Amazon aims to counter discontent over last year’s sale with offers of TVs, toys and meetings with celebrities.
10 questions to ask a marketing vendor.
The rise and spread of robust computing and analysis has been a double-edged sword. On the one hand, it’s now possible to measure the effectiveness of marketing initiatives in real time. This promises to help retailers get the best return on their online marketing investments, and avoid throwing good money after bad. On the other hand, the internal transparency created by response data places a great deal of pressure on marketers to improve results.
To aggravate matters, the world of multi-channel marketing becomes more complex every month. Instead of crafting a strategy for a single interactive channel, marketers must now develop strategies to push messages through multiple channels such as banner ads on web sites, e-commerce site promotions, social media and ads embedded in search and e-mail. In each instance, marketers have to understand how customers use these entry points to get to their products. They must react nimbly, swapping offerings in and out depending on traffic flow and strategic objectives. They must continually analyze data and adjust spending and targeting accordingly. The winners in this new environment will be those who can figure out not simply how to respond to a changing marketplace, but also how to create a business ecosystem that continually adapts, learns and improves.
Cutting through complexity
Sounds complicated. Fortunately, there are plenty of marketing services organizations willing and able to help companies optimize their marketing strategies. Marketers can choose from among strategic interactive agencies, multivariate testing firms, web interaction and personalization managers and decision engines, to name a few. In the end, all these approaches share certain characteristics. Each leverages data to assist in making better marketing decisions. Each dynamically controls content to deliver more effective messaging. And each provides reports on the effectiveness of the measures. But, as the standard warning goes, results may vary.
Providers of marketing services today differ in approach, tactics, the scope of services they offer, and ultimately, in the outcomes they produce. Few offer clients the ability to not just improve their marketing results, but also to create and install systems that will transform the company into a learning organization, one that can make sense-and more effective use-of the deluge of data generated. In order to work intelligently and effectively with marketing solutions providers, marketers must first figure out whether the provider possesses the right capabilities and fit. Simply asking for answers to these ten questions can go a long way to providing the information and perspective needed to make a decision.
1) A/B, multivariate, or audience?
This isn’t a trick question. Rather, it aims to seek out what type of methodology a provider uses. A/B testing is the simplest form of optimization, in which one sets up a control (say, a one-page checkout window) and then tests another option against it (say, a checkout process in three windows) to see if it delivers a statistically significant improvement. Such tests are easy to manage and implement, but they are time consuming, not particularly sophisticated, and generally offer the opportunity only for slow, incremental improvements.
Multivariate tests employ more sophisticated algorithms to run multiple tests in parallel. Instead of A/B, think: ACDE/BDGH/AGDC, etc. A marketer using multivariate tests may try out several calls-to-action, price points, content messages, headlines, and images simultaneously in order to produce a new winning control in less time. By conducting a small number of tests and analyzing results, analysts can isolate which factors work best in conjunction with one another. Multivariate works well for the minority of marketers who are seeking a quick, one-size-fits-all solution.
Audience segmentation takes a fundamentally different approach. A/B and multivariate tests may provide guidance on which offers were best, on average. But audiences aren’t average; they differ in their tastes, habits and buying behavior. Audience segmentation seeks to produce a winner for each user that sees an ad, hits the web site, or receives an e-mail. Credit card companies and offline direct marketers have employed such tactics for decades, sending platinum cards to one group, low-interest-rate teasers to another, or balance transfer options to still another. Technology allows interactive marketers to offer much more sophisticated and effective versions of this tactic. Model-based solutions can deliver different messages to different segments, continually measure the responses, and alter text or images subtly as conditions warrant.
2) Can you deal with constraints and flexible attribution?
Many solutions are off-the-shelf offerings. Just follow the instructions and let the technology work its magic. Based on stated preferences and the mandates of a program, a collaborative filtering system will spur a site to offer a certain book at a certain price to a certain group of browsers. If only the marketplace were so simple. What if inventory is running low on a particular title? Or what if a publisher offers special incentives to anyone who can sell 100 copies of a particular book in 24 hours? A market optimization provider should be able to afford clients the ability to work under constraints and flexible attributions. For example, it should be able to dictate that a certain offer needs to be shown no more than 50,000 times, or a minimum of 10,000.
3) Product, People, or Both?
Some marketing services firms deliver tools to do this testing and optimization that in fact are power tools. It typically takes a good amount of domain expertise and statistical background to deploy them effectively. And marketers don’t always have the resources or staff to handle them. Other providers function more like old-fashioned advertising agencies, essentially delivering people who try to improve marketing processes. This approach may provide temporary results, but doesn’t leave any technological system behind it that allows for continuous improvement.
An effective provider should offer a combination of optimization and analytics, of products and people: a product that functions as a strong, malleable and effective platform capable of learning and improving; and people who understand business needs and best practices, who can devise overall strategy and execute it, and then train clients to use the system themselves or remain engaged.
4) Can you cross channels?