Are you ready for the future of web advertising?
2012-08-10 04:24 - 1,450 reads
Let me start with a warning. This blog post is going to be dry and boring, but you need to persevere because the subject matter is so darned important.
I am going to explain something called Real Time Bidding. It is a revolutionary new way of buying online advertising, and is replacing the traditional model.
In a nutshell, it involves altering the price paid for each advertising space based on the commercial potential of the person viewing it.
Real Time Bidding has massive implications for your business – in a very, very good way.
Okay, with that stern preable done, let’s begin.
Buying advertising space online has been the same since the internet began. Advertisers buy slots on webpages, paying the website-owner either by duration (one month, a year etc), per impression (eg, £4 per thousand), by clicks, or by conversation (meaning when the customer actually buys something after clicking on the advert).
The problem with this standard approach is that advertisers show the same advert to browsers irrespective of who they are. This is crude and wasteful.
Real Time Bidding is a process which re-invents the purchasing process.
Instead of block-buying advertising spaces by the thousand, companies buy each slot individually via an auction triggered each time a user clicks onto the webpage.
Real Time Bidding explained
When an internet-user navigates to a web-page, the space where an advert is normally shown is blank. An auction is held to fill that space. The auction takes place in a thousandth of a second – so fast the internet-user has no idea an auction ever took place.
The auction is a simple highest-bidder-wins model. When the auction is complete the winner gets to display their ad to the internet-user.
The key is the way the auction functions. Bidders do not have a standard price, rather they each have a formula which creates a different bids from auction to auction.
How much should they bid? The trick is to calculate potential value of the internet-user who will see the advert. This is done by by looking at dozens of pieces of data supplied by the internet-user.
For example, is the internet-user browsing with a Mac or a PC? If the advertiser is selling a luxury good such as a Rolex watch they may only wish to target high-net worth individuals, and therefore will be prepared to pay three times as much for a Mac user than someone who owns an ageing Windows XP machine.
Maybe the advertiser wants to target potential customers in a specific geographic region. The IP address of the user will tell the advertiser where they are located to within a few hundred yards. The advertiser can thus decide to bid less for users who are outside their catchment area, and increase their bids for users located close to their core market.
Now you may ask: “How can advertisers factor these things into an auction which takes place in a thousandth of a second?”
The answer is algorithms. Advertisers create a set of values attached to each data set. The bid process is automated – hence the extraordinary speed of the auction.
A working example will illustrate what is going on.
Let’s take an umbrella company in Leicestershire. This umbrella retailer has a clear picture of the internet user it wants to target with its adverts. So it will state:
- If the user is on an iPad we are willing to pay a premium of 5p per ad, Lion OSX 4p, Windows 7 2p, and if the user is on Windows XP we don’t want to bid at all.
- If the user is located in Leicester we’ll pay a premium of 2p per ad more than if they live in Leicestershire. If they live outside the county of Leicestershire then we won’t bid at all.
These rules are woven into the bidding formula. Each time a web-page is shown an auction is triggered, the formula comes into play and the advertiser places a bid in the auction derived from the formula.
The genius of the system is that advertisers reserve their budget for internet users who really fit their target profile. When a red-hot prospect hoves into view the advertiser can ramp up their campaign, spending more if necessary to capture the attention of the internet-user. When no-hopers turn up then the advertiser can bid low, or not at all, saving money.
Tweaking the process
You might wonder just how much information can be gleaned about an internet user. After all, the user is anonymous (unless logged in, but we’ll assume they are not). Their name, bank balance and purchase history are thus all secret.
This is true, but even anonymous internet-users give away an awful lot of information.
Click here to see what information you are currently broadcasting to the world: ipswift.com. The results might surprise you.
Google allegedly uses 57 signals to personalise your search results even when you are not logged into your Google account.
Each one of these signals can be evaluated to determine the potential value of each internet-user to your business.
It’s raining data
You don’t have to rely on merely signals and data supplied by the user. You can factor a ton of other things into your algorithm. Our umbrella retailer, for example, may wish to advertise more aggressively when it is raining. It can incorporate a data-stream from the Met Office into its its bidding formula. When the heavens open it can increase the amount it is prepared to pay for advert spaces. When the rain stops the adverts automatically die away.
Inventory can be factored in. If sales are slow and a firm has too much of one product line, the advertising budget for that product line can be increased automatically.
Any data stream can be added to the algorithm.
Before you start think Oh dear, this all sounds horribly complex, let me say – Yes, you are right. It is. Which is why Real Time Bidding firms offer their expertise. Firms such as Infectious Media employ dozens of mathematicians who do the algorithm work for you. They look at your sales data and will create a bidding formula which, if done right, will ensure your adverts have the maximum return-on-investment.
Customise each advert
The way Real Time Bidding auctions off advertising space is clever.
But what is also remarkable is the ability to customise every advert which gets shown.
Traditional advertising shows the same image and text to every internet-user.
Not with Real Time Bidding.
By using template adverts with interchangeable components it is possible to tailor-make each and every advert to match the needs of the individual internet-user.
Suppose our umbrella seller wants to flog umbrellas in Leicestershire, Northamptonshire and Bedfordshire. By looking at the IP address of the internet user the advertiser can tell where the user lives. So it can change the text on its ad to read: “The finest umbrellas in Northamptonshire” to the Northants users, “Keeping Bedfordshire dry since 1967” to the denizens of that county.
The text and image of each advert can be customised on any of the data points used by the advertiser (“iPad users will love our range of Apple-themed umbrellas”; “It’s raining in Bedford, so why not buy an umbrella from us” etc).
There is no limit to the amount of customisation.
How big is it going to be?
Real Time Bidding is going to be massive. Facebook is launching its Real Time Bidding platform, the Facebook Exchange, or FBX. This will allow advertisers to make use of the staggering wealth of data it collects about users. The customisation potential is huge. Facebook’s $100bn flotation was predicated on the popularity FBX will have for advertisers.
Research by The Online Ad Exchange reports that 34 per cent of all online display advertising will be Real Time Bidding by 2017
Real Time Bidding allows you to customise the amount you pay and the advert you show according to the precise nature of every individual web-browser.
It is the future of online advertising, and, despite being tricky to grasp, is most definitely worth knowing about.
Like this post? You may also enjoy: