# Spindle AI implementation cost: Agentforce analytics, Tableau Next and Data 360 Spindle AI joined Agentforce to work on agent observability, ROI forecasting and scenario modelling. There is no Spindle licence to buy, so the cost sits in Agentforce, Tableau Next and Data 360, and the project is mostly about agreeing a baseline before anyone forecasts a return. **Published:** 2026-07-28 **Canonical:** https://synconai.com/spindle-ai-salesforce-implementation-cost **Author:** SynconAI Architecture Team ## In short Spindle AI is not sold as its own product. Salesforce said it would join Agentforce to work on agent observability, ROI forecasting and optimisation, so you price the stack it feeds: Agentforce Flex Credits at US$500 per 100,000, Tableau Next from US$40 per user per month, and Data 360 credits at US$500 per 100,000, as at October 2026. Services cost follows your baseline and data model. ## Sources - **Salesforce Signs Definitive Agreement to Acquire Spindle AI** (retrieved 2025-11-21) https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-spindle-ai/ > Salesforce completed its acquisition of Spindle AI on November 21, 2025. - **Salesforce Analytics Pricing** (retrieved 2026-10-11) https://www.salesforce.com/analytics/pricing/ > The full power of the Tableau Next platform is now included in Core, Advanced, and Max editions for Sales, Service, and Industry Clouds. - **Agentforce Pricing** (retrieved 2026-10-11) https://www.salesforce.com/agentforce/pricing/ > Flex Credits are a new unit of payment for Agentforce that align cost to the business value your AI agents create. - **Data 360 Pricing** (retrieved 2026-10-11) https://www.salesforce.com/data/pricing/ > Data 360 pricing scales with your usage. ## What to take away - There is no Spindle AI SKU. The announcement placed it in Agentforce's Agent Observability and Self-Improvement pillar, so you price Agentforce and the analytics around it. - Tableau Next starts at US$40 per user per month billed annually, and Salesforce says it is now included in Core, Advanced and Max editions of Sales, Service and Industry Clouds. - A forecast of agent ROI is only as good as the baseline. Measure handle time, cost per case or conversion before the agent goes live, or there is nothing to compare against. - Scenario models for pricing, packaging and sales mix need finance in the room. Budget their time, and treat model outputs as inputs to a decision, not the decision. ## The guide Spindle AI built software that let business teams model decisions before making them: what happens to revenue if we change packaging, shift the sales mix, or reprice a tier. Salesforce completed the purchase on 21 November 2025, and the announcement set out where the team would go. After the deal closed, Spindle AI would join Agentforce to work on the "Agent Observability and Self-Improvement pillar for Agentforce 360", with custom agentic analytics, ROI forecasting and continuous optimisation named as the focus. The announcement also described what Spindle did before the deal: its platform let business teams "simulate the impact of complex, data-driven decisions, like pricing and packaging", along with go-to-market and sales-mix changes. Those two sentences point in slightly different directions, and a buyer should notice. One is about measuring and improving your agents. The other is about modelling commercial decisions. As at October 2026 we could not find a Salesforce price list or product page that sells Spindle AI by name, so this guide prices the products that do exist and that the work would run on: Agentforce, Tableau Next and Data 360. It then sets out the two kinds of project a buyer is likely to mean when they ask about Spindle, because they cost differently. ## What you pay Salesforce today All figures are US dollar list prices as at October 2026 from Salesforce's own pages, with Australian dollar figures where the page shows them. | Product | Published list price | AUD | Billing basis | Source | | --- | --- | --- | --- | --- | | Agentforce Flex Credits | US$500 per 100,000 credits | AU$700 | Consumption | [Agentforce pricing](https://www.salesforce.com/agentforce/pricing/) | | Agentforce add-ons (Sales, Service, Field Service) | US$125 per user per month | Not shown | Per user | Agentforce pricing | | Tableau Next | Starting at US$40 per user per month | AU$56 | Billed annually | [Analytics pricing](https://www.salesforce.com/analytics/pricing/) | | Tableau Cloud | Starting at US$15 per user per month | AU$21 | Billed annually | Analytics pricing | | CRM Analytics | Starting at US$140 per user per month | AU$196 | Billed annually | Analytics pricing | | Data 360 Flex Credits | US$500 per 100,000 credits | Not shown | Consumption | [Data 360 pricing](https://www.salesforce.com/data/pricing/) | | Data 360 Profiles | US$240 per 1,000 profiles per year | Not shown | Annual | Data 360 pricing | Two points on that table save money. First, the analytics page now says: "The full power of the Tableau Next platform is now included in Core, Advanced, and Max editions for Sales, Service, and Industry Clouds." If you are on one of those editions, Tableau Next may already be on your contract, and a partner who prices it as a new purchase has not checked. Second, the Flex Credits on the Agentforce and Data 360 pages are the same unit. The Data 360 page says you can move credits between Data 360 and Agentforce, so estimate both workloads against one pool rather than buying two. The Agentforce pricing page defines the unit as follows: "Flex Credits are a new unit of payment for Agentforce that align cost to the business value your AI agents create." That phrase, business value, is exactly what an ROI forecast tries to prove. The analytics project is the one that tells you whether the credits are earning their keep. ## Two projects people mean by "Spindle" **Project A: measuring and improving your agents.** You already run Agentforce, or are about to, and you want to know whether it is working. Which subagents resolve the request, which hand off, how many actions a typical conversation takes, what each outcome costs in credits, and what it saved. This is the observability side the announcement named. The data comes from agent sessions, Case and Opportunity outcomes, and consumption reported in Digital Wallet. **Project B: modelling commercial decisions.** Finance, sales operations or product marketing want to test pricing, packaging, territory or sales-mix changes before making them. The data comes from opportunities, quotes, orders, product and price tables, and often from systems outside Salesforce. This is the scenario-modelling side Spindle was known for before the acquisition. Both run on similar plumbing, but they need different people, different data and different tests of success. A quote that blends them into one "AI analytics" line will be hard to hold to account. ## What an implementation is made of **1. Decide the question and the decision.** Write down the decision the analysis will inform: renew or expand an agent, change a price, move a product into a bundle. If nobody can name the decision, the project will produce dashboards nobody acts on. **2. Establish the baseline.** For Project A, measure the current handle time, cost per Case, resolution rate or conversion rate before the agent changes them. For Project B, reconcile the revenue, margin and volume figures to the finance ledger. Without a reconciled baseline, any forecast is a guess with a chart on it. **3. Data model and semantics.** Build the data model in Data 360 where cross-system data is needed, and define metrics once in Tableau Semantics so that "resolved", "cost per Case" and "net revenue" mean the same thing on every dashboard. This package is usually the largest, and it is where disagreements between teams surface. **4. Agent telemetry.** For Project A, capture which actions ran, which subagent handled the request, the outcome and the credit cost. Connect that to the business outcome on the record. Without that join, you can count actions but not value. **5. Scenario models.** For Project B, build the models: price elasticity assumptions, mix rules, constraints from capacity or contracts. Agree with finance which assumptions are inputs a person sets and which the model estimates. **6. Dashboards and Tableau Next delivery.** Put the outputs where decisions are made: in Tableau Next inside Sales or Service, in Slack, or in a review pack. Keep the number of views small. **7. Validation.** Back-test forecasts against a period you already know the answer to. A model that cannot explain last quarter should not set next quarter's prices. **8. Operating rhythm.** Decide who reviews the numbers, how often, and what triggers a change. Self-improvement without a person accountable for accepting changes is drift. ## A worked estimate for measuring one agent Arithmetic on the public prices helps keep Project A in proportion. Suppose a service agent handles 20,000 conversations a month and averages three actions each. That is 60,000 actions, which at 20 Flex Credits per action is 1.2 million credits a month, or about US$6,000 at the published list rate of US$500 per 100,000. Over a year, the agent's own consumption is in the order of US$72,000 at list, before any discount or included credits. That is the figure the ROI analysis has to beat, and it gives you a sense of scale for the measurement project. Spending heavily on analytics to prove the value of a modest agent is poor value. Spending nothing, and renewing a large credit commitment on faith, is worse. The right budget for measurement sits somewhere in between, and it should be set after you know the size of the consumption it is measuring. Notice also what the estimate does not include: the Data 360 credits used to ingest session and outcome data and to query it for dashboards. Those come out of the same credit unit, so a busy observability dashboard competes with the agent for the same pool. Ask the partner to show both draws side by side. The same logic applies in reverse for Project B. A pricing model that informs a change worth a meaningful share of annual revenue can justify far more effort than a dashboard about agent handle time. Scale the analytics spend to the size of the decision, not to the novelty of the tool. ## Risks particular to forecasting value Three risks recur in this kind of project. The first is attribution: when an agent goes live at the same time as a new knowledge base, a staffing change or a seasonal dip, the forecast will happily credit the agent with all of it. Build a comparison group or a staged rollout if you can. The second is metric drift: definitions agreed in the first month get quietly changed by a later report, and the trend line breaks. The semantic layer exists to stop that. The third is false precision: a scenario model that outputs revenue to the dollar invites more confidence than its assumptions deserve. Ask for ranges and for the assumptions behind them. ## What drives the effort up or down **Whether the baseline exists.** If your service team has measured handle time and cost per Case for a year, Project A starts quickly. If it has not, the first weeks are spent building the measurement, and the agent may already have changed the numbers you wanted to compare against. **Number of source systems.** Project B nearly always needs data from the finance system, billing platform or data warehouse. Each source adds ingestion, reconciliation and Data 360 credit consumption. **Metric disagreements.** If sales and finance define revenue differently, the semantic layer will expose it. Resolving that is valuable, but it is also time, and it should be scoped rather than discovered. **Edition position.** If you are on Core, Advanced or Max, Tableau Next may already be included. If you are on an older edition or Pro Suite, analytics licences are an additional purchase, and the choice between Tableau Next, Tableau Cloud and CRM Analytics should be made deliberately. **Appetite for automation.** "Continuous optimisation" sounds attractive. Letting a model change agent instructions or prices without a human approval step is a governance decision, and building the controls for it adds cost. ## Getting quotes you can compare Tell every bidder which project you mean, A or B, or both in sequence. Give them your edition, the agents you run or plan, the systems holding the data, and the decision you need to make. Ask each partner to show how they will establish the baseline and how long it needs to run before results are meaningful. Ask them to name the metrics they will define in the semantic layer and who signs them off. Ask for a Data 360 credit estimate with the assumptions written down, including query volume, which is often forgotten. Ask how they will validate a forecast before anyone uses it. Ask Salesforce for the licence position separately. Check whether Tableau Next is already included in your edition, whether your existing Flex Credit pool covers the extra consumption, and whether any part of Agentforce's observability features is already on your contract. ## Costs that are commonly left out - **Data 360 query credits** from dashboards that refresh often or are used heavily. - **Finance and operations time** to agree definitions and validate models. - **Historical data loads** to give models enough history, which can be the largest single consumption event. - **Additional analytics licences** for users outside the editions that include Tableau Next. - **Governance design** for any automated change the models recommend. - **Ongoing model maintenance** as products, prices and agents change. - **Training** for the people who must read scenario outputs and know their limits. ## Where the other Spindle AI questions live The licence question on its own is on [Spindle AI licence cost](/spindle-ai-salesforce-licence-cost). How to scope the method, with the baseline and trace defined before any figure, is on [Spindle AI and Agentforce](/spindle-ai-agentforce). The acquisition is covered on [what happened to Spindle AI](/what-happened-to-spindle-ai), and the connector and write path are on [integrating Spindle AI with Salesforce](/integrating-spindle-ai-with-salesforce). If you used Spindle AI before the acquisition, [should I migrate from Spindle AI](/should-i-migrate-from-spindle-ai) covers the choices. When you are ready to brief a team, [hire a Spindle AI Salesforce partner](/hire-spindle-ai-salesforce-partner) lists the questions to put to them. For a budget holder: check what your edition already includes, price the extra Flex Credits from the public pages, and fund the baseline and the semantic layer properly. An ROI forecast for agents is useful only if it is built on numbers your finance team already trusts. ## Questions ### Can I buy Spindle AI from Salesforce? Not as a named product, as far as we could find as at October 2026. Salesforce said Spindle AI would join Agentforce to work on agent observability, ROI forecasting and optimisation. You buy Agentforce and the analytics around it. ### Is Tableau Next already included in my Salesforce edition? Salesforce's analytics pricing page says Tableau Next is now included in Core, Advanced and Max editions for Sales, Service and Industry Clouds. Outside those, it starts at US$40 per user per month billed annually. ### Do I need Data 360 for agent ROI analysis? Usually, if the outcome data lives in more than one system. Data 360 Flex Credits are US$500 per 100,000 and can be moved between Data 360 and Agentforce, so estimate both workloads against one pool. ### What is the most important work package? The baseline. Measure handle time, cost per case or conversion before the agent goes live, and reconcile commercial figures to finance before modelling scenarios. Without it, a forecast has nothing to be compared against. ### Should a model be allowed to change prices or agent instructions on its own? Only with controls you have designed on purpose. Most organisations should keep a person approving changes, at least at first. Building automated change safely is a separate governance cost. --- SynconAI is a Salesforce Select Partner and an OpenAI Select Partner, delivering architect-led Salesforce and AI programmes across the United States and Australia. Related service: Agentforce consulting (https://synconai.com/salesforce-agentforce-consulting).