The Uncomfortable Practice Nobody Advertises
Most agencies would not put it this plainly in a sales deck, but a fair number quietly run two different numbers for the same core service: one rate for a brand-new client signing on for the first time, and a different, often lower, rate for a client who has been around for years. Neither number is published side by side with the other.
Framed uncharitably, that looks like an inconsistency waiting to be discovered. Framed accurately, it is closer to a deliberate segmentation decision most agencies just never say out loud.
Why a New Prospect and a Long-Tenured Client Are Not the Same Buyer
A 2025 client-agency tenure study attributed to the ANA and 4A’s, cited in agency-benchmark research, puts average agency-of-record tenure at close to seven years today, more than double 2016’s reported 3.2 years. No direct link to the original study was located in this research, so treat this as a named industry benchmark rather than an independently re-verified number.
Even treated cautiously, a population of clients averaging that much tenure is a genuinely different group from a prospect being pitched for the first time. One has years of demonstrated value, trust, and switching cost behind them. The other has none of that yet, and is actively comparing the agency’s number against competitors it may never think about again.
What Anchoring Explains About Charging Two Different Numbers
Anchoring bias, established by Tversky and Kahneman’s 1974 research, shows that an initial reference point measurably shapes a person’s later numeric judgment. A new prospect’s anchor is whatever number gets stated in the pitch, weighed against competing pitches happening in the same window. A long-tenured client’s anchor is a completely different number: whatever they have been paying, adjusted gradually over years of an ongoing relationship.
Because those two anchors were never the same number to begin with, pricing the two groups differently is not treating one group unfairly relative to the other. It is pricing each group against the reference point that governs their own decision.
Where This Crosses From Segmentation Into a Problem
The practice stops being defensible the moment it becomes concealment rather than segmentation, specifically if a long-tenured client would feel misled to learn a brand-new client is paying meaningfully less for the identical scope of work, with no difference in relationship history to explain it.
The line that keeps this practice legitimate is simple: the price difference should track something real, tenure, proven reliability, reduced onboarding cost, not just an arbitrary split designed purely to extract more from whichever group tolerates it longer.
How Agencies That Do This Keep It From Backfiring
Agencies that run this deliberately tend to keep the new-business number separate from client-facing pricing conversations entirely, so the two rarely sit next to each other where a client could compare them directly. They also tend to keep the gap modest rather than dramatic, which reduces both the odds of discovery and the damage if it happens.
None of that eliminates the risk entirely. It manages it, on the theory that a defensible, tenure-based price difference survives scrutiny in a way an arbitrary one never does.
Is This Value-Based Pricing in Disguise?
There is a more generous reading available. Value-based pricing sets a price according to what an outcome is worth to a specific buyer rather than a single cost-plus number applied uniformly. A long-tenured client has already proven the relationship works, which is itself a kind of demonstrated value a brand-new prospect has not yet earned the benefit of.
Seen that way, a lower rate for a proven, long-tenured client is not really a discount at all, it is a price that reflects a lower-risk, lower-cost-to-serve relationship, priced honestly against what that specific population represents to the agency.
Sources
The external data in this article draws on the sources below. Figures described in the text as estimates or industry triangulations are directional and are not attributed to a single dataset.