Platform partner ground game: case studies, economics, and how to keep score
Public sources only. Nothing here comes from inside phData or Anthropic.
Confidence tags: [V] = verified, the linked page was read for this brief. [R] = reported, seen in search results or secondary coverage but the primary page was not read. [I] = inferred, Ray's synthesis.
Why this is in the vault
On 2026-11-02 the founder asked what an exceptional "ground game" partner looks like. These are the services firms that turn a vendor's product into production consumption. He wanted to know whether this is a good business, and how he should score himself as a Deal Solutions Architect (DSA) at phData. phData is a public Snowflake and Anthropic partner. Anthropic took part in phData's 2026-10-28 funding round. This brief collects the deal history with corrections, the ecosystem-multiplier studies, the tier rules that vendors publish, and a personal scoreboard he can use.
1. Case studies (with corrections)
| Partner | Platform | Acquirer / event | Date | Value | Tag |
|---|---|---|---|---|---|
| Appirio | Salesforce, Workday | Wipro | Announced 2016-10-20 | $500M headline. Wipro's FY2019 20-F books $475.7M. Revenue was "cresting $200m" | [V] press release, 451 Research; 20-F figure [R] SEC |
| Bluewolf | Salesforce (first consulting partner, about 2001) | IBM (into iX) | Announced 2016-03-31 | Undisclosed. Re/Code reported about $200M. 451 puts it at 2.7x revenue | [R] IBM UK newsroom; 2.7x [V] 451 |
| Cloud Sherpas | Google, Salesforce, ServiceNow | Accenture | Announced 2015-09-15, closed 2015-10-20 | Undisclosed. About 1,100 staff. 451 puts it at 2.7x revenue | [R] Accenture newsroom; 2.7x [V] 451 |
| Onica | AWS (Premier Consulting, managed service provider (MSP), 9 competencies) | Rackspace | Announced Nov 2019, closed late 2019 | Undisclosed. 350+ consultants, founded 2014 | [R] Rackspace |
| Cloudreach | AWS Premier, plus Azure and Google Cloud | Atos | Announced 2021-12-07, closed 2022-01-04 | Undisclosed. 1,000+ certifications (unsourced: not on linked page) | [R] Atos |
| 2nd Watch | AWS Premier, MSP | ST Telemedia (Temasek) took a majority stake. This was not a strategic SI exit | Oct 2019 | Undisclosed. Existing VCs kept significant stakes | [R] STT newsletter |
| Hashmap | Snowflake Premier (also Databricks) | NTT DATA Services | Announced Dec 2020, closed 2021-01-04 | Undisclosed | [R] NTT DATA |
| Hakkoda (added) | Snowflake Elite | IBM | Closed 2025-04-02 | Undisclosed | [R] IBM newsroom |
| phData | Snowflake, AWS, Anthropic | Gryphon majority growth investment (Dec 2024). New funding Oct 2026 with Anthropic participating | 2024-12-16 / 2026-10-28 | Undisclosed | [V] Gryphon post; [V] 2026 funding |
| PwC Consulting (enterprise resource planning, ERP, era) | SAP, Oracle, ERP and customer relationship management (CRM) | IBM | Announced 2002-07-30, closed 2002-10-01 | About $3.5B on about $4.9B revenue, roughly 0.7x revenue | [R] IBM 8-K |
phData's public Snowflake awards. phData won the 2020 Emerging Partner of the Year and the 2021 Regional System Integrator (RSI) Innovation award. It then won the Americas RSI award in 2022 and 2023, and Americas Partner of the Year in 2024 and 2025. In 2026 it won Implementation Partner of the Year, which phData describes as its 7th consecutive year, plus its first AI Partner of the Year. Its public numbers include 280+ Snowflake technical certifications and 255+ Snowflake projects, and phData calls itself #1 in advanced certifications. [V] phData 2026, [R] 2022. phData's own pages count the streak in different ways.
ERP-era systems integrators (SIs). Accenture describes itself as SAP's largest services partner, with a 40+ year relationship. IDC (2022) and Forrester rank it a Leader in SAP implementation. [R] Accenture/IDC. Commentators commonly say ERP implementation services cost 1-2x the license. One public case reports services at twice the software cost. [R] houseblend, information-age. This search found no reliable 1990s ratio for SAP services to SAP license revenue.
2. Is it a good business, and where does the value accrue?
Ecosystem multipliers. Each study measures partner revenue per $1 of vendor revenue, and every one was paid for by the vendor:
| Vendor | Multiplier | Study | Tag |
|---|---|---|---|
| Salesforce | $6.19 per $1 by 2026 ($5.80 in the 2019 study, unsourced: not on linked page) | IDC 2021, Salesforce-sponsored | [R] Salesforce IR |
| Microsoft | $8.45 for services partners, $10.93 for software partners (n=638, 2024). The 2017 figure was $9.01 | IDC, Microsoft-commissioned | [R] Microsoft partner blog (fetch failed) |
| SAP | $10 for partners selling their own cloud IP, $8.81 for services/own solutions | IDC Jan 2023, SAP-sponsored | [R] SAP News |
| AWS | "Up to" $7.13, reached by Expert partners. Focused partners get only $1.26. Expert partners are 31% of the base. The 2022 Canalys figure was $6.40 | Omdia 2025 and Canalys 2022, not IDC | $7.13 and $1.26 [V] AWS APN blog; 31% [R] (Omdia coverage, not on the APN blog), Omdia; 2022 [R] Canalys PDF |
Two caveats. First, these multipliers count revenue opportunity, not margin. Omdia says this directly in its study coverage [R]; the AWS blog only frames the figure as services revenue per $1 [V]. Second, the multiplier rises as a partner covers more of the lifecycle: advise, build, manage, plus its own IP. In the AWS data, the Expert multiplier is about 5.7x the Focused one ($7.13 vs $1.26: figures [V], ratio [I]).
Valuation: services vs software. IT services deals have a median of about 1.3x EV/Revenue and 10.2x EV/EBITDA, against about 3.0x and 15.2x for software (Aventis/Mergermarket, 2015 to mid-2025). [R] Aventis. Within services, data & analytics firms show a median of 16.3x EV/EBITDA. That compares with 10.4x for IT services overall and 9.8x for IT consulting, a premium of roughly 57-66%. Aventis says almost no Snowflake partner deal discloses a multiple, so these are proxies. [V] Aventis Snowflake M&A. Livingstone (2026) reports "AI-native" services firms at more than 3x revenue. [R] Livingstone
What drove the premium exits. The 2015-16 cloud-platform pure plays (Bluewolf, Cloud Sherpas, and ServiceNow partner Fruition Partners) sold in a tight band of 2.4-2.7x revenue, with Appirio in the lower part of the band. Wipro's own value at the time was reportedly about 2.0x (unsourced). [V] for comps via 451. PwC Consulting sold for about 0.7x in a mature, regulator-forced ERP market. [R] Ray's read [I] is that buyers pay a premium for three things:
- Scarce certified talent on a platform that is still in its rising phase. Incumbents were late to Salesforce, Google, and AWS and bought the talent rather than building it.
- Vendor standing. Signs include top tier, partner-of-year awards, and "first partner" status. Cloud Sherpas held Salesforce Global Strategic status, was one of 4 ServiceNow Master Partners, and won Google Partner of the Year four times. [R]
- Recurring run revenue. Onica, 2nd Watch, and Cloudreach all held MSP status, which turns project revenue into annuity revenue.
The same pattern shows up on the downside. Once a platform matures, services pricing falls back toward commodity consulting multiples.
Where value accrues [I]. Most dollars and most margin go to the vendor. Its product sells at software multiples, and its consumption grows after go-live. Partners collect a large revenue pool but are valued at services multiples. Partners gain in three ways: (a) owning IP and managed services, which matches the SAP and AWS multiplier data; (b) specializing early in a rising platform, which gets them a scarcity premium at exit; (c) data & analytics specialization, which earns the D&A premium. A ground game that only does projects is a good cash business but a poor equity story unless it moves up that stack.
3. What great ground games have in common
Each trait below is linked to the cases above. The trait list itself is [I].
- They committed to one platform before it was obvious. Bluewolf was Salesforce's first consulting partner (about 2001). Appirio was founded in 2006. Onica (2014) and phData (2014) were founded early in their platforms' growth, and phData won Snowflake's Emerging Partner award in 2020. Being early compounds into status that a late entrant cannot buy.
- They built certification depth. Cloudreach reportedly had 1,000+ certifications (unsourced), Onica had 9 AWS competencies, phData has 280+ Snowflake certifications, and Hakkoda was "Elite." Every vendor tier uses certifications as its first gate.
- They produced production references and won awards repeatedly. phData has won Snowflake awards seven years in a row. Cloud Sherpas won Google Partner of the Year four times. Awards are how vendors publicly say "this partner gets customers live."
- They added a run business to the build business. Onica, 2nd Watch, and Cloudreach were MSPs, and AWS Premier requires an MSP, DevOps, or CloudOps competency. Revenue after go-live is what makes the multiplier, and the multiple, larger.
- They added adjacent platforms near the top. Cloud Sherpas covered Google, Salesforce, and ServiceNow. Cloudreach covered AWS, Azure, and GCP. Hashmap covered Snowflake and Databricks. phData covers Snowflake, AWS, and Anthropic. These firms go deep on one platform first, then widen. The first platform pays for the second.
- They kept pace with the vendor's next wave. phData's 2026 AI Partner of the Year award and its Anthropic funding show the move from data platform to AI agents. Hakkoda was sold to IBM to "fuel AI transformations." [V]/[R]
4. How a partner should score its own execution
How vendors score partners (public tier docs)
| Vendor | What counts | Top-tier threshold | Tag |
|---|---|---|---|
| Anthropic Services Track (2026-06-03) | Active certified people (a current cert and Claude use in the last 90 days). Joint customers in production (trailing 12 months). Public customer stories. Verified quarterly, with the same bar for every firm | Select: 10 certified / 2 production / 1 story. Preferred: 100 / 15 / 3. Global Premier: 1,000 / 100 across 3+ regions / 15 stories, plus a joint business plan with exec sponsors | [V] Anthropic |
| AWS Services Path | Accredited and certified people, launched opportunities with monthly recurring revenue (MRR), competencies, business plan, sustained attainment | Premier: 25 technical certs (10 Pro/Specialty), 50 launched opportunities at $50K+ total MRR, 3 competencies (1 MSP, DevOps, or CloudOps), >6 months sustained | [V] AWS tiers |
| Snowflake Partner Network | Certifications, closed pipeline, deal registrations, customer success stories | Select / Premier / Elite. Thresholds are not public | [V] Snowflake |
All three programs gate tiers on production outcomes. Anthropic and AWS no longer count sold deals: they count customers live in production and launched opportunities with MRR attached. Signed statements of work (SOWs) and proofs of concept (POCs) do not count. [V] for the criteria, [I] for the reading.
Partner self-scorecard [I] (Ray's proposed definitions; no public industry benchmarks were found):
| Metric | Definition | Why it matters |
|---|---|---|
| Pilot-to-production rate | Share of POCs and pilots that reach a production workload within 6 months | This is the Anthropic and AWS tier currency |
| Time to production | Days from signature to the first production workload | Shorter time means consumption starts earlier and the reference arrives sooner |
| Influenced consumption | Vendor consumption or MRR attributable to the partner's deals, measured at 12 months | This is what the vendor actually optimizes for |
| Co-sell / referral rate | Share of deals sourced from the vendor or co-sold with it, compared with partner-sourced deals | Shows whether the vendor trusts you with its accounts |
| Certification density | Active certifications per delivery head, weighted to advanced and professional levels | Tier gate. Anthropic counts a cert only with 90-day usage |
| Production references | Count of public, named customer stories per year | Required at every tier. The scarcest item at scale |
| Run attach rate | Share of build engagements that convert to managed or run services | The multiplier and valuation lever |
5. What this means for an individual DSA
Emmitt Smith and Derrick Henry are useful models because their value came from volume and durability. They were not highlight-reel players. For a DSA, the equivalent is a high number of carries that end in production [I]:
- Score yourself on vendor tier currency, not bookings. The useful counts are customers live, launched opportunities, and public stories you helped make happen. A DSA whose deals reliably reach production adds directly to the counts that decide tier status.
- Design every deal for production from day one. Each scoping document should name the production workload, the first-consumption date, and the reference ask. A POC with no path to production is a carry for no gain.
- Keep certifications active. Anthropic counts a cert only if you used Claude in the last 90 days, so a credential you don't use stops counting. Track advanced certs on both Snowflake and Anthropic.
- Work where Snowflake and Claude overlap. phData's public record combines Snowflake implementation depth with AI work. Its case studies include Claude running on Snowflake Cortex, such as a disputes workflow with a claimed 192% ROI. [R] case study. Being the person who links the two platforms is rare, and rarity is what earns a premium.
- Durability comes from repeatable patterns. Reusable scoping templates, architectures, and reference-ready outcomes let one DSA run more deals per quarter without quality falling. That reuse is how a person goes from making plays to piling up yardage.
Mapping against Ray Data Co
- RDCO is dormant and phData is the main bet. This brief strengthens the credibility-for-phData-sales thesis. Public content on "agents in production" directly supports the scarcest item in the Anthropic and AWS tiers, which is production references.
- The cert path maps onto Anthropic's active-certified-practitioner metric. Certs plus 90-day usage count in the tier model. Certs alone do not.
- The "agents in production" credibility play should be measured in the same units vendors use: production go-lives and public stories, not posts or subscribers.
- There is no case for RDCO as a services firm. Services trade at about 10x EBITDA and depend on early specialization plus a run business, and phData already has both at scale. [I]
Related
- [[2026-06-09-consultancy-partnership-tracks-velocity]]
- [[2026-09-11-phdata-snowflake-partnership-warm-start-map]]
- [[2026-11-02-ark-anthropic-s1-training-paying-off]]
- [[2026-06-11-phdata-anthropic-partner-tier-cert-leverage]]