ChatGPT
Claude
PerplexityTwelve people. Four active projects, one of them slipping. The architect who knows the client’s integration is on another engagement for the next three weeks. The one person who writes decent user stories is the same person running the steering committee.
That is the shape of a boutique Salesforce practice, and no methodology document written for a firm of two thousand addresses it. The constraint is not process maturity. It is that your senior people are load-bearing on multiple things at once, and a playbook that assumes role separation is describing someone else’s company.
What changed in 2026 is that the quality of your answer to this now shows up on your Salesforce partner record.
Salesforce replaced 170 badges with 28 competencies, and delivery produces the inputs
Salesforce overhauled its Consulting Partner Program in early March 2026. Salesforce Ben’s report on the change, published 3 March, sets out what actually moved.
Base, Ridge, Crest and Summit collapsed into two tiers. Select covers proven delivery partners meeting clear performance standards. Summit is for top-tier partners aligned to strategic growth. Roughly 170 legacy badges went with the old tiers, replaced by 28 core competencies that take over from Navigator, and two recognition levels now sit on top of those: Accredited for demonstrated capability, Expert for scaled delivery excellence.
Then comes the part a delivery lead should read twice. Each competency is measured by three inputs: certifications, completed projects, and customer satisfaction scores. The practical effect, as Salesforce Ben reads it, is that consultancies will be known for specific AI, product and portfolio competencies instead of general Salesforce knowledge. Salesforce’s stated expectation is that partners investing in specialization and lifecycle execution win larger engagements and see better expansion and renewal rates.
Salesforce has also said FY27 introduces lifecycle incentives, which tie financial reward to driving value at stages across the customer lifecycle instead of concentrating it at the point of sale.
Two of those three scored inputs come straight out of how you run projects. Certifications are a hiring and training decision. Completed projects and customer satisfaction are produced by delivery and by nothing else.
A caveat on precision, because this is where most published summaries of the change overreach. Minimum competency counts, reference requirements and the exact CSAT bar sit behind the partner portal, and Salesforce has not published them openly. Several partner blogs state specific thresholds; none of them cite a source. Treat the structure above as solid, and get the numbers from your own partner record or your partner account manager before you plan against them.
A small firm cannot win on project volume, so satisfaction carries the tier
Follow the arithmetic of that scoring model from a boutique’s position and a strategy falls out of it.
Completed project count is a volume measure. A global integrator closes more Salesforce projects in a quarter than a twelve-person firm closes in three years, and no amount of process improvement changes that ratio. Competing on that input is competing on headcount you do not have.
Customer satisfaction does not work that way. It is a per-engagement score, and a small firm running four projects with senior people on every one has a structural advantage over a large firm running four hundred with mixed seniority. CSAT is the one measured input where being small helps.
Competency coverage sits between the two. Earning it is a concentration decision, not a scale decision, because 28 named competencies reward depth in a few areas over thin presence across all of them. A boutique that picks two clouds and one industry and certifies hard there gets recognized faster than one chasing whatever the last prospect mentioned.
So the playbook has a strategic purpose beyond running projects well. It should be built to produce the two things that are in reach: consistently high satisfaction, and evidence. Which means the question driving every decision below is not whether a practice is elegant, but whether it survives contact with a twelve-person team and leaves a trace afterwards.
The playbook is five artifacts, not a methodology document
Boutiques that try to write a methodology produce a forty-page document nobody opens. The ones that build a working playbook ship a small number of reusable artifacts and enforce them.
Five carry most of the value.
A discovery question set per solution area. Not a generic workshop template, a specific list of what must be established before you quote Service Cloud, or Experience Cloud, or an integration. This is the artifact that stops a junior consultant missing the question that later costs a change request.
A standard story and acceptance criteria format. Including the access conditions a Salesforce build depends on, which generic agile templates leave out entirely. How that format should be constructed, and where generated stories reliably fail, is covered in writing Salesforce user stories and acceptance criteria.
A solution design template with the decisions pre-listed. Objects, sharing model, automation choice, integration pattern, data volumes. The template’s job is to make the omissions visible, because a design doc that forgets the sharing model produces a build that forgets it too.
A test case library keyed to the story format. Reused across clients, because the same object does the same things at most companies, and rewriting UAT scripts from scratch for every engagement is pure waste.
A go-live and handover checklist. This is the artifact that most directly moves CSAT, because the client’s lasting impression of your firm forms in the two weeks after go-live, not during the build.
The discipline that makes these work is boring and absolute: one owner per artifact, a version number, and a rule that improvements made on a project go back into the template within the week. A playbook that only improves when someone has spare time never improves.
Where GetGenerative.ai changes what a small team can produce
The artifact list above has an obvious problem in a twelve-person firm. Producing discovery notes, stories, designs and test cases to a consistent standard is exactly the work that gets compressed when a project runs hot, and it is the work that produces the CSAT and the evidence the partner program now scores.
That is the specific gap the platform addresses. Discovery turns workshop sessions into structured requirements instead of a notebook only one person can read. Design converts those into stories and low-level solution outlines against your own format. Metadata reads the client org, so a design is written against the configuration actually present rather than what the sales call implied. Test generates the case library from the same stories, which is how the UAT pack stops being written the week before UAT.
For a boutique the economics are different from a large firm’s. A GSI absorbs artifact production in a delivery pyramid of junior staff. A boutique has no pyramid, so the same work lands on senior people whose time is the firm’s entire margin. Moving that production off senior hands is not an efficiency gain at the edges, it is the difference between the playbook being followed and the playbook being skipped.
Teams running a practice this way can review plans and credits for consulting teams.
Salesforce now lists FDE partners and AgentExchange as separate routes to visibility
Two distribution routes appeared on Salesforce’s partner page that did not exist for small firms a couple of years ago, and both suit a boutique better than they suit a large integrator.
The first is a dedicated Forward Deployed Engineering partner category. Salesforce describes it as specialized Forward Deployed Engineering partners to power an agentic strategy, and it sits as its own search route alongside the general consulting partner finder. A category with fewer firms in it is a category a small specialist can rank in, which is the opposite of the dynamic on a general partner directory.
The second is AgentExchange, which Salesforce positions as a unified marketplace for agents, apps and capabilities, including skills, plugins and MCP servers. For a boutique holding a truly reusable asset, this turns an internal accelerator into a listed product with its own discovery path.
Salesforce’s own framing is worth reading for what it concedes. On its partner program page it calls the route for implementation firms the Consulting Track, offers reach to 21 million active users and agents across more than 95 countries, and states that 70 percent of Salesforce implementations are led by consulting partners. That figure is the reason the program measures delivery at all: most of what customers experience as Salesforce was built by a partner team. Salesforce also puts AgentExchange app installation at 91 percent of its customers, which is what separates a marketplace listing from a vanity badge.
Neither route rewards breadth. Both reward being the obvious choice for something narrow, which is the positioning a twelve-person firm can actually defend.
Scope control is the lever a small firm fully owns
Every boutique has the same conversation. The client asks for one more thing, it seems small, the relationship matters, and somebody says yes.
The pattern is not a discipline failure. It is that a twelve-person firm has more relationship exposure per client than a large firm does. Losing one client out of four is a quarter of your revenue, so the pressure to accommodate is proportionally larger and the person feeling it is usually the founder.
Three mechanics make this manageable without damaging the relationship.
Separate the decision from the conversation. The person hearing the request should not be the person approving it, even in a firm small enough that both are in the same room. A stated rule that changes go to a weekly review converts an awkward no into a process.
Price the change in days and show the schedule effect alongside the fee. Clients accept scope discipline far better when the conversation is about the go-live date moving than when it is about an invoice.
Keep a visible log of accepted changes. At the end of a project, a client who sees eleven accommodations understands why the final week was tight, and that conversation protects the satisfaction score the partner program now reads.
What this page deliberately does not cover is how to price the work in the first place, which is a separate decision with its own trade-offs, set out in fixed price, time and materials, or outcome-based pricing.
Two people in four roles is the real capacity model
Large-firm delivery assumes distinct roles: engagement manager, business analyst, architect, developer, tester. A boutique assigns those five roles across two or three people, and the playbook has to say how.
Combinations that work share a trait. Keep the architect and the lead developer separate where there is real build volume, because the architect who is also writing Apex stops thinking about the data model at exactly the wrong moment. Let one person hold engagement management and business analysis, since both are client-facing and the handoff between them is where requirements get lost. Never let the person who built something be the only person who tests it.
Failure modes cluster too. A founder holding sales, delivery oversight and the hardest technical problem is the single most common failure mode in a small Salesforce practice, because the sales half is unpredictable and the technical half is on the critical path, and when both spike the project absorbs it.
One practice separates boutiques that scale from ones that stay dependent on a founder: every engagement names a second person who could take it over. Not as a formality, as a person who has attended the client calls and can read the design doc. That is also the only real defense against the bus factor a twelve-person firm carries.
Agents change which artifacts a small team can afford
Agents are entering boutique delivery in two distinct ways, and conflating them produces bad decisions.
The first is agents that help you deliver. Generation of requirements, designs, test cases and documentation, which is the artifact production problem above. The value here is real and immediate for a small firm, because it converts work that previously required senior time into work that requires senior review.
The second is agents you deliver for clients. This is a service line, and the program change makes it strategic, though not in the way most summaries of the change suggest. Salesforce has published no Agentforce requirement for either tier. What Salesforce Ben raised instead is the open question: community feedback indicates Agentforce adoption is thinner than the messaging implies, and what happens to partners without an AI practice remains unconfirmed. Read that as room to build evidence, not a deadline already missed. A boutique with two completed and measured Agentforce engagements still sits in a different competitive position from one with none, and two is reachable at small scale.
Two cautions belong in the plan. Agent work carries consumption cost that moves with usage, so a fixed-price Agentforce engagement with no consumption assumption stated is a margin risk, and the mechanics are set out in what Agentforce actually costs. That cost model is still moving: in late September 2026 Salesforce signalled that successful MCP and direct API calls from registered agents will consume Flex Credits as Headless Platform Interactions, with the multiplier unpublished and 30 days of notice promised before metering starts. Metering is limited to active production orgs, so sandbox and Developer Edition work stays free. A fixed-price agent engagement signed now should name the consumption assumptions it rests on and say what happens when they change. And the first agent engagement should be scoped small and measured, because the case study matters more than the fee, and an unmeasured deployment produces no evidence at all. Salesforce’s own partner page leads with outcome numbers from named firms, pipeline generated and autonomous case resolution rates, which tells you the format the evidence is expected to take.
What to build first if you are starting from nothing
Most boutiques reading this have fragments: a proposal template, somebody’s good discovery questions, a half-finished runbook.
Sequence the gap-filling by what compounds fastest.
Start with the story and acceptance criteria format, because everything downstream keys to it and it is the cheapest artifact to standardize. Add the go-live and handover checklist next, since it has the shortest path to the satisfaction score. Build the discovery question set third, one solution area at a time, drawn from the last five projects rather than invented. Then the design template, then the test library, which is the slowest to accumulate and the most valuable once it has.
Give each artifact an owner and a version, and put a recurring thirty minutes in the calendar to fold project learnings back in. A boutique playbook does not fail because the artifacts were wrong. It fails because nobody was responsible for them after the project that created them ended.
The broader question of which delivery model a firm is actually running, and what each one costs to operate, sits in the comparison of in-house, partner and AI-led delivery, and the failure patterns the playbook exists to prevent are catalogued in why Salesforce implementations fail.
Boutique partners raise six recurring questions about delivery
What should a Salesforce delivery playbook contain?
Five reusable artifacts rather than a methodology document: a discovery question set per solution area, a standard story and acceptance criteria format, a solution design template with the key decisions pre-listed, a test case library keyed to the story format, and a go-live and handover checklist. Each needs a named owner and a version number.
How did the Salesforce partner program change in 2026?
Salesforce overhauled the Consulting Partner Program in early March 2026. The four consulting tiers, Base, Ridge, Crest and Summit, collapsed into Select for proven delivery partners and Summit for partners aligned to strategic growth. Roughly 170 legacy badges were replaced by 28 core competencies that supersede Navigator, with Accredited and Expert recognition levels on top. Each competency is measured on certifications, completed projects and customer satisfaction. Salesforce has not published the specific thresholds, so confirm those against your own partner record.
Why does customer satisfaction matter more for a small Salesforce partner?
Because it is the one measured input where being small is an advantage. Completed project volume favors large firms by definition, but satisfaction is scored per engagement, and a firm putting senior people on every project has a structural edge over one staffing from a pyramid.
How should a boutique Salesforce partner assign delivery roles?
Across two or three people, with three rules. Keep architect and lead developer separate where build volume is real, let one person hold engagement management and business analysis, and never let the builder be the only tester. Name a second person on every engagement who could take it over.
Does a small Salesforce partner need Agentforce experience?
Salesforce has not published an Agentforce requirement for either tier, and what happens to partners without an AI practice is still unconfirmed, so treat claims of a hard requirement with caution. The competitive case stands on its own, though. Two completed and measured Agentforce engagements put a small firm in a different position from one with none, and two is reachable at small scale if the first is scoped tightly and instrumented.
How do you stop scope creep without damaging the client relationship?
Separate the person who hears the request from the person who approves it, price changes in schedule impact rather than fee alone, and keep a visible log of accepted changes so the end-of-project conversation is grounded in a record both sides recognize.
