Until 2005, my career had been in enterprise software. Selling software was simple. We put it in a pretty box and shipped it off with a perpetual license. Then in 2005 I joined Brightcove; we launched a SaaS application for video publishing, and realized marketing SaaS was a whole new game.
We tried a variety of different approaches and learned a bunch through trial and error. Below you’ll find a few insights we picked up.
1. SaaS Products are Easier to Launch and Harder to EOL
In the packaged software world, we spent a lot of time (usually half the release cycle) trying to make sure we wouldn’t have to re-manufacture the CDs or release a patch right after the product was shipped. That made development cycles longer and changed the dynamics of QA.
One of the deceptive things about SaaS systems is that they are relatively easy to launch. Build some software, throw it up a server, and you’re off and running. You don’t have to get it perfect the first time out the door, because you can fix things on the fly. But there’s a rub.
Say the product does ok. You’ve got 5,000 satisfied customers, you’re bringing in single digit millions, but you’re looking at the growth and realizing — this thing is just not going to be as big as you thought, and you have a better idea. If it were packaged software, you wouldn’t care. Adobe still sells HomeSite and the last time anyone touched that code was in 2002.
With SaaS it’s all different. If you decide you don’t want to keep building a product, you still have to maintain the system (and incur those costs), especially if you hope to sell other products to the same customer base. If you discontinue the service, you shut all your customers down disrupting their businesses. More importantly, you violate the unspoken SaaS promise: that you will keep the system running.
Middling SaaS services can become an anchor on your company as you try to figure out how to EOL them without screwing your customer base and brand.
2. SaaS Pricing is Engineered into the Product
As marketer, I care a lot about price, and I’m used to being able to control it independent of the product. If I’m selling my software in box, whenever I want I can change the pricing. But SaaS services have the pricing engineered into the product, since its almost always based on metering usage in some way. That metering and the integration between the metering and your billing system is all engineering work. Marketing is totally dependent on engineering to implement pricing strategies — pricing is a feature of the product.
The fundamental SaaS pricing question is dollar per X, where X (and there could be more than one) is an index of value in the customer’s mind and has some correlation to cost on the vendor side. In some cases this is simple. SalesForce.com is indexed to seats, which works great because every sales rep adds a relatively predictable amount of cost and revenue. What about social media platforms? Where is value being created (surely not per seat on the backend)?
Once you decide how you’re going to index value and how that’s going to translate into pricing, you have an engineering project. When you decide to change it, you have another engineering project that competes with your other priorities. What’s more, if you’ve already signed agreements with customers under the old pricing, you end up maintaining those old pricing models, which adds to engineering and accounting costs.
All this puts more pressure on marketers to get pricing right from the beginning. You have to think ahead more carefully about how to price because once you’ve built it into the project you’ll find your hands tied.
3. Marketing Can’t Just Rely on the Launch
One of the cool things about running a SaaS product line is that you can add new functionality whenever you want. There is a continuous release cycle. But this creates a challenge for the experienced enterprise or packaged software marketer who is used to doing big annual launches.
The traditional marketing cycle is very waterfall. You spend a few months planning, then you spend 8-9 months implementing all your launch activities (new collateral, advertising, etc). Then you do your big launch and put into the market a new “marketing platform” (pricing, messaging, channel strategy, advertising, collateral, etc.) This platform remains largely unchanged; you just build on it for the next year, which frees you to work on the next big launch.
But with a continuous release cycle this approach breaks down. New functionality is getting dribbled out every quarter, so it’s harder to create those big bang product announcements. When should you execute the product review campaign? Do you update all the product collateral every month as a features change? What do you tie announcements too? With everyone going agile you also face the challenge that your competition is constantly changing. Competitive briefs are out of date a few weeks after being published. Slide decks are always being changed and update. It’s easy to feel like you’re stuck on a hamster wheel constantly doing tactical work with no time for strategy.
We never found the perfect solution to this problem at Brightcove, but a few things seemed to work:
a) Disconnect from the Product – There are lots of ways to generate PR and run campaigns that are not tied to product releases. It forces you to think more about stories, trends, and controversy rather than products and features.
b) Stay Digital – Print collateral is dead. Don’t make it. PDFs and web pages get the job done and they are much faster to update. If you have to print, use a service that prints on demand to keep inventory low.
c) Create Your Own Cycle – We built a collateral release cycle that we could manage. At any given time, some stuff is out of date, but the whole team is more efficient with a predictable schedule.
d) Staged Launches – If want to drive product news, pick a launch date, and role up the last 6 months of features into an announcement.
4. The Best Customer Communication Tool is the Product
Marketers are used to depending on email communication to reach their customers, and they miss the new reality that the product user interface is the best way to reach your users. Except in the situations where you price per seat, I believe leaving per seat pricing out encourages individual user registrations and less sharing of login credentials, which gives you much better data about users.
As you design and develop the product think about how it can incorporate a communication system that marketing can manage outside of engineering for delivering everything from notices about new releases, to feature up-sells, to user tutorials that build customer satisfaction and loyalty. Every time a user logs into your product is an opportunity deepen your relationship with them. Start simple, but make sure there is a road map for how communication will work in the product experience.
5. Build for Reliability from Day 1
Reliability is the air that a SaaS company breaths. You take it for granted. But when it’s gone, you suddenly risk death. Agile engineering makes a strong case for only building what you need now. If you don’t need scale, don’t build for it. This is one area I’d challenge.
Engineering teams for SaaS products should be architecting for reliability from day one. Moreover, QA, systems engineering, and operations should have a deep mantra about reliability. There are increasingly well established patterns for building SaaS systems that stay up — don’t ignore them. The coolest features in the world instantly become useless if the system goes down. Business-to-business-to-consumer systems afford very little tolerance for down time because an outage affects your customers’ perception with their customers.
This should caution marketing to balance front end, customer facing features, with time for engineering to work on reliability and performance enhancing features. The cost of SaaS systems going down is very high in terms of customer satisfaction, loyalty, word of mouth, brand reputation, and potentially PR.
6. Analytics Should Power Your Business Decision Making
Why speculate when you can monitor behavior directly in the product? Amazon is legendary for using actual user data about behavior to test and tune their service.
Unfortunately, this is hard to do. Most people underestimate the sophistication, skills and resources you need on your team to make use of data. It’s relatively easy to instrument a SaaS product so you can collect a lot of data. It’s much harder to analyze the data and make it useful.
That means you need to build deep analytical and statistical skills into your team from day one, and make it a fundamental part of how you figure out which features are working and which are not. The data from the product is incredibly powerful if you have the tools and people to turn it into information and insight.
7. Pay Attention to COGS
In packaged software the cost of goods sold (COGS) is very predictable. But in SaaS COGS can quickly get away from you cutting into gross margins. There tends to be a constant battle between keeping COGS low and deploying the hardware and systems engineering you need to maintain your service.
Generally, COGS create an ongoing debate between finance and engineering, but marketing can play a role and should be paying attention because it’s a fundamentally important metric. The entire management team should be involved in decisions that weigh the tradeoffs between costs savings and reliability.
This is the start of a list. I’m sure there will be more. Gartner forecasts that SaaS will reach $11.5 Billion 2011. I’m looking forward to hearing comments and reflections from other executives making sense of how to get SaaS businesses to sing.