Managing Advertising Inventory Yield
General Principles and Cases from Inventale Business Practice
History of the Question
The term “yield management” first appeared in the air-industry, where carriers had to solve the same task for many years – set a valid price for each passenger seat on each flight during the year, taking into account seasonal demand for different flight directions. The primary goal – revenue maximization – was reached by selecting optimal prices while retaining high sell-out of seats.
Gradually this process was enriched by sophisticated algorithms and approaches and the term “yield management” firmly entered into the terminology of those business-directions where a similar task was handled – the selling of “perishable” goods or services (eg., hotel rooms), the demand of which changes seasonally, at the best price possible.
The term is equally popular in the online-advertising sphere. However, from a publishers’ perspective the task is even more complicated, since apart from fluctuations of demand (i.e. advertisers’ interest in the inventory), they have to deal with changing offers, as the number of impressions on a website depends on the activity of its visitors, and this figure is very unsteady and requires careful forecasting.
Ad Inventory Yield Management
Thus for a digital publisher, yield management is a complex of measures for maximizing revenue from their ad inventory, achieved through a balance between optimal cost of each impression and high level of sell-out (Sell-Through Rate or STR).
This process is frequently described as: selling the right product to the right customer at the right time for the right cost.
The Right Product
The term “product” (sometimes “package”) in online advertising – is a set expression, indicating a combination of inventory and targeting most valuable for a customer. For instance, a combination “Geo: London; Interests: Cars; Income: High” can be of utmost interest to advertisers dealing with the automotive industry in the British capital. During the last few years major publishers have tended to sell their inventory pointwise: narrow slices, optimal for a specific ad campaign, to allow the raising of the total value of the inventory.
For “packing” audience and inventory properly one needs to know their buyers and fluctuating market situation. An ideal tool for this is a system which allows the creation and monitoring of different products, the analysis of their historical STR, delivery, revenue and average cost, as well as building a forecast by these metrics, influencing inventory yield.
There is indeed no one ideal formula or a universal solution for packaging inventory and audience into products, and the success lies in the hands of an expert. In other words, having the right software at hand is a necessary but insufficient factor for working with products.
The Right Customer
One of the popular tendencies in yield management is a use of systems providing competitive auctions of impressions, when each impression goes to the buyer, offering the highest price. Here we are talking about such interrelated technologies as RTB and header bidding.
While advantages are apparent, the drawback of this approach is that a real-time auction cannot guarantee a stable purchase of impressions at the highest possible price. For this reason, many publishers prefer direct deals when selling their premium inventory while trying to squeeze the maximum of the remnant on ad exchanges.
In any case, the critical factor of yield management is finding a right balance between different channels, through which inventory is sold. Managing your network as a single system allows choosing of the right sales channel and a way of monetization for each impression – such an approach is often called “holistic yield management”.
The Right Time
To support a balance between supply and demand one needs to know their impressions volume; which part of it is already guaranteed to advertisers and which can still be sold.
A regular forecast by the entire network allows minimizing the number of unsold impressions, while detailed forecast on potential new campaign delivery prevents the situation when the number of booked impressions exceeds the number of impressions that will be delivered in reality. Moreover, an ability to forecast the volume of available inventory by a narrow targeting slice for several weeks ahead often helps to choose the optimal time period when delivery of a specified ad campaign will go smoothly.
To avoid unsuccessful delivery it is essential to monitor campaigns in real time. However, not only their current status is essential, but also a forecast, considering both inventory availability fluctuations and competing campaigns.
In general, the correct choice of delivery time allows reaching optimal STR and avoiding both campaign under-delivery and an excess of unutilized impressions.
The Right Cost
As mentioned previously, to choose the best offered price publishers often use different auction or impressions allocation systems. Yet apart from the possibility to select from available offers, making steps for raising the total inventory value is also important. The price a buyer is ready to propose for an impression depends on user data provided by the publisher. Gender, age, income, interests, place of residence, actual user location allow the advertiser to address their ads to the targeted audience, and the cost of these impressions will be higher. The context in which the ad is shown also plays an important role: page topic, video contents, the genre of the watched film influence the effectiveness of the accompanying adverts and consequently their cost. Besides, a value of impressions grows if the publisher can prove that the ads have been shown and not just downloaded into the part of the screen not visible to the user.
However, the more complex the structure of segments and targeting is, the more advanced tools are to be used. Just gathering data on website users is not enough – for a successful ad campaign it is important to know how the audience will be distributed in the future by each segment and slice (as well as their overlapping), the number of unique users, visiting each particular webpage, and how frequency capping (FC) or campaign priority will influence the volumes of available inventory.
Thus high cost of ad inventory can be reached through usage of valuable data on each impression (audience, context, quality) in conjunction with specialized systems for managing this data (reporting, analysis, forecasting).
How Our Clients Are Reaching High Revenue
What tools can help in solving the tasks described above? Below are several cases from our experience in helping our clients with challenges they were facing.
Raising the Effectiveness of Direct Deals
Practically all Inventale’s clients are major publishers that for supporting direct sales have to answer advertisers’ and ad agencies’ queries every day. For effective inventory commercialization, sales teams are always searching for answers to questions like:
- What is the total volume of available impressions for a specified time period?
- What part of these impressions falls into a specified overlapping of dates, targeting (eg., geo and KV), inventory and settings (FC, priority and the like)?
- Inside this overlapping what part of impressions are unique users?
- What running campaigns will be competing for the impressions with the new campaign? What number of impressions will they be competing for? What part of the impressions will be available for sure?
- How available and booked impressions will be distributed among websites and segments within the overlapping?
In addition to everyday sales processes, a timely forecast for more extended time periods is essential for our clients for building a long-term sales strategy. Having analyzed our clients’ sales teams queries; we implemented a specialized module Inventale Forecasting. The system has an easy-to-use interface that allows configuring a query of any complexity. One inputs any parameters of an ad campaign or the whole network (inventory, targeting, settings, dates), and as a result the publisher gets a forecast on available and booked impressions and their distribution within the query, overlapping with other campaigns, unique users and other important metrics.
Knowing the volume of available inventory our clients keep their sell-out at the highest level (usually the figures increase by 10-20% after our clients start using the system), and detailed knowledge of the inventory considerably raises its value for pointwise (individual, direct) sales (by 10-15%).
For some of our clients, it was essential to get the forecast not via Inventale interface, but via their booking system used by the advertisers for buying their inventory. For these clients, we implemented access to the forecast via Inventale API. This approach allowed supporting the sales process within a single interface, when buyers got a forecast on the inventory they were interested in, with an ability to book it immediately.A Point of Control Over Your Ad Campaigns
Some our clients were facing the same problem – their ad server did not provide any handy tools for monitoring running ad campaigns.
The clients needed an easy-to-use tool, bringing together all the ad campaigns from the network, showing both their current status and an accurate forecast reflecting all traffic fluctuations, competing campaigns (including the ones, not started yet) and not tracked by standard external events algorithms (such as local holidays or special events).
To solve this problem, Inventale has developed Inventale Monitoring module, a consolidated screen to monitor ad campaigns delivery processes. We have tried to deliver both the detailed information on all campaigns for thorough analysis (via UI and excel), and simple tools for tackling daily tasks. For instance, filtering by a level of urgency and degree of delay from the goal allows traffickers to quickly get the list of campaigns requiring the most attention today, and in one click go to the ad server for further actions.
A Holistic Approach to Yield
One of our clients applied to us recently with the following case. Their sales team get a salary plan each year that is divided into calendar months and the total available inventory (loosely speaking, each website has its revenue goal for each month). For effective plan execution, the sales team need answers to such questions as:
- in what part of the network and on what days there are still many available impressions, and where there is a shortage
- what is the average impression cost in each inventory package
- how each metric, influencing final revenue, is distributed over time
- how much money each package and the network as a whole will bring, and how much more one can expect in the nearest future
- how realistic is the sales plan both for the network and each sales package
- how effectively this or that inventory is used
The solution, offered by Inventale, is a dashboard providing key revenue metrics (profit, delivery, STR, еCPM) for each part of the inventory and for the whole network. The metrics have been visualized in graphs and diagrams, enabling one to monitor inventory monetization dynamics and sales plan execution. We have also added color flags and several experimental metrics, helping to choose an optimal CMP for plan execution.
Another case we are planning to support in the near future is a yield analysis for longer time slots. Users will be able to choose any time period and study each metrics dynamics, as well as compare metrics by days on a graph or via Excel-report.
We want to develop a universal tool for analysis and reporting, giving a very detailed, precise and useful picture on network monetization in the past and the future, and helping to make key decisions in revenue management.
Instead of Conclusion
As practice shows, for raising effectiveness publishers continuously have to develop and look for new technological solutions. However, high-level tasks for raising revenue stay the same: deliver the right product, find the right customers, choose the right time, set the right price.
Based on our experience we can say that many successful publishers achieve high yield through introducing new technologies, but keeping the focus on high-quality analytical data and accurate forecast. After all yield management in the first place is a process of analysis and balanced decision-making.