We had the opportunity to attend the Dreamforce 2016, held in San Francisco, from October 4th till 7th 2016. As usual, it was a massive affair, Salesforce claims over 160k attendees. Noticeable from last year, our conversation with exhibiting partners was more positive than last year, so Salesforce has done something right. 


Take a look at my musings on the event here: (if the video doesn’t show up, check here)

No time to watch – here is the 2 slide condensation – overall and for platform (if the slide doesn’t show up, check here):
Want to read on? Here you go: 


Salesforce joins the AI frenzy – No conference in 2016 without AI, so Salesforce made no difference, Einstein (like most other announcements) leaked weeks before Dreamforce. The main message is that Einstein works out of the box, no data scientist required, and of course works on all Salesforce data. But the examples were almost all based on scoring and recommendations, capabilities that Salesforce had already, or just acquired and are really the conservative bottom of AI. E.g. scoring does not pass the muster to be AI. But it works and that’s what matters. Basically Salesforce has put all predictive analytics (or I call it ‘true’ analytics – see here) under the Einstein marketing umbrella. And now needs to figure out more on what the platform is, how data will be aggregated, where that all will run, what it means for privacy and security etc. etc. [Salesforce points out that everything starts with trust at Salesforce, and that Salesforce is committed to not share data across customer boundaries. Did you catch that as an attendee / watcher? I missed it.]. But a fair start that would have looked better with less marketing hype, but this is Salesforce. (Read below what Salesforce does to enable developers to build AI applications, that is interesting).

Commerce Cloud – Salesforce has a proven track record to help CRM specialists, the humans. It’s e-commerce capabilities were less developed and the recent acquisition of Demandware has changed that picture (see here for my colleagues take on the acquisition). that. Salesforce store / e-commerce capabilities are now more beefed up and legitimate than before, so time for Salesforce customers to look at the offering. Commerce Cloud certainly is the cloud with the most product in Salesforce’s cloud arsenal.

Another Quip at productivity – Another acquisition – Quip – has been integrated with Salesforce and tries to solve the separation of business applications and productivity applications – a long, practically forever challenge for business users. With a document collaboration model that brings together both structured, unstructured and document information, including the usual side bar for collaboration / notes it certainly is a plausible approach that Salesforce customers should check out.

Salesforce unshackles itself from the past – Salesforce is one of the SaaS pioneers, and has built on a very successful platform – Force – but it is showing its age since some time and it was not designed and built for the cloud age. Since years Salesforce is dancing a limbo between products built on force (e.g. Sales Cloud and Service Cloud) vs. those built on Heroku (e.g. Marketing Cloud) (and Heroku runs on Amazon’s AWS for the less technically inclined readers). Dancing is a great activity, but not the best for an enterprise software vendor, but given the success of the force platform - not an easy thing to address and solve. But Salesforce has made a substantial commitment to AWS (see our analysis here) and at Dreamforce said for the first time that force applications will run on AWS. Likewise, developers can be building force applications will be able to move their DEV ant TST environments to AWS. Both are key developments that allow the conjecture that at some point the Salesforce platforms may again be truly combined, at the moment it looks like on AWS Cloud. And that is a key takeaway, as life gets better for all players in the Salesforce ecosystem, customers, partners and Salesforce itself.


Events have arrived – The Force platform on an object model level showed its age as it could not model modern entities such as e.g. Events. Now with a new Events object being created as a first class Salesforce object, this gap from the past gets addressed. And it is key for the future of Einstein – as Machine Learning (and maybe later AI) cannot work on just ‘business data’ alone, they need to model events that resulted from business activities and know if an outcome is positive – or not so positive. So a key step in the Salesforce entity / object model. Unfortunately, it does not find itself on the platform side, too – where due to the acquisitions that Salesforce has made, there is a large variety on design how to model events and outcomes. It would be good to see Salesforce providing a common event model on the Heroku side of the landscape, too – we should see if that will happen. But for now a good move. 

Force to AWS – Mentioned already above, probably the most important takeaways for customers. When Salesforce succeeds in running the force apps (mostly Sales Cloud and Service Cloud) on AWS, it will change its need to invest into, build, operate and maintain data centers. A very important move that should ultimately allow Salesforce to put more R&D on product development, but also help customers with more local instances of Salesforce, which helps both with performance and data residency requirements. Allowing developers to develop in force on AWS is another key step here. We will be watching.

PredictionIO comes to Heroku – This was actually the most important announcement for Einstein going forward. More developers to build machine learning applications with the help of the PredictionIO capabilities. Salesforce needs more AI applications and empowering developers, who in general are keen on this and hear from all tool makers how they can build Machine Learning / AI aps is important. Didn’t have a chance to see how easy Salesforce is making this, but this will be key for Einstein success going forward.

Kafka runs on Heroku – A consequence and productization of last year’s IoT Cloud announcement, but an important step to ingest data from other sources into Heroku, which becomes more and more the future of Salesforce Application Development – both internally and for the developer community. Let’s hope Salesforce did not have to do too many tweaks to Kafka for it to run on Heroku, so that Salesforce developers can take advantage of the high speed of innovation we see on the open source side. Future uptakes of Kafka in Heroku will write that story.


A good Dreamforce for Salesforce. Einstein took center stage, and certainly was a ‘must do’ announcement for Salesforce – given the announcements by the competition in the same area. But the real advancements, with short and longer term effect are on the platform side – the move to Heroku / AWS. Something that could have happened a few years ago now finally seems to take place, we don’t know what made Salesforce execs wave the start flag – but realizations like at customers that common data center based infrastructures do not scale for next generation applications may have happened in a Salesforce office, too. So it is good news for Salesforce customers that a potential end of the dual infrastructure reality is in sight, a move to AWS / Heroku can only help customers. More modern development, DevOps tools, best practices like CI / CD become tangible now, very important to make sure Salesforce not only remains an attractive and viable platform, but also to make its SaaS applications more innovative.

On the concern side, time is of the essence for Salesforce. Key competitors in e.g. CRM as well as platform have made that move already and do not have to pay the tax for a dual platform in all phases of the software development lifecycle. The problem is that Salesforce product development speed is not moving breathtaking fast, e.g. an indicator is that the Sales Cloud has not fully moved to the new Lightning user experience / platform, that was announced …. at Dreamforce… 2014. Granted, Sales Cloud is a massive product, with lots of screens, but 2+ years for a new UI is clearly in negative record territory for any leading SaaS vendor. On the flipside Salesforce is making the right moves to make developers faster and more efficient – when it happens for (paying) platform developers and ISVs, it also happens for Salesforce in house developers. How much will be built in the next 12 months will give an answer to that.

Overall a good Dreamforce for Salesforce, granted platform is not the topic 160k attendees will get excited, but platform messages work and are important (see Salesforce ‘cousin’ Workday making the platform theme the one theme of their keynote – last week) – even for end user audiences. I can’t wait the Salesforce marketing skills and spend to be unleashed for platform – maybe at Dreamforce 2017? We will keep watching.

Find more coverage on the Constellation Research website here and checkout my magazine on Flipboard and my YouTube channel here.

Business Research Themes