Dreamforce 2018: Einstein Analytics Becomes the Default Analytics Layer

We're writing this from the Moscone Center floor on day one of Dreamforce 2018, and the short version is this: if you're a Sales Cloud Enterprise or Unlimited customer and you haven't thought seriously about Einstein Analytics until now, that calculus just changed. The bundling announcement that came through this morning is the kind of shift that takes a few days to fully absorb, and we wanted to get our initial read down while the conversations are still fresh.

Some context before we get into the details. Einstein Analytics has only been called Einstein Analytics since April of this year. A lot of our clients still say "Wave" out of habit, and frankly so do some of us when we're not paying attention. The rebrand was more than cosmetic — it signaled a repositioning from a standalone BI product into something Salesforce wanted to treat as a native intelligence layer across the platform. What we're seeing announced at Dreamforce this week is the next logical step in that repositioning.

The Bundling Shift and What It Actually Means

The headline is straightforward: Einstein Analytics is now bundled with Sales Cloud Enterprise and Unlimited editions. If you're an existing customer at those tiers, you're getting access. If you've been evaluating it as a separate line item on a proposal and wondering whether the budget will come through, the conversation is now different.

We want to be careful here, though, because "bundled" does not mean "unlimited." Our licensing team has already been in three separate conversations today trying to clarify exactly what the included allotment looks like versus what you'd need to purchase on top. The short answer is that the bundled access covers a meaningful subset of the full platform — enough to get real work done, but structured in a way that larger deployments will still need to think carefully about user provisioning and dataset volume. We'll publish a more detailed licensing breakdown once we have the full spec in writing rather than relying on booth conversations.

What this does change immediately is the organizational dynamics around adoption. For the past several years, Einstein Analytics (or Wave, depending on when you first encountered it) required a separate purchase decision and a separate champion in the buying process. That friction was real. We've seen solid implementations stall for months because the analytics budget sat in a different cost center than the Salesforce renewal. Bundling removes that gate. If your CRM admin has access to Einstein Analytics as part of what your company already pays for, the question shifts from "should we buy this" to "why aren't we using this yet," and that's a much easier internal conversation to have.

Einstein Discovery Going GA: The Part We're Most Interested In

The announcement we're spending more time on than the bundling news is Einstein Discovery reaching general availability. This has been in various stages of preview and pilot for a while now, and the GA designation matters because it means we can actually commit to it in production implementations without the usual caveats.

For readers who haven't worked with Einstein Discovery directly, the core idea is statistical story generation — you point it at a dataset and it surfaces the factors that are most predictive of a given outcome, along with narrative explanations in plain language. The "story" metaphor is intentional. The output isn't a dashboard or a report; it's a structured analysis that tells you what's driving the metric you care about, what the most significant variables are, and what the model suggests you do about it.

The story automation capabilities announced here are the part that closes a workflow gap we've been working around for a while. Previously, running a new story required manual initiation. With automated story refreshes, you can tie the analysis cycle to your data pipeline cadence and have updated recommendations surfaced without someone having to kick it off manually. In practice this means the operations team that checks their sales pipeline numbers every Monday morning can have an updated Discovery story waiting for them rather than looking at analysis that's a week stale.

We do want to flag a realistic limitation here. Einstein Discovery works best when you have clean, consistent historical data across a meaningful time horizon. We've seen it produce impressively coherent stories on datasets with two or three years of CRM history. We've also seen it produce confident-sounding but misleading output on thin or inconsistently populated data. The GA release doesn't change that underlying reality. The model is only as good as what you feed it, and the narrative framing can sometimes make the output feel more authoritative than the data warrants. Our standard practice is to do a data quality audit before we point Discovery at anything and treat the first story output as a hypothesis to interrogate rather than a recommendation to act on.

Lens-Only Datasets: A Smaller Announcement With Real Operational Value

This one didn't get much stage time, but lens-only datasets are something our implementation team has been asking for in some form for a while. The basic problem they solve: there are analytical use cases where you want users to be able to explore data in a lens — slicing, filtering, pivoting — without that data being used as a source for dashboards that other users might take action on.

The governance use case is straightforward. You might have a dataset that contains data your analysts are allowed to explore but that you don't want embedded in a published dashboard that gets broadly distributed. Previously the controls around this were coarser than you'd want. Lens-only datasets give you a cleaner way to enforce that boundary at the dataset level rather than trying to manage it through dashboard permissions after the fact.

From an implementation standpoint, this also helps with the "exploratory sandbox" pattern we use with a lot of clients. We build a set of curated, governed dashboards for operational use, and separately we want analysts to have room to explore the underlying data without the risk that an experimental lens accidentally ends up in a production context. Having a formal dataset type that supports this separation makes the architecture cleaner.

The Licensing Math: Run It Before You Celebrate

We want to be direct about something, because we've already seen a few clients this week assume that bundled means their analytics costs just dropped to zero.

The bundled access is valuable and the barrier-to-entry reduction is real. But if you have a deployment of any meaningful scale, you need to sit down and actually run the numbers before you restructure your analytics strategy around it. Key questions to work through:

How many of your users actually need full Einstein Analytics access versus read-only? The licensing model distinguishes between viewers and builders, and the bundled allotment structure reflects that. If you're expecting to give interactive exploration access to a large portion of your sales org, understand what that costs before you make commitments.

What does your dataset volume look like? Einstein Analytics pricing has historically included constraints around the volume of data you can bring into the platform. If you're working with large external data sources or complex multi-object data flows, understand where the limits are.

What's your Einstein Discovery usage pattern? Discovery stories are analytically intensive. Understand how automated story refreshes affect your consumption before you set up schedules that run more frequently than you actually need.

None of this is a reason to slow down. The bundling announcement is genuinely good news for the ecosystem and we expect it to accelerate adoption significantly. But "bundled" is a starting point for the licensing conversation, not the end of it.

How This Changes Our Implementation Playbook

For clients we're currently working with who are on Sales Cloud Enterprise or Unlimited, our immediate recommendation is to start treating Einstein Analytics as a first-class citizen in the CRM architecture conversation rather than a parallel track. That means:

Data model decisions made in the CRM should account for analytics consumption. This is something we advocate for already, but bundled access makes it easier to justify the investment in cleaner field naming, consistent picklist values, and historical data hygiene when the analytics layer is included in what you're already paying for.

Start with a lens-based discovery phase before you build dashboards. One pattern we've had good results with is giving a small group of power users lens access to key datasets before we build any fixed dashboard content. What they actually explore and find useful tells you more about what should be in a production dashboard than any amount of requirements gathering does.

Take Einstein Discovery seriously for your top three or four operational metrics. You don't need to boil the ocean. Pick the metrics that drive the most actual decisions in your sales org — pipeline conversion, average deal size, time to close, whatever matters most — and run Discovery stories against them. Even if you only use them internally to validate or challenge your existing assumptions, the exercise is valuable.

The Broader Picture: What Comes After This

We'll be honest about what we don't know. The analytics market is moving fast. There are capable independent BI tools that integrate with Salesforce data, and the right architecture for any given client still depends on what they're already invested in, what their data team looks like, and what use cases they're actually trying to serve.

What Salesforce is clearly doing with this bundling move is establishing Einstein Analytics as the default answer to the question "how do I analyze my Salesforce data" rather than one option among several. For pure Salesforce shops, that makes a lot of sense. For organizations that have significant data infrastructure outside of Salesforce, the picture is more nuanced.

The Tableau acquisition hasn't happened. There's no roadmap visibility into what the broader Salesforce analytics story looks like two or three years out. What we have right now is a significantly more accessible version of a product that, when implemented thoughtfully, produces real business value. That's enough to work with.

We'll be posting a follow-up piece later this week once we've had the licensing conversations we need to have and sat through the technical sessions on the Einstein Discovery architecture. If you're at Dreamforce and want to compare notes, find us in the Einstein Analytics theater sessions this afternoon.

This post was written during Dreamforce 2018 and reflects what we knew as of the announcements on September 25. Licensing details are based on booth conversations and pre-release documentation and may be clarified as official documentation is published.