Why this is in the vault
A sharp framework for how much professional-services revenue a software company should carry, using Guidewire CFO Jeff Cooper's "sub-20%, falling" rule and a worked P&L showing why shrinking services revenue makes the top line look worse before it looks better.
Mapping against Ray Data Co
The load-bearing connection is phData's capacity model: Guidewire's fix for services-capacity risk is building a certified-partner ecosystem so implementation capacity scales without proportional headcount ("every consultant who can successfully implement Guidewire gives them capacity without another person directly sitting on Guidewire's payroll"). That's structurally the same lever behind project_cert_bonus_and_deal_placement — each additional certified consultant at phData is capacity and better deal placement, not just a resume line, which is exactly the "certified consultants outside your own payroll" logic Cooper describes. The piece's bigger warning also applies directly: phData is a services-heavy business bundled with platform/software work, and CJ's point that cutting services mix tanks reported top-line growth before subscription/recurring revenue catches up is a live tension for any services firm trying to look more "software" to a buyer or investor — worth keeping in mind for how phData frames its own mix story.
The core argument
Services dollars are worth less than subscription dollars (lower margin, one-time, no multiple), but implementation work is often the gate that makes new ARR possible — so services capacity, not sales headcount, can become the actual growth constraint. Cooper's rule of thumb: keep services under ~20% of revenue and falling, hand repeatable implementation work to trained/certified partners, and keep only the genuinely novel work in-house. The catch: deliberately shrinking services revenue makes the P&L look worse for a few quarters (lost services revenue books instantly; new subscription revenue books ratably), which is dangerous for a PE-backed company with EBITDA covenants or incentive plans tied to total revenue unless the mix shift is modeled and explained to investors up front.
Curation section
- Run the Numbers podcast — Brian Neider (Managing Director) on "the hierarchy of metrics BS," spotting misleading EBITDA add-backs, and how AI is changing software defensibility. Cross-promoted, not written by CJ.
- Quote of the issue — "Your best race won't be the fittest or the smartest. But it will be the bravest." (unattributed).
⚠️ Sponsorship
Two distinct disclosures in this issue:
- Tabs (third-party, paid) — top-of-issue banner ad for the Tabs Billing Lag Calculator, standard "Mostly Metrics is proudly powered by Tabs" sponsor block with UTM-tagged links. Tabs is Mostly Metrics' Quote-to-Cash series-standing sponsor, but this issue is not a Quote-to-Cash installment — the banner placement is independent of topic, confirming Tabs buys general newsletter real estate, not just series-specific slots.
- Looking for Leverage (CJ self-promotion, not third-party) — the entire core essay is explicitly a cross-post from CJ's other newsletter, with two separate "Subscribe for free" CTAs for it embedded in the issue. This is CJ growing his own second publication, not a paid sponsor, but it means the "research" here wasn't written fresh for Mostly Metrics — treat the framing as optimized for Looking for Leverage's PE-operator audience (hence the EBITDA-covenant/PE-sponsor angle in the back half) rather than Mostly Metrics' VC-backed SaaS audience.
Related
- [[2026-11-01-mostlymetrics-revenue-vs-ebitda-multiple]]
- [[2026-10-20-mostlymetrics-gross-margin-cogs-classification]]
- [[project_cert_bonus_and_deal_placement]]
- [[project_credibility_for_phdata_sales]]