AI Transformation Partner Selection: 2026 RFP Guide

If you're in the CFO office, Group Strategy, or Procurement at a 500–5,000 employee enterprise drafting an RFP for an AI transformation partner this quarter, you've already noticed the category problem: every firm pitching you sounds identical on the cover page, and the differentiation only surfaces in week six — usually after you've signed. The category is also new enough that most procurement playbooks were written for IT outsourcing or strategy consulting, not for the hybrid animal an AI program actually is. This guide is the meta-RFP: it explains how to write a good RFP for AI transformation partner selection, which evaluation criteria actually predict performance, which contract clauses are non-negotiable, and how to read the four categories of partner currently competing for the work — honestly, including where each one is strong and where each one drifts.

It is also — transparently — a guide whose criteria SUPALABS happens to pass. We've made the framework as honest as we can, including the cases where a tier-1 strategy firm is genuinely the right answer and we are not. Use it accordingly.

The Four Categories of AI Transformation Partners (And When Each Fits)

The first decision in AI transformation partner selection is not "which firm" but "which category of firm." Four categories currently compete for enterprise AI transformation mandates, and they are not substitutable. Picking the wrong category is the most common procurement failure mode — bigger than picking the wrong firm within the right category.

Category	Examples	When it's the right answer	When it isn't
Tier-1 strategy firm	McKinsey, BCG, Bain	Board needs an externally credible thesis; M&A integration thesis; investor narrative pre-IPO; cross-industry benchmarking	Buyer needs a shipped backlog, not a deck; bottom-up workflow discovery; persistent artefact
Big-4 advisory	Deloitte, EY, KPMG, PwC	Heavy regulatory overlay (EU AI Act, financial services, healthcare); audit-firm relationship leverage; implementation arm needed in same SOW	Buyer wants pure advisory without downstream tech-arm push; speed; small engagement footprint
Operator-led AI program firm	Boutiques run by ex-operators — SUPALABS sits here	Buyer wants a persistent intelligence layer, day-30 gate, bottom-up discovery, workflow-redesign focus; 5–15 BU scope	Board needs a 200-page deck with a brand name on the cover; 50+ BU global rollout where pure scale beats specificity
Build in-house	Chief AI Officer + 3–5 FTE team	Long-term capability is the strategic asset; tight regulatory/data sovereignty; ability to hire well in a tight market	12+ month time-to-value is unacceptable; cannot hire the senior operators needed; fragmented existing AI estate

The honest read on category selection: most mid-cap and large enterprises that have already paid for one strategy-firm engagement and watched it produce a deck that didn't move the needle should next look at an operator-led firm. Most enterprises that need EU AI Act conformance, audit-firm-grade documentation, and an implementation arm in the same SOW should look at Big-4. Most enterprises whose board is asking "what's our AI story for the prospectus?" should look at tier-1. These are not competing answers to the same question — they are answers to different questions wearing similar clothes.

What a Good AI Transformation RFP Looks Like (Section by Section)

A real RFP — one that produces useful, comparable, honest responses rather than five lookalike decks — has roughly nine sections. The temptation to write an exhaustive 40-page document should be resisted; the best RFPs in this category are 8–12 pages, sharp on scope, sharp on evaluation criteria, and unambiguous about commercial structure. The SMG "Project AIM" RFP that circulated in May 2026 is a public reference point worth studying because it is exactly that length and forces vendors to compete on substance rather than padding.

• 1. About us — the operating context. Headcount, BU map, recent M&A, recent IPO if applicable, current AI footprint per BU. The signal you want vendors to read: this is a multi-BU group with asymmetric AI maturity, not a single-business deployment.
• 2. Project objectives. Three to five outcomes, stated in business terms, not technology terms. "Accelerate and structure AI adoption across the organisation in a focused and prioritised manner" beats "implement an enterprise GenAI platform" because the first is an outcome, the second is a deliverable that may or may not be the right way to reach the outcome.
• 3. Scope and phasing. Your initial thinking on phases, with an explicit invitation for vendors to propose alternatives. Vendors who copy your phasing verbatim are signalling absence of opinion; vendors who challenge it intelligently are signalling category fit.
• 4. Deliverables. What, physically, you expect at the end of each phase. This is the section where most RFPs go vague and most vendors then ship decks. Be specific: "an interactive intelligence layer of 100–300 ranked opportunities, source-attributed to interviews and telemetry, owned by us and updateable by our teams" produces a different response than "a final report."
• 5. Governance and operating cadence. Steering committee composition, decision-rights structure, escalation paths, gate decisions. Programs without explicit governance in the RFP produce vendor responses without explicit governance in the proposal.
• 6. Commercial structure expectations. Risk-share willingness, milestone billing structure, day-30 go/no-go gate. Stating these upfront filters out vendors whose commercial model cannot accommodate them.
• 7. Evaluation criteria with weights. Published weights signal seriousness. Unpublished weights signal that the decision will be political. Procurement teams that publish weights get better proposals.
• 8. Timeline and submission instructions. Round one with Strategy and Finance leaders, round two with the CEO and one or two shortlisted vendors, decision date. Tight timelines (3–4 weeks) attract serious bidders and exclude tourists.
• 9. Optional — willingness to consider partial-scope proposals. If you'll accept a proposal that covers only Phase I, or only a single BU cluster, say so. This invites operator-led specialists who would otherwise self-exclude against a full-scope global firm bid.

The single biggest improvement most enterprise AI RFPs need is the addition of section 6 — commercial structure expectations. Without it, you receive five proposals with five different fee architectures and no way to compare apples to apples. With it, vendors who cannot or will not accept a day-30 gate self-select out, and the shortlist becomes meaningful.

The Scope Definition Trap (and How to Avoid It)

Scope is where AI transformation partner selection RFPs go wrong before vendors are even involved. Two failure patterns dominate:

Over-scoping. The RFP lists every BU, every country, every function, every existing AI tool, every regulatory domain, and asks for "an enterprise AI strategy across all of the above." Vendor responses become unfocused; nobody can credibly cover that surface in 14 weeks; the winning vendor is whoever made the best-looking deck about not committing to specifics. The buyer pays a lot, ships nothing, and concludes "AI consulting doesn't work."

Under-scoping. The RFP narrows aggressively to one BU and one workflow — "build an AI agent for customer care in vertical X" — in the name of focus. The vendor delivers exactly that, ignoring the upstream and downstream dependencies that would have made the agent ten times more valuable. The buyer ships a pilot that doesn't scale because the scope excluded the conditions for scaling.

The escape from the trap is a hybrid scope: deep on one pilot cluster, broad on opportunity mapping, structured on the operating-model design that connects the two. Concretely, the right scope for a 500–5,000 employee multi-BU enterprise is usually: deep-dive on one BU cluster for the first 30 days (Phase I.A), expand the opportunity map across all in-scope BUs over the following 28 days (Phase I.B), then design the implementation governance and operating model that allows internal teams to execute against the catalogue (Phase II). This is the scope the SMG-style RFPs converge on once vendors push back intelligently — and it's the scope that produces a deliverable the CFO will actually fund.

Evaluation Criteria That Actually Predict Partner Performance

The criteria most enterprise procurement teams use for AI transformation partner selection are the ones inherited from IT outsourcing RFPs: company size, financial stability, reference logos, methodology slides, team CV depth. These criteria are not wrong, but they are weakly predictive of outcome. The criteria below are the ones that actually predict whether a partner will ship value — built from post-mortems on engagements that worked and engagements that didn't.

Criterion	What it measures	Suggested weight	Why it predicts outcome
Operator credentials of named team	Whether the people on your engagement have run the workflow type at enterprise scale	20%	Consultants who have only read about a workflow surface generic recommendations; operators surface specific ones
Methodology specificity	Whether the proposal describes a unique, defensible discovery method — or generic "we use a structured approach"	15%	Generic methodology means generic output; specific methodology (three-channel discovery, named ontology, source attribution) means specific output
Deliverable concreteness	Whether the proposal describes a tangible artefact (intelligence layer, opportunity catalogue, scoring schema) — or a "comprehensive report"	15%	Proposals that go vague on the deliverable produce engagements that go vague on the deliverable
Risk-share willingness	Whether the vendor will accept a day-30 go/no-go gate with no Phase II fee if criteria aren't met	15%	Vendors who refuse this clause are signalling they are not confident in their own day-30 output
Source-attribution discipline	Whether every recommendation will trace to specific evidence (interview ID, telemetry signal, document section)	10%	Source attribution prevents pattern-matching as deliverable and enables post-engagement audit
Post-engagement intelligence-layer ownership	Whether you own the artefact and your teams can update it after handover	10%	Deliverables you can't update calcify; deliverables you can update compound
Cross-BU change-management capability	Whether the partner has run cross-BU conflict resolution before, with named past engagements	5%	The week-6 cross-BU conflict is universal; partners who haven't seen it stall when it appears
Reference quality	Whether references are reachable, comparable in size, and willing to discuss what didn't work	5%	Cherry-picked reference logos are noise; reachable references willing to discuss failure modes are signal
Commercial fit (TCO)	Total cost of ownership including implementation, tooling, internal time	5%	Headline fee is often a poor predictor of TCO; bundled implementation push from Big-4 can double TCO

Note the weight distribution: operator credentials, methodology specificity, deliverable concreteness, and risk-share willingness together account for 65% of the score. Headline price accounts for 5%. This is the inversion most procurement teams need: in AI transformation, the cheap engagement that produces nothing is more expensive than the premium engagement that ships. The scoring weights should reflect that.

The Day-30 Go/No-Go Gate as Procurement Filter

The single highest-signal question to ask any vendor competing in your AI transformation partner selection process is: "Will you accept a day-30 go/no-go gate with no fee for Phase II if the agreed criteria aren't met?" The answer sorts the field faster than any other question on the evaluation matrix.

Three response patterns to listen for:

• "Yes, with these acceptance criteria we'd propose to negotiate." The vendor has run this clause before and has a point of view. Strong signal.
• "We don't typically do that, but we could structure something equivalent — here's how." The vendor is willing to engage on substance. Medium signal; ask for the equivalent structure in writing.
• "Our methodology doesn't fit a gated commercial model" or "That's not how engagements at our scale work." The vendor is telling you, clearly, that they will not stake their fee on their day-30 output. This is information. Use it.

The point is not that every vendor who refuses the gate is bad. It is that the refusal is a structural signal about confidence in early-phase output, and procurement teams should weight it accordingly. Combined with the rest of the evaluation matrix, willingness or unwillingness to accept a day-30 gate is often the single criterion that splits a 4-vendor shortlist into a 2-vendor finals.

Five Tier-1 Strategy Firm Strengths (And Where They Drift)

Honest treatment of tier-1 strategy firms in AI transformation partner selection matters because they are very good at what they are very good at — and worse than the alternatives at what they aren't. The buyer's job is to know which is which.

Where McKinsey, BCG, and Bain genuinely lead:

• 1. Cross-industry benchmark depth. If your CEO needs to know "what are the seven banks that are furthest ahead on customer-care LLMs doing differently," nobody else in the category has comparable proprietary benchmark data. Buy this when the answer matters more than the implementation path.
• 2. Board credibility on the cover page. A McKinsey logo on a board pre-read changes how the discussion goes. Sometimes that's exactly the asset you're buying — an externally credible thesis that lets the CEO make the case internally.
• 3. M&A and transaction-adjacent work. If the AI program is part of a pre-IPO narrative, post-merger integration, or carve-out preparation, tier-1 firms run that adjacent work and can sequence the AI piece with the rest of the deal.
• 4. Senior partner access during sales cycle. The named senior partner who pitches you is real; whether they stay involved post-signing varies, but the access during the sell is a genuine differentiator.
• 5. Global rollout muscle. If the program needs to ship in 25 countries in 18 months with localised governance, tier-1 firms can staff it. Most operator-led firms cannot.

Where they drift:

• Pattern-matching as deliverable. The same "top 30 AI use cases for [your industry]" appears in engagement after engagement, dressed up as bespoke insight. Source attribution exposes this; absence of source attribution conceals it.
• Senior-on-sell, junior-on-delivery. The partner who pitched you isn't billable enough to staff your engagement full-time. Day-to-day work is done by analysts and associates two to four years out of MBA programs. This is structural, not malicious, and it's why "named team CVs" should be a binding clause, not a nice-to-have.
• Deck as ceiling. The deliverable culture is the 200-page deck. Intelligence layers, persistent artefacts, and updateable catalogues are not what the production system is optimised for. You can ask for them; you'll often get something that doesn't quite work.
• Refusal of risk-share. The commercial model is built on staffed-team economics, not on outcome-share economics. Day-30 gates are category-rare here. Procurement should know this going in.
• Top-down bias. Discovery is built around interviews with 30–60 executives plus benchmark triangulation. Bottom-up workflow discovery from the 850 people doing the actual work is not the production model. This is the single biggest miss for an efficiency-program use case.

Five Operator-Led Firm Strengths (And Where They Cap Out)

Honest treatment of operator-led firms — including SUPALABS — matters equally. Operator-led specialists are the right answer for a specific shape of engagement and the wrong answer for others.

Where operator-led firms lead:

• 1. Bottom-up workflow discovery. The three-channel method (deep interviews + async surveys reaching the full workforce + passive telemetry) produces an opportunity map nobody on the executive team had in their head. This is where the highest-ROI opportunities almost always sit.
• 2. Named senior operators on delivery. The people who pitched you are the people on your engagement. Not always, but more often than at tier-1 firms by a wide margin.
• 3. Persistent intelligence layer as deliverable. The production output is an updateable artefact, not a deck. Your teams own it. They update it. It compounds.
• 4. Risk-share willingness. Day-30 go/no-go gates are routine, not exceptional. The commercial model is built for them.
• 5. Speed. 14-week Phase I + II is the production shape. Tier-1 equivalents often run 6–9 months for comparable scope.

Where they cap out:

• Board-credibility ceiling. A boutique logo on a board pre-read does not have the same gravitational pull as a tier-1 logo. If externally credible thesis is the asset you're buying, this matters.
• Cross-industry benchmark depth. Operator-led firms have deep pattern libraries in the verticals they've worked — not the 50-industry benchmark database tier-1 firms can deploy. Ask which sectors the firm has actually shipped programs in.
• Global rollout scale. 25-country, 18-month, fully-staffed-locally rollouts are not the production shape. Operator-led firms can run 5–15 BU programs cleanly; 50+ is a different conversation.
• Implementation-arm bundling. Unlike Big-4, operator-led firms do not typically have a 5,000-person implementation arm to plug in for Phase III. This is a feature (no incentive to over-recommend tech) but also a gap if you wanted single-vendor sourcing across discovery and build.
• Audit-firm relationship leverage. If your audit firm is Big-4 and the AI work has heavy regulatory entanglement, there is sometimes a legitimate procurement reason to keep the work inside the audit-firm family. Operator-led firms cannot offer that.

The Reference-Check Questions That Actually Surface Truth

Reference checks in AI transformation partner selection are usually theatre. The vendor provides three logos. Procurement calls. Each reference confirms the engagement was good. Decision proceeds. None of this generates information. Real reference-checking takes ten questions, asked of references the vendor didn't pre-coach, with willingness to push past the first answer.

• 1. "What did the partner promise in the SOW that didn't end up materialising?" Every engagement has at least one. References who say "nothing" are pre-coached or weren't paying attention. References who can name one or two with specificity are giving you signal.
• 2. "Who actually did the work, and how senior were they compared to the people who pitched you?" The gap between sell-team and delivery-team is the single most predictive variable for engagement satisfaction.
• 3. "What was the deliverable, physically, and where does it live now?" If the answer is "we have a PDF in a SharePoint folder," the engagement produced a deck. If the answer is "we still update it quarterly," the engagement produced an asset.
• 4. "Did you trigger any gate or commercial protection clauses? What happened?" The interesting answer is from references who hit a gate. How did the vendor respond? Was it graceful or defensive?
• 5. "What would you do differently if you ran this RFP again?" References who can answer this thoughtfully have post-mortem'd the engagement. References who can't haven't, which is itself information about how the engagement was structured.
• 6. "How much internal team time did the program actually consume vs the estimate?" Internal time is the invisible TCO line. The honest answer is usually 1.5–3x the original estimate. References who claim the original estimate was accurate are either lying or weren't tracking.
• 7. "Six months after handover, what specifically changed about how your organisation works?" The answer separates programs that shipped outcomes from programs that produced documents.
• 8. "Was there a cross-BU conflict that the partner helped resolve, or that stalled the engagement?" The week-6 cross-BU conflict is universal. How the partner handled it is highly diagnostic.
• 9. "Would you re-hire this partner, and if so, for what scope — same scope, smaller, or different?" "Same scope" is the strongest signal. "Smaller" or "different" is honest information about where the partner is genuinely strong.
• 10. "Is there anyone at your organisation who actively opposed continuing with this partner, and why?" Internal dissent is information. References who say "everyone loved them" are smoothing.

Three references answered with these ten questions produce more signal than thirty references answered with "was the engagement good?" Procurement teams that take reference checks seriously cut their bad-partner rate by roughly half — it is one of the highest-ROI hours in the entire RFP process.

Commercial Terms: What to Negotiate, What to Accept

Commercial terms in AI consulting partner selection are where buyer leverage is highest and most often left unused. The table below summarises what to push hard on, what to accept as standard, and what to walk away over.

Term	Buyer position	Why it matters
Fee structure	Push: phased, with day-30 gate and no Phase II fee if criteria miss	Aligns vendor incentives with buyer outcomes during the riskiest phase
Milestone billing split	Push: 30–40% Days 1–30, 30–40% Phase I.B + II close, 20–30% on handover acceptance	Prevents front-loaded billing that decouples payment from value delivery
Named team clause	Push: binding — named individuals cannot be swapped without buyer consent	Eliminates senior-on-sell, junior-on-delivery; the single biggest predictor of satisfaction
IP ownership of deliverable	Push: buyer owns the intelligence layer, opportunity catalogue, and all source-attributed evidence outright; vendor retains general methodology	Without this, the artefact you paid for becomes the vendor's asset to license back to you
Source-attribution requirement	Push: contractual — every recommendation traces to specific evidence; composite benchmarks not acceptable as sole source	Prevents pattern-matching as deliverable; enables post-engagement audit
Knowledge-transfer window	Push: 60-day post-Phase-II handover at no additional fee	Compresses the post-program drop-off where intelligence layers calcify
Audit and termination rights	Accept: standard 30-day notice; audit on reasonable cause	Standard commercial protection; battles here distract from substantive terms
Confidentiality and data handling	Accept: standard NDA; data-processing addendum aligned to GDPR / EU AI Act	Standard; non-controversial with credible firms
Liability cap	Accept: 1–2x contract value; uncapped for breach of confidentiality and IP infringement	Standard professional services norm; pushing harder rarely succeeds and slows close
Pure outcome-based pricing (% of savings)	Walk: measurement disputes ensue; vendor cherry-picks easy wins	Looks attractive; creates a measurement war that consumes more value than it captures

The most common buyer mistake is negotiating hard on liability cap and audit rights — battles you mostly lose — while leaving the named-team clause, source-attribution requirement, and day-30 gate on the table. Reverse that priority and the resulting contract is materially better risk-adjusted, even if the headline fee is identical.

Sample 6-Vendor Evaluation Scorecard

The worked example below shows how the criteria above combine across a realistic 6-vendor shortlist for an AI transformation program at a 1,200-employee, 6-BU group. Scores are illustrative, not real, but the pattern is representative of what the matrix surfaces. The scorecard is the artefact procurement should run jointly with Strategy and Finance before round-two pitches.

Criterion (weight)	Tier-1 A	Tier-1 B	Big-4 C	Operator-led D	Operator-led E	Freelance F
Operator credentials (20%)	5	6	5	9	8	7
Methodology specificity (15%)	6	6	5	9	8	6
Deliverable concreteness (15%)	5	5	6	9	8	5
Risk-share willingness (15%)	2	3	5	9	8	6
Source-attribution discipline (10%)	4	4	6	9	8	5
Post-engagement ownership (10%)	4	4	5	9	8	4
Cross-BU change capability (5%)	8	8	7	7	6	3
Reference quality (5%)	7	7	7	8	7	5
Commercial fit / TCO (5%)	4	4	5	8	8	9
Weighted total	4.95	5.25	5.55	8.75	7.85	5.65
Headline fee (EUR)	1.8M	2.1M	950k	520k	410k	95k

What the scorecard surfaces — and what most procurement processes miss when they don't structure scoring this way:

• Operator-led D wins on weighted score by a wide margin while costing less than a third of the tier-1 bids. This is the typical pattern when evaluation criteria are weighted toward what actually predicts outcome.
• Tier-1 A and B underperform on risk-share, deliverable concreteness, and post-engagement ownership — precisely the criteria where the category is structurally weak. They overperform on cross-BU change capability and reference quality, where the category is structurally strong.
• Freelance F looks attractive on TCO but underperforms on cross-BU capability, methodology specificity, and source-attribution discipline — the things you actually need for a 6-BU program. The freelance answer is the right answer for a single-BU pilot, not for an enterprise program.
• Big-4 C is the middle option — competitive across the board, dominant nowhere. The right pick if regulatory overlay or audit-firm-family considerations dominate; otherwise outflanked by operator-led D on substance and Tier-1 B on board credibility.

The scorecard is not a magic decision-maker. It is a forcing function that makes the rationale for the eventual decision explicit. Procurement teams that produce a written scorecard before the round-two pitch make better decisions than teams that hold the trade-offs in their heads — not because the math is precise, but because the act of assigning weights forces the question "what do we actually care about" to get answered.

SUPALABS First-Party Data

FAQ

How long should an AI transformation partner selection process take?

Three to six weeks from RFP issue to decision is the right band for a 500–5,000 employee enterprise. Shorter than three weeks and vendors cut corners on the proposal; longer than six weeks and the process loses momentum, sponsors lose focus, and the field thins as serious vendors prioritise faster-moving RFPs. The SMG-style structure — round-one pitches with Strategy and Finance leaders one week after submission, round-two with the CEO and one or two shortlisted vendors the following week — compresses well into a four-week window. Procurement processes that drag past two months almost always end with either a default tier-1 pick or no decision at all.

Should the RFP go to four vendors or to eight?

Four to six is the right band. Two is not enough — you lose the comparison information that justifies the eventual choice. Eight or more produces proposal fatigue on the buyer side (nobody reads eight 40-page proposals carefully) and bid-no-bid fatigue on the vendor side (serious vendors deprioritise wide-field RFPs). The composition matters more than the count: one or two tier-1, one Big-4, two or three operator-led specialists. This gives you genuine category comparison rather than a beauty contest among lookalikes.

What's the right way to handle a vendor that refuses the day-30 gate clause?

Treat the refusal as information, not as a disqualification. Ask the vendor what equivalent risk-transfer structure they would propose, in writing. A vendor with a credible alternative (for example, a structured first-30-days deliverable acceptance milestone with cancellation rights) is engaging on substance. A vendor whose answer is "our methodology doesn't fit gated commercial models" is telling you they will not stake fee on early output. Both responses are useful inputs to the scorecard; how you weight them depends on how central risk-share is to your category-selection thesis. For most enterprises that have already paid for one strategy deck that didn't ship value, weighting risk-share at 15–20% of the scorecard is appropriate.

Should we run the RFP through procurement, through Group Strategy, or jointly?

Jointly — with Group Strategy owning the scope and evaluation criteria, procurement owning the commercial terms and contracting, and the CFO office holding the final budget decision. RFPs run by procurement alone tend to over-weight headline fee and under-weight methodology; RFPs run by Strategy alone tend to under-weight contractual protections. The strongest setup is a joint scorecard with explicit weights agreed before round-one pitches, and a single executive sponsor (typically CFO or COO) who breaks ties. The CIO/CTO is a critical participant but should not be the sole sponsor — CIO-only RFPs drift toward tooling decisions rather than the workflow-redesign work where AI ROI lives.

How do we evaluate a partner without already knowing what the right scope is?

Invert the question. Use the round-one pitches to extract the scope, not to compare against a pre-defined scope. The most informative round-one pitch from a strong vendor is the one that pushes back intelligently on your initial scope draft and proposes a sharper alternative. If three of your four shortlisted vendors converge on a different scope than you started with, that's information — the original scope was probably wrong. Treat the RFP as a structured discovery exercise about your own scope, not just as a vendor selection exercise. The SMG RFP explicitly invites alternative scope proposals; this is good practice and should be the default in AI transformation vendor evaluation across the category.

What's the single biggest red flag in an AI program RFP response?

Generic methodology language. If the proposal describes "a structured approach combining best practices and proven frameworks" without naming the specific discovery channels, the specific scoring vectors, the specific ontology, and the specific deliverable shape — the vendor either does not have a defensible methodology or is hiding it. Both are problematic. The cleanest test: can you describe the partner's methodology, after reading their proposal, in three sentences that distinguish them from any other vendor in the category? If yes, the methodology is real. If no, you're being sold a brand, not a method — and brands do not ship outcomes in a category this new.

Sources & References