Stephen E. Arnold: Online Pricing – Every Angle Except One – Affordable Effective Search

IO Impotency
Stephen E. Arnold

Online Pricing: Well, Who Knows?

I read “Doubling SaaS Revenue by Changing the Pricing Model.” If you are involved in charging for online pricing, you will want to work through this write up by the founder of Kalzumeus Software. What makes the article valuable is its real world data. The article describes a “highly configurable pricing plan” with a low-cost option model which, based on the author’s analysis, is inefficient. The revenue friction is that low-ball customers cost more to support than informed customers. The “free” crowd will find the Kalzumeus analysis disturbing.

The path to pricing happiness according to the write up is via a hybrid approach. The trick is to use a tiered pricing plan with options. Listen to the Kalzumeus argument. Note: we tidied up the grammar for clarity.

The advice … was that Server Density switches to a SaaS pricing model with 3 or 4 tiers segmented loosely by usage, and break with the linear charging. The advantages [of this approach are]:


Trivial to buy for non-technical stakeholders: name the plans correctly and they won’t even need to count servers to do things correctly. (“We’re an enterprise! Of course we need the Enterprise plan!”)

Predictable pricing. You know that no matter what the system admins do this month, you’re likely to end up paying the same amount.

Fewer decisions. Rather than needing to do capacity planning, gather data internally, and then use a custom-built web application to determine your pricing, you can just read the grid and make a decision in 30 seconds.

More alignment with business goals. Unless you own a hosting company, “number of servers owned” is not a metric your CEO cares about. It only tends to weakly proxy revenue. Yes, in general, a company with 10 servers tends to have more commercial success than a company with 1 server, but there are plenty of single-server companies with 8 figures of revenue.

The Kalzumeus article includes some real world data and information about the “pull” of the case example’s “highly configurable pricing plan” and the revised “tiered plan.” The net of the numbers in my opinion is that charging more steers commercial and serious customers to a service level more in line with the customers’ cost goals. The flip side, in my opinion, is more important; that is, a business or serious customer is likely to be a prospect for for-fee services and support. Support and service are the handmaidens of profitability for some online businesses. Free sounds so darned good that the cheerleading makes it difficult to concentrate on the economics of support for the free user segment.

There are some interesting challenges in pricing. The write up does not dig into the upsides and downsides of options. Search and content processing vendors have put themselves in a position which some whiz kids think are flexible, peril free, and magnetic. In short, some search and content processing vendors do not have a pricing method which works in a reliable manner.

Let’s look at a few real life examples and then take a no holds barred look at what works and what does not work.

First, consider the appliance vendors. Search appliances have been around a long time. Google rolled out its first appliance in 2004. When it appeared, Thunderstone and Index Engines were selling appliances. I have not been able to determine if Thunderstone developed the first search appliance. Search appliance pricing is based on the taxi meter method of charging. Now vendors mask the taxi meter model in a number of clever ways, but once the “limit” of the appliance is reached, the customer can buy or license another appliance. Appliances are expensive to use when the volume of data grows rapidly. You can check this effect by contacting an appliance vendor and requesting pricing for units which can handle one million documents and 100 million documents. If you don’t fancy phone time with a sales representative, you can get prices for the Google Search Appliance from GSA Advantage. The prices go through the roof when one buys post sale service and support, clustering, and fail over. The taxi meter pricing model means that once the system gets running, licensees pay to process more documents. In my experience, most customers with large amounts of textual information to process do not like taxi meters.

Second, consider the open source model. Most open source search and content processing vendors offer a free or lower cost package. The licensee can use the free or low cost product and purchase training, support, service, and engineering services on an as needed basis. The open source vendors have a pricing advantage because for a new project, the no cost or low cost package eliminates administrative hassles and allow engineers already on the payroll to handle the implementation and other chores. However, search and content processing are expensive systems to deploy in an enterprise. Appliance vendors and Microsoft are quick to point out that basic search is not a big deal. Unfortunately search and content processing are very big deals. Smaller firms do not like expensive consulting services.

Third, is the bundle pricing method. Enterprise vendors like the dominant enterprise players such as IBM, Microsoft, and Oracle want “lock in.” The idea is that once a big company standardizes on IBM, Microsoft, or Oracle systems, that client will not be too eager to jettison existing IBM, Microsoft, and Oracle systems. These companies are not alone in the pursuit of “lock in.” The reason is that once staff has certification in Oracle Database, IBM DB2, or Microsoft SQL Server, management finds that shifting to a different approach becomes a consultants’ treasure chest. Certified staff wants to stick with what they know even if there are options available. Amazon and Google want to disinter mediate traditional information technology structures to break through the log jam against change. And bundling? A client familiar with Windows and Word could get an option to embrace SharePoint, stick with client access licenses for Microsoft’s productivity applications, and get some financial inducements to implement Fast Search. How do you price a bundle? Microsoft has a pricing advantage and can use add ins, options, and discounts to cement its products in an organization. The method works for cloud and on premises products. Small and mid-sized customers can feel like herded cattle when exploring the bundling option.

Fourth, some search vendors offer price lists. These vary widely from vendor to vendor. However, most of the deals are shrouded in mystery. Coming up with a price quote for a large government funded project induces anxiety. My experience in getting and evaluating bids for government search projects has taught me to be prepared for many phone calls, inquiries, and questions. Figuring out what a search system costs when the bidder must conform to the line items in a statement of work is a tough job. Quite a bit of brain power goes into making it difficult to get a “firm number” for the cost of a system. Variables, options, “if then” statements, and the vagaries of government funding almost guarantee opacity and overruns. In short, the search and content processing vendors don’t have a “fixed price”. The vendors price to match available dollars. Some customers are turned off by complicated and option riddled pricing schemes.

There are other pricing options; for example, the deferred license fee for certain nonprofit prospects, discounts based on the prospect’s past or future purchasing behavior, and hybrid models. The point is that pricing for search and content processing systems is highly variable, often confusing, and designed to obfuscate the fully loaded costs of a find ability system.

The trick to growing revenue in search and content processing boils down to getting customers and selling high-margin services. The old fashioned, license fee works for some vendors; for example, low profile dtSearch and the basic Thunderstone search appliance. For other vendors, pricing is less important than the package of support, training, and engineering services the customer buys.

The reason search is shifting to a service has more to do with revenue than with cost savings for the customer. Search and content processing vendors are packaging “solutions” for verticals, rolling out ready-to-run solutions for financial services or customer support applications, and repositioning their technology to swizzle away from once lucrative market sectors to more promising markets. Think search morphing into big data or analytics for examples.

The problem is that few vendors can provide a simple, clear answer to the question, “What does your search system cost?” The pricing problem has been around since the days of STAIRS III and Personal Library Software. Pricing issues seem to be a permanent feature of the search and content processing landscape.

Stephen E Arnold, August 28, 2012