Consider a mid-size brand consultancy managing 45 active client accounts. Quarterly review time rolls around. The team pulls the CRM report. All but two accounts show green. Those two green-turning-amber flags get attention. Everyone else goes into the next quarter with minimal intervention.
Two months later, three accounts that showed green on the CRM report either downscope significantly or decline renewal. Post-mortems reveal that all three had been cooling for at least two months before the review. But none of that cooling showed up in the CRM, because it was happening in the parts of the relationship that CRM data does not capture.
This is not an exceptional scenario. It is a common one. And it points to a structural problem in how agencies monitor account health.
The one signal that does show up in your CRM
To be precise about what CRM data can and cannot detect, start with what it can. The one signal that CRMs reliably capture is response latency, specifically whether clients are responding to communications more slowly over time. If a client who typically replied within a day or two starts taking a week, that is a behavioral change that can in principle be tracked in a CRM.
In practice, most agency CRMs are not set up to track this at the contact level with any granularity. But the data is there in principle, and for agencies that track email open or response rates, it does surface. It is a real signal. It is also a lagging one. By the time response latency visibly degrades, the relationship has already been cooling for a while.
The four signals below do not show up in CRM data at all, because they live in the qualitative layer.
Signal two: declining specificity in survey verbatims
Most agencies run periodic satisfaction surveys. Net Promoter Score, quarterly check-in surveys, post-project feedback forms. The scores from these surveys go into reports. The open-text responses, the verbatims, are often read by the account manager once and then filed somewhere.
What happens to verbatim quality over the life of a relationship is worth tracking systematically. Early in a relationship, satisfied clients tend to write more specific verbatims. They name particular deliverables, reference team members, describe concrete value. As a relationship cools, the verbatims become more generic. "Good work as usual." "Meeting expectations." "Happy with the team." These responses are not negative, so they do not flag any alert. But the specificity has been replaced with a form of polite distance.
A client who shifts from "the strategic framework you presented in March really reframed how we think about the segment" to "work is meeting our expectations" over three survey cycles has meaningfully disengaged, even if their NPS score did not change.
Signal three: meeting note tone shift
Post-meeting summaries and notes, whether written by the agency team or shared collaboratively with the client, carry a lot of relational information. The language used to describe client reactions, the questions they asked, the topics they engaged with: all of this reflects the quality of the relationship at that moment.
When meeting notes shift from describing substantive exchanges to describing status updates, that is a signal. When the client stops asking questions about the agency's approach and starts only asking about delivery dates, the nature of the relationship has changed. When the client used to bring up adjacent business challenges and now keeps every conversation strictly on the contracted scope, the relationship has narrowed.
These shifts are often subtle in any individual meeting. Across six or eight sessions, they become visible as a pattern.
Signal four: email tone changes
Email tone is a reliable relationship barometer that most account managers read intuitively but do not track systematically. The warmth of an opening, the presence or absence of personal notes, the degree to which the client volunteers information beyond the immediate question: these all reflect where the relationship is at.
A client who starts an email with "Hi [first name], hope the quarter is going well, quick question for you..." is in a different relationship state than a client whose emails open with "Following up on the deliverable due last Tuesday." Both may be perfectly reasonable communications. But the second client has mentally separated their professional regard for the team from their transactional requirements of the contract.
Tracking email tone across time is hard to do manually at scale. It requires reading and annotating each thread. But it is genuinely informative when done.
Signal five: scope ambition declines
In healthy agency-client relationships, there is typically some level of ongoing conversation about what else could be done. Clients who see value from the engagement tend to think about adjacent problems they might want help with. They bring up ideas that are outside the current scope, even if casually.
When clients stop doing this, when conversations become strictly bounded by the existing contract with no exploratory discussion, the relationship has typically moved into a maintenance phase. The client is no longer thinking about the agency as a strategic partner. They are consuming the agreed deliverables and evaluating whether those deliverables justify renewal.
This is not necessarily a sign the relationship is ending. Some clients are just not expansive thinkers. But for clients who used to have those conversations and have stopped, it is worth investigating.
Why the unstructured layer is where the real signals are
The pattern across these four signals is that they all live in unstructured text: survey verbatims, meeting notes, email threads. This is precisely why CRMs miss them. CRMs are designed around structured fields, activities, and scores. They are not built to analyze text for relational content.
This is not a criticism of CRMs. They are doing exactly what they were built to do. But for agencies where the primary value delivered is relational and qualitative, not transactional and measurable, the CRM is measuring the wrong things. It is capturing the shadow of the relationship, not the substance of it.
The agencies that catch cooling accounts in time are typically the ones that have some process for reading the qualitative layer, whether that is an exceptionally diligent account manager, a regular internal review that includes note analysis, or a system that helps them do it at portfolio scale.
If you want to understand how Avara approaches this problem at scale, the product walkthrough covers how we extract signals from the sources your team already produces.