An agency client health score is a repeatable way to identify which accounts need intervention, which are stable, and which are ready for a broader conversation. It should not replace account-manager judgement. It should make that judgement more consistent, visible and easier to act on before renewal pressure arrives.
Many agencies already hold the necessary signals in project tools, CRM records, reporting dashboards and meeting notes. The common failure is not missing data; it is treating every account as equally healthy until a cancellation email proves otherwise.
A useful model combines delivery performance, client engagement, commercial value, satisfaction signals and renewal risk. The objective is practical: give leadership and account teams a weekly view of where to protect revenue, unblock work and pursue credible expansion.
Start with the operating decision, not the spreadsheet
Before assigning points, define the decisions the score must support. In my professional judgement, a health score that has no owner, cadence or prescribed action becomes another dashboard metric that nobody trusts.
For most SEO, paid media and digital marketing agencies, the score should answer five questions:
- Which accounts need a retention intervention this week?
- Which accounts are operationally blocked and need senior help?
- Which accounts are commercially unhealthy despite appearing satisfied?
- Which stable accounts have evidence-based expansion potential?
- Where does the agency have a systemic delivery or capacity problem?
Give each question an accountable owner. Client services may own interventions, delivery leads may own blocked work, finance may validate margins, and leadership may approve save plans or strategic investment. The score is shared infrastructure, not a performance-management weapon for one department.
Use a weekly review for at-risk accounts and a monthly portfolio review for trends. This timing is practitioner judgement: weekly is frequent enough to catch deteriorating relationships without creating constant administrative churn.
Build five score components
Use a 0–100 score, with five components that are individually visible. Visibility matters because an overall score of 62 is less useful than knowing it reflects strong delivery but weak stakeholder engagement.
| Component | Suggested weight | What to measure |
|---|---|---|
| Delivery performance | 25% | Milestones met, work accepted, implementation velocity and unresolved blockers |
| Client engagement | 20% | Attendance, response times, access to decision-makers and approval progress |
| Commercial value | 20% | Gross margin, payment status, scope stability and strategic fit |
| Satisfaction signals | 20% | Direct feedback, meeting tone, advocacy, escalations and confidence in the plan |
| Renewal risk and growth | 15% | Contract timing, procurement signals, budget changes and validated opportunities |
These are starting weights, not universal rules. Label them as practitioner judgment until you have enough renewal and churn history to test whether a different weighting better predicts outcomes. A performance-led SEO retainer may weight delivery more heavily; a complex enterprise account may need more emphasis on engagement and renewal process.
1. Delivery performance: measure controllable progress
Delivery should assess whether the agency is doing what it committed to do, not whether every marketing outcome has already materialised. Search visibility, pipeline and revenue are important, but they have lags and are affected by client implementation, seasonality, competition and measurement quality.
Track on-time completion of agreed milestones, age of open dependencies, response time to critical issues and the proportion of recommendations implemented. For SEO, distinguish agency-owned work from client-owned development and approval work. Otherwise, the score punishes the wrong party and conceals the real bottleneck.
Google’s Google Search documentation is useful as a primary reference for understanding search systems and available tools, but it does not prescribe how an agency should score client health. The scoring design remains a management choice.
If implementation delay is a recurring issue, pair the health score with a documented workflow such as this guide to improving SEO implementation velocity. A weak delivery score should point to a specific next action, not merely a red status.
2. Client engagement: capture relationship reality early
Engagement is often the earliest useful warning signal. Repeatedly postponed meetings, lost access to decision-makers, slow approvals and vague feedback may precede churn long before a client says they are unhappy.
Use observable fields wherever possible: executive sponsor attended the last two reviews, decisions were made within an agreed period, requested inputs were supplied, and the client completed the agreed approval steps. Add a short account-manager assessment, but keep it structured: positive, neutral or concerning, with a one-sentence reason.
Do not mistake quiet clients for happy clients. Some are simply busy; others are disengaging. The distinction emerges when attendance, decisions, feedback and commercial signals are reviewed together.
3. Commercial value: include profitability without making it punitive
An account can be pleasant, strategically important and still commercially unhealthy. Include current gross margin against the account’s target, aged receivables, unapproved out-of-scope work and the effort required to service the account.
Margin should not be used to downgrade a new account that is deliberately being onboarded heavily, or a strategic account with an explicitly approved investment case. Record the exception and its end date. This is practitioner judgement, but it prevents a score from treating intentional investment as accidental failure.
For a consistent margin calculation, use a defined cost basis and agreed treatment of leadership time, tools and contractors. This companion guide on measuring SEO client profitability provides a practical structure for that work.
4. Satisfaction signals: favour evidence over sentiment theatre
A quarterly satisfaction question can help, but it is a weak standalone predictor when it is anonymous, infrequent or answered by someone outside the buying group. Combine it with evidence from executive reviews, unsolicited positive or negative feedback, escalations, reference willingness and confidence in the next-quarter plan.
Use a simple rubric. A score of 100 might require recent positive senior feedback plus clear agreement on priorities. A score of 50 could indicate neutral feedback and unresolved concerns. A score of 0 should be reserved for an explicit escalation, loss of trust or stated intention to review suppliers.
Do not manufacture precision. A consistent, documented 0/50/100 rubric is generally more reliable than asking account managers to choose among 21 vaguely different sentiment values.
5. Renewal risk and growth: score signals, not optimism
Renewal risk should increase as the contract end date approaches without a documented renewal path. Other negative signals include procurement activity, budget freezes, a new marketing leader, a reduced scope request, competitor involvement or a request for handover materials.
Growth potential is different from account health. A client may be healthy but have no current budget or need to expand. Keep an opportunity flag alongside the score rather than inflating health because an account looks valuable. Qualify expansion with a documented problem, buying stakeholder, likely budget route and evidence that the agency can deliver the added service profitably.
For example, an SEO client with strong content performance but persistent conversion friction may warrant a CRO proposal only after the agency can connect search activity to commercial outcomes. A SEO-to-CRO feedback loop can provide the evidence needed for that conversation.
Calculate the score and set intervention thresholds
Calculate the weighted score as: (Delivery × 0.25) + (Engagement × 0.20) + (Commercial × 0.20) + (Satisfaction × 0.20) + (Renewal/Growth × 0.15).
Consider an account with delivery 80, engagement 55, commercial 70, satisfaction 60 and renewal/growth 40. Its score is (80×0.25) + (55×0.20) + (70×0.20) + (60×0.20) + (40×0.15) = 20 + 11 + 14 + 12 + 6 = 63.
That score is not automatically a crisis. The decision threshold is what makes it useful. Here is a concrete practitioner-designed rule: accounts below 60 enter a recovery plan; accounts from 60 to 74 require a named watch item; 75 and above are stable. Add an override: any explicit non-renewal signal, unresolved executive escalation or contract end within 90 days without a renewal meeting caps the account at “at risk”, regardless of its calculated score.
The cap avoids a common contamination problem: excellent historic delivery can mask an immediate commercial threat. Review the thresholds after two renewal cycles. Compare prior scores with actual renewals, downgrades and expansions, then adjust only if the pattern is meaningful and documented.
Make data quality and ownership non-negotiable
A health score is only as credible as its inputs. Define each field, its system of record, update frequency and owner. Keep the first version deliberately small. A mature agency may automate data pulls from a CRM, project platform, finance system and reporting stack; a smaller agency can begin with a controlled monthly sheet.
At minimum, audit missing fields, stale contract dates, duplicated client records and unexplained score changes. A structured approach to marketing data quality monitoring is directly relevant here. Automation accelerates a bad process as efficiently as a good one.
Bing’s Webmaster Tools can supply useful search-performance context for relevant accounts, but no platform metric should become a direct proxy for relationship health. Use platform data to inform delivery discussions, then retain human review for scope, confidence and commercial context.
Turn scores into specific retention and growth plays
Each status should trigger a standard response. For at-risk accounts, hold a short internal review within five working days, identify the root cause, assign an executive sponsor where needed and agree a client-facing recovery plan with dates. Do not respond with a generic “check-in”; explain the issue, the corrective action and the decision needed from each side.
For watch-list accounts, resolve one leading indicator before it becomes a renewal problem: obtain implementation approval, restore stakeholder access, reset a mismatched KPI or clarify a scope dispute. For stable accounts, schedule forward-looking conversations around proven needs, not a reflexive upsell.
At portfolio level, look for clusters. If several clients show low delivery scores because of client-side development delays, improve onboarding, dependency mapping and executive expectation-setting. If margins are falling across accounts, examine capacity, pricing and scope governance rather than asking account managers to “work smarter.”
FAQ and conclusion
How often should an agency client health score be updated?
Update operational inputs weekly for active accounts and review the full portfolio monthly. Contract dates, major escalations and renewal signals should be updated as soon as they change. This cadence is practitioner judgement, designed to balance timely action with administrative effort.
What is a good agency client health score?
A good score depends on your definitions and history. Start with clear bands such as at risk below 60, watch from 60 to 74 and stable at 75 or above. Validate those thresholds against actual renewals, scope reductions and margin outcomes before treating them as reliable.
Should account managers score their own clients?
Yes, but not alone. They hold essential context, while delivery, finance and leadership should validate relevant fields. Use structured evidence and an override log to reduce optimism, inconsistency and surprise changes.
Conclusion: The best agency client health score is not the most complicated model. It is the one your team updates consistently, understands at a glance and uses to make earlier decisions. Start with five visible components, document the scoring rules, apply intervention thresholds and review whether the score predicted real outcomes. Over time, it becomes a practical early-warning system for retention and a disciplined way to identify expansion opportunities that genuinely fit the client’s needs.