You are reviewing your best testimonials before a homepage refresh, and one of them stops you cold. It is a great quote — specific, credible, from a recognizable customer — and the entire thing is about a feature you have already decided to deprecate. Maybe you are folding it into a broader workflow, maybe it is being replaced by something better, maybe it is quietly being retired because almost nobody used it. Either way, in three months the thing this customer is praising will not exist in the form they describe.
Now you have a small but real dilemma. Leave the quote up, and you are showcasing a capability a new buyer might sign up expecting — and not find. Pull it, and you lose one of your most persuasive pieces of proof over a detail most readers would never notice. Neither the "leave it" nor the "delete it" reflex is right on its own. The correct move depends on what the quote is really doing on the page.
First, separate the feature from the outcome
Read the testimonial twice. The first time, notice the feature it names. The second time, ignore the feature entirely and ask: what outcome is this customer actually crediting you with?
Almost every good testimonial is a two-layer object. There is the surface layer — the specific feature, screen, or workflow the customer points at — and there is the underlying layer — the result they got, the problem that went away, the thing they can now do that they could not before. A quote like "the automated weekly digest saved my team the Monday scramble" names a feature (the weekly digest) but sells an outcome (no more Monday scramble).
This distinction decides everything. If the outcome survives your deprecation — if the customer will still get "no more Monday scramble" through whatever replaces the digest — the testimonial is salvageable. If the outcome itself disappears with the feature, the quote is genuinely stale and should come down. Most of the time you are in the first case, and the fix is lighter than you feared. This is the same "read past the adjectives to the underlying claim" discipline we apply to contradictory testimonials, and the reason attribution details like job title and company matter — the surface wording is rarely the real evidence.
The four situations, and what each one calls for
1. The outcome survives; only the feature name changes. You are renaming or absorbing the feature, but the customer still gets the same result. Here you have two honest options. You can keep the quote as-is with a light date stamp (a testimonial is a historical record of a real experience, and dating it makes that explicit — see should you put a date on a testimonial). Or, better, you go back to the customer and ask whether they would refresh the quote to describe the outcome rather than the retired feature. A customer who was happy enough to give you a quote is usually happy to update it, and an outcome-framed testimonial ages far better than a feature-framed one.
2. The outcome survives, but through a genuinely different path. The Monday scramble is still solved, but now by a live dashboard instead of an emailed digest. Keeping the old quote prominent would set a false expectation — a buyer reads "weekly digest," signs up, and finds a dashboard. Do not display this one against the feature it names. You can still use it as general proof ("saved my team hours every week") in a place where the mechanism is not the point, but strip or soften the specific feature reference. The rule: a testimonial should never make a promise your current product cannot keep.
3. The outcome disappears with the feature. You are retiring the capability outright and nothing replaces the result. This quote is now describing a product you no longer sell. Retire it from all live placement. You do not have to delete the record — it is a real thing a real customer said — but it has no business on a page a prospect will read, because it is proof of value you have chosen to stop delivering. If several of your strongest quotes are anchored to a feature you are killing, that is worth a second look at the deprecation decision itself; strong, specific praise is a signal that the feature mattered to someone.
4. The feature is the whole reason they bought. Occasionally the deprecated feature is not incidental to the testimonial — it is the testimonial, and it is also, quietly, the reason that customer chose you. Before you touch the quote, that is a product-and-retention conversation, not a copy one. Reach out to the customer directly. You may learn the feature has a small but devoted base you were about to alienate.
Do not let the copy outrun the product
The failure mode to avoid is leaving feature-specific praise on a high-traffic page long after the feature is gone, because nobody owns the job of reconciling testimonials with the roadmap. It is an easy gap to fall into — testimonials feel like permanent assets, while features change constantly, and the two are almost never reviewed together.
Build the check into your deprecation process. When a feature is scheduled for sunset, one line on the checklist should be: "Which testimonials reference this, and what happens to each?" If you tag testimonials by the feature or outcome they describe — the same metadata hygiene that lets you ask a customer to refresh an outdated testimonial without guesswork — this becomes a thirty-second query instead of an archaeology project. The point is not to police your social proof into blandness. It is to make sure every quote on your site is proof of something your product still does, so that the confidence a testimonial builds is confidence a new customer will actually get to keep.
The one-line rule
A testimonial can outlive the exact feature it praises, but it must never outlive the outcome it promises. Keep the quotes whose results you still deliver, refresh the ones whose mechanism changed, and retire the ones whose value you chose to stop shipping — and your social proof stays honest through every roadmap change.