How to Measure Client Health in a Digital Marketing Agency
A practical framework for agency client health scoring: which SEO and PPC signals to combine, how to weight them, and how to act on the result.

Why agencies need a health measure at all
Most agencies already report on performance. The problem is that reports are written one client at a time, once a month, and read after the fact. By the time a downward trend is visible in a monthly deck, the conversation with the client has often already changed.
A client health measure is not a replacement for reporting. It is a way of summarising many signals into one comparable view, so account leaders can see across the whole portfolio which relationships need attention first.
What a useful client health score combines
A health score should reflect the outcomes the client actually hired you for. For most search and paid media engagements, that means combining:
- Organic visibility: ranking coverage, keyword movement and Search Console impressions and clicks.
- Organic outcomes: sessions and, where tracked, conversions from organic search.
- Paid efficiency: cost per conversion, conversion rate and CTR relative to the account's own history.
- Direction of travel: whether the last few periods are improving or deteriorating.
- Progress against the starting point of the engagement.
Compare clients to themselves first
A local service business and a national ecommerce brand will never have comparable absolute numbers. The fairest approach is to normalise each metric against the client's own baseline and recent history, then combine those relative changes.
This is the idea behind measures such as a Baseline Progress Index (progress relative to where the campaign started) and a Recovery Index (recovery relative to a previously observed low).
Weighting SEO and paid media
If a client buys only SEO, paid media signals should not drag their score. If they buy both, weight each channel by its share of the engagement or its importance to the client's goals. Keep the weighting simple and documented — a score nobody can explain will not be trusted by account teams.
Turning a score into action
A score is only useful if it changes what the team does next. A simple operating rhythm works well:
- Review the portfolio weekly, sorted by lowest health and largest negative change.
- Assign an owner and a next step to every client on the watchlist.
- Record notes so the reasoning behind decisions is not lost between meetings.
- Revisit after the next data period to see whether recovery has started.
Limits worth being honest about
Health scores describe performance signals; they do not measure client satisfaction directly and they cannot guarantee whether a client will stay. Relationship context, budget changes and business events outside search matter too. Treat the score as an early prompt for a conversation, not a verdict.
Worked example: a mixed-channel account
Consider a synthetic account whose organic clicks fall 18% while paid conversions hold steady. A combined score may look stable, but the channel split exposes an issue worth investigating. First confirm date ranges and tracking, then compare priority landing pages and search demand. Do not change campaign strategy solely because a composite score moved.
Document what the score includes, which sources are missing and how engagement weights were chosen. Review the account with its owner and record the next investigation. This is a suggested operating method, not Client Guardian's published scoring formula.
Start with a measurement contract, not a dashboard
Before assigning weights, write down the decision the score will support. An account lead choosing this week’s investigations needs something different from a finance team estimating renewal revenue. This guide concerns campaign-performance attention: where results are deteriorating, where progress has stalled, and which accounts need a closer conversation. It does not turn performance measurements into a probability of cancellation.
Create a short measurement contract for each engagement. Record the business objective, the channels included, the conversion definition, the reporting timezone, the baseline period, and the person responsible for checking data quality. “Generate qualified enquiries” is more actionable than “increase traffic”, but it still requires agreement on what qualifies an enquiry and when the sales team will return that information.
Keep the contract versioned. A conversion action changing from form submission to qualified lead creates a break in the series. A redesigned site, a changed budget, or an expanded keyword set may also change comparability. Mark those changes before interpreting a lower score as weaker execution. Otherwise a sensible improvement to measurement can look like a sudden performance failure.
Build a scorecard with observable components
Separate outcomes, efficiency, visibility, and context. Outcomes describe useful activity such as qualified leads or sales. Efficiency connects those outcomes to cost. Visibility describes opportunities to be seen, including search impressions and priority-query coverage. Context explains whether the data can support a decision. A tracking failure belongs in context, not in the same bucket as a genuine collapse in lead demand.
For an SEO engagement, choose a small set that fits the objective: non-branded clicks to commercial pages, relevant conversions when reliable, and visibility for an agreed query group. For paid media, qualified conversion volume and acquisition cost may be more useful than click volume. Do not include every available metric merely because the source exports it. Highly correlated inputs can count the same problem twice.
Search Console’s Performance report defines clicks, impressions, CTR, and position within Google Search. Google Ads conversion measurement depends on configured actions and settings. Those are measurement definitions, not business-value guarantees. Use the official references below to agree what each source actually represents, then document the gap between a platform conversion and the client’s commercial outcome.
| Layer | Example input | Question to ask |
|---|---|---|
| Outcome | Qualified enquiries | Is the qualification process consistent? |
| Efficiency | Cost per qualified enquiry | Are spend and outcomes from comparable periods? |
| Visibility | Non-branded commercial clicks | Did demand or ranking coverage change? |
| Confidence | Tracking and source freshness | Can the team trust this comparison? |
A transparent normalisation method
The following calculation is an illustrative agency method, not Client Guardian’s scoring formula. For a higher-is-better input, divide the current value by the agreed baseline and multiply by 100. A baseline of 200 qualified organic enquiries and a current value of 180 gives 90. For a lower-is-better input, reverse the ratio: baseline acquisition cost divided by current acquisition cost, multiplied by 100. A rise from $40 to $50 gives 80.
These ratios are progress indices, not automatically bounded health scores. A value of 120 means twenty percent above the chosen reference for that input; it does not mean 120 percent healthy. If the agency wants a zero-to-one-hundred score, define and disclose a mapping with sensible limits, including what counts as meeting the objective. Do not silently clamp everything above baseline to 100 and then imply further improvement did not happen.
Zero needs special handling. A baseline of zero makes the usual ratio undefined. A current acquisition cost cannot be calculated when there are no conversions. Treat those cases as insufficient evidence or a distinct no-outcome condition. Show the underlying spend and volume instead of inventing a ratio. Small counts also produce dramatic movements: one additional lead can double a total of one without establishing a durable trend.
Worked calculation: weights without hidden assumptions
Imagine a synthetic engagement with already mapped channel-health values of 70 for SEO and 85 for paid media. The agreed engagement weights are 60 percent SEO and 40 percent paid. The illustrative combined value is (70 × 0.60) + (85 × 0.40) = 76. The result is easy to reproduce, and the team can still see that organic performance needs more attention than the overall number suggests.
Now suppose paid-media data is unavailable because an export failed. Do not substitute zero and report 42 as if campaign performance collapsed. Show the organic result of 70, mark the paid component unavailable, and label the combined result incomplete. If you reweight available components for a temporary view, state that rule explicitly and prevent comparison with an earlier full-coverage score as though nothing changed.
Run a sensitivity check before adopting the weights. With equal channel weights, the same inputs give 77.5; with 80 percent SEO and 20 percent paid, they give 73. If a small weighting adjustment reverses the order of two clients on the watchlist, their apparent difference is fragile. Review them together rather than letting a decimal point dictate which client deserves attention.
- SEO 70 × 60% = 42
- Paid 85 × 40% = 34
- Combined = 76
- Review weaker channel
Choose periods and baselines that survive scrutiny
A baseline should be representative of the engagement starting point, not selected because it produces an attractive improvement story. Use complete periods and preserve the original reference. If the starting period contained a tracking outage or an exceptional promotion, explain why another reference is needed. Keep both the old and replacement basis available so the team can distinguish a methodology change from a performance change.
For seasonal accounts, compare recent complete periods with both a rolling reference and a comparable seasonal period where the history exists. Keep weekday composition in mind: a short window containing more weekends can look weaker for a business-to-business account. Never compare a partial current month with a complete previous month without a clear partial-period adjustment and caveat.
Recovery deserves its own explanation. Moving from a low of 40 to 60 is an improvement of 20 points, but the account may still be below an earlier baseline of 80. Show the current level, the low, and the baseline together. A Recovery Index can help describe direction, while a Baseline Progress Index can describe distance from the starting reference; neither should hide the absolute business outcome.
Calibrate attention bands with human review
There is no universal threshold that makes an account healthy. A value of 65 cannot mean the same thing across every scoring method, industry, and campaign stage. Begin with descriptive bands linked to documented review actions, then evaluate whether those bands actually identify worthwhile investigations. Avoid presenting suggested cut-offs as industry standards when they are internal operating choices.
Use a rehearsal period. Let account leads review the proposed watchlist alongside the raw metrics and their relationship notes. Record false alarms, missed problems, and recurring causes of uncertainty. If a high-spend client with poor lead quality remains green, the score may be measuring platform efficiency instead of the agreed objective. If every seasonal account turns red together, the comparison window may need revision.
Keep performance and relationship evidence adjacent but distinct. A client can have improving campaigns and still be concerned about responsiveness or budget. Another may accept weak short-term results during an agreed rebuild. Record the concern in plain language with an owner and review date rather than pretending that a single composite number captures the entire relationship.
Frequently asked questions about client health
How often should the score be reviewed?
Use a weekly account-prioritisation meeting when the source cadence supports it, and a deeper monthly review of assumptions and outcomes. Low-volume campaigns may need longer evidence windows even if the team checks for technical failures more frequently. Refreshing a screen daily does not make yesterday’s conversion count statistically persuasive.
Should every client use the same weights?
Use the same documented method where possible, but not identical channel weights regardless of scope. An SEO-only client should not be penalised for missing paid-media activity. Compare accounts with awareness of coverage, confidence, and objectives; use the score to order investigation, not to rank client worth.
Can this framework predict churn?
No. These measurements describe observed campaign signals and attention needs. A churn probability would require a separately validated model, clearly defined renewal outcomes, and evidence about prediction quality. A watchlist is a practical review mechanism, not proof that a client will leave.
References & further reading
Sources for further context—not evidence of a product integration.
Keep the thought going.
Related readingClient Retention Intelligence: From Reporting to Account-Level Decisions
How agencies can move from monthly reporting to account-level decisions that support client retention, using health indicators and risk watchlists.
Marketing Performance Intelligence vs. Traditional Marketing Reporting
The difference between traditional marketing reporting and performance intelligence, and why agencies need both to prioritise their work.
SEO Performance Monitoring: Metrics, Trends and Early Warning Signs
Which SEO metrics to monitor across client accounts, how to read trends rather than snapshots, and the early warning signs worth acting on.


