible to outrank that offer with a high enough bid. On Trivago and Kayak, position 1 is always given to the lowest price.

Bringing it all together

With an understanding of these levers you can adjust execution as necessary. For example, it is possible to use the data within Koddi to see how effectively you are responding to market demand, how good your core data is, and how complete your distribution is. It is also possible to test different optimization methodologies to and focus in on different key metrics. If you’re wondering if there is additional revenue out there, it is just a matter of scratching down some basic numbers. If your actual revenue is way below your addressable demand, you know exactly where to focus to improve things across all facets of your campaign at one time.

Want to learn more? Download our ebook on metasearch optimization for 21+ pages of great content covering metasearch basics, advanced optimization techniques, how to organize to drive high performance, the future of metasearch, and much more.

Categories
Metasearch