The marketing of performance
Much of the technology I consult about boils down to performance. There are many sub-categories — parallelization, scalability, low latency, interactive response, price/performance, and more. But basically it’s about computers operating faster, within realistic resource constraints.
There are three kinds of benefits performance can offer:
- It can allow you to do things more simply and/or cost-effectively (e.g., with less hardware or less tuning).
- It can allow you to do things better.Examples include:
- Faster-loading web pages for your customers.
- Faster-responding queries for your business analysts.
- Better prices on your algorithmic trading.
- Better analytic results, perhaps from:
- Using more data.
- Running more queries.
- It can allow you to do something that would be impractical otherwise (usually because of expense).
These benefits are easily confused. When a prospect says “I can’t do X with existing technology”, what she really means is often “I can’t afford to do X well enough to matter.” When a vendor says “We make it cheap and easy to do Y”, what prospects hear is commonly “Great! Now we’ll be able to do Y within our resources and budget.”
Given the breadth of the subject, it’s hard to generalize comprehensively about the marketing of performance claims. But my observations include:Â
1. For “cheaper” to be a strong message, you have to be significantly cheaper in TCO (Total Cost of Ownership), not just in system acquisition/floor space/power costs. But raw performance is often not the biggest driver of cost, given:
- The cost and risk of technology adoption in immature markets, when experience and expertise are hard to come by.
- The cost and risk of technology adoption in mature markets, when expectations and switching costs may both be high.
2. Hardware/software purchase/license cost is a directly important performance consideration mainly to two classes of users:
- Those who want to do a lot more of something than they have been doing to date, and quail at the expense unless they change vehicles. But that’s in the “better/more” category of my taxonomy, not the “simpler/cheaper” one.
- Those who want to do something they couldn’t afford before. That goes in the “impractical” section.
I have real trouble thinking of a pure “we out-benchmark the other guys and so we’re cheaper” story that ever has won.
3. But “simpler” is a benefit that should not be overlooked. It speaks to all of operational cost, operational risk, and resource availability. Analytic RDBMS vendors brag about how little tuning their systems require. In both the Hadoop and NoSQL/NewSQL markets, ease of scaled-out cluster management is a major criterion.
4. An important sub-case of “better” is “do lots more”. Scenarios I run across frequently include (and these overlap a lot):
- We want to analyze a lot more data!
- We want to do a lot more analysis on our data.
- We’re hitting a wall with Oracle Standard Edition, and Oracle Enterprise Edition/Exadata cost much more than we want to pay.
5. If you think your story TRULY is “Our performance is so great it makes the otherwise impossible possible,” you’re kidding yourself. First, you have competitors, who also make it possible. Second, if what you’re newly making possible is all that bloody important, then probably people have already been making do to get it done as best they can, even in an inferior way.
Yes, I know there are a few exceptions. I invite you to mention them in the comment thread. 🙂
If you want to say “It can’t be done without us” as part of your marketing flair, be my guest. But please remember that what you’re saying isn’t actually true.
6. Overall, the most fruitful performance-related business-benefit positioning usually straddles “better” and “impractical without us.” For the richer or more sophisticated buyers, you’re “better”. For the laggards, you’re taking them by the hand and leading them to the Promised Land.
7. Actually, the middle layer of the layered messaging model may be more important than the top one. Your “metric” kinds of benefits may be clearer than your business benefit stories.
8. Anyhow, it’s hard to market on performance only, since performance stories are often hard to differentiate from each other. So the rest of your technical benefits may be what sets you apart from your close competitors. As just two examples:
- Multiple generations of data management technologies have been differentiated in large part by their associated development tools. Examples include prerelational DBMS in the early 1980s, relational DBMS in the later 1980s, and CEP/streaming tools in the present era.
- Memory-centric business intelligence tools are rarely differentiated from each other on performance grounds.
Comments
20 Responses to “The marketing of performance”
Leave a Reply
I’d argue that Sybase’s initial market entry, circa 1987 was pretty close to a straight performance, we out-benchmark the other guy message. However, to your point, it was well-wrapped in a higher-level product positioning message which was “an OLTP relational database” as opposed to either slow query-oriented RDBMSs or the fast/OLTP non-relational databases.
Long way of saying that I can’t think of a sustained position every built purely on performance. Market entry strategy? Maybe.
Amen. Everything you said makes sense, including the bit about marketing flair 🙂
There will be spikes in performance metrics as folks move to flash memory based systems and move more data into caches that are accessible faster.
(Read: Better but more expensive, until flash price points move closer to disk prices). But again, everyone has access to those technologies.
So something more is needed than “bigger, better, faster” marketing. And that i believe would be more use case based – finding use cases which customers have a hard time doing today and educating them on adopting the newer technologies.
Seems to me that “time to value” as in “how long it takes to get ______ up and running and available for users is the most effective sales pitch.
Pure run-time metrics (Query A runs faster on platform X than on platform Y) assume a commodity use case. It’s possible for some vendors to establish a short-term advantage when the metrics are well-defined, but other vendors usually catch up.
Radical performance improvements create the ability to do something that is currently impossible. That is the value driver, and effective sellers connect the dots for the customer.
Thomas,
“Time to value” is one effective sales pitch, and ever more important as the speed of competitive response keeps picking up. But to say it’s the single best kind of pitch in all of IT would be quite a stretch, unless you’re defining it so broadly as to include “good UI”.
Dave,
I’d say Sybase fits my model of marketing better performance by straddling the “better” and “can’t do it without us” claims.
Further, as you point out, they illustrate my point that making some kind of a core technical claim is good.
Curt,
Across multiple software vendors where I’ve worked in the past twenty years, “time to value” is the one constant. It’s also the principal driver towards Hadoop, SaaS, Cloud, and many other innovations driving IT today
Regards
TD
Hi Curt-
Excellent post! I really enjoyed reading this. Not only does this help vendors, like us, who do market on performance & the value it can provide operationally, financially and to the business, it provides users with insight into how to navigate the benefits of “performance”.
Couple of points I really liked:
1. “Cheaper” point:
– Cost and risk of technology in immature markets, when experience and expertise is hard to come by. I can’t think of a better example of this than Hadoop, especially expertise like MR programming.
– Cost of mature markets, switching costs are high. A great example here is DI/ETL.
5. “Our performance is so great it makes the otherwise impossible possible.” You invited comments in this thread. 🙂
Check out this post in WSJ/CIO:
http://blogs.wsj.com/cio/2012/04/18/comscores-big-data-push-wins-it-new-clients/?mod=google_news_blog
Using multiple products, including from Syncsort, comScore delivered:
– operational impact by enabling processing times to be slashed to six hours from 48
– financial impact by processing twice as many records – 900 billion! – using the same amount of resources
–Keith
Keith,
Thanks for the kind words!
That said: Your comScore case is a surely an example of “more” or “better”, but it doesn’t sound to me like “newly possible”.
Thomas,
I disagree about “time to value” being the principal driver for Hadoop, especially when the Hadoop is on premises. I’d put cost higher on the list. Ditto “newly possible”.
Further, I think the importance placed upon time to value ebbs and flows. Around the Y2K crunch, it was crucial. But that’s also about the time when BI went from having short time-to-value to taking a lot longer, and the industry kept growing even so.
Curt,
Disagree to your heart’s content! If you listen to folks at Google/Facebook/Whatever talk about why they ventured into Hadoop in the first place, they usually say stuff like “we didn’t have time to create schemas etc etc and we were drowning in data so we had to put it SOMEWHERE”. That sounds to me like “time to value”.
Regards
TD
Thomas,
I haven’t heard Google talk much about its Hadoop use. 🙂
I do grant the importance of dynamic schemas in the world of internet companies.
Curt,
Nice post. I like how you “net-out” what is really the sales driver in the analytics / Hadoop space.
Agree with you on the “time to market” point – people are trying Hadoop to see if there is a cost effective way to manage analytic insights using hadoop that translates into dollar savings / business improvement. Judging by the average sizes of Hadoop deployments people are deploying, it doesn’t seem that they are in a rush to replace existing approaches, but rather exploring for opportunities to improve and gradually adopting.
At the end of the day, it is my feeling buyers make data decisions based on who is selling it, track record, and service levels. The true differentiators is with the team / company vs. the technology. There are enough vendors chasing deals such that there is parity in performance and many other technology factors. Would be interested to hear your insights on this.
Thanks
Joe,
Buyers try to dually optimize product and vendor desirability. At one extreme, only one vendor passes the desirability tests. In many cases, there’s a minimum level of vendor accomplishment, after which product characteristics take over. Or, similarly, there can be a minimum standard for product proof points.
If one simplifies that matter too far, one will be oversimplifying. There are a bunch of different buying-process templates that show up in real life.
[…] First, there aren’t as many jaw-dropping use cases as one might think. For as I wrote about performance, new technology tends to make things better, but not radically so. After all, if its applications […]
[…] First, there aren’t as many jaw-dropping use cases as one might think. For as I wrote about performance, new technology tends to make things better, but not radically so. After all, if its applications […]
[…] Marketing the benefits of performance Categories: Columnar database management, Data warehouse appliances, Data warehousing, Database compression, Exadata, Netezza, Oracle, Vertica Systems Subscribe to our complete feed! […]
[…] modeling is a performance feature, and performance can have several kinds of benefit, which may be summarized as “cheaper”, “better”, and “previously […]
[…] Then it became more about selling “business value” to organizations who needed more convincing about the benefits of great price/performance. […]
[…] views on the marketing of productivity benefits are similar to what I wrote about the marketing of performance:Â There are three kinds of benefits performance can […]
[…] illustrated why that’s so when I wrote about the marketing of both performance and […]