Proving retention ROI without claiming more than the data supports

"Is this working?" is the question behind every retention retainer review, and it's rarely the real question. What a client is usually asking is: would we be selling roughly the same amount without you? That's a much harder question, and pretending attribution data answers it is where a lot of agency reporting quietly overreaches.

The question behind the question

A client comparing your retainer cost to your reported results isn't really doubting that the flows exist or that they converted — they're asking about the counterfactual: what would have happened anyway. Attribution models don't measure counterfactuals. They measure which messages were present near a conversion event, which is a different and much more answerable question.

What attribution can and cannot answer

Attribution can tell you, precisely, what Klaviyo credited to a given flow under its model. It cannot tell you what would have happened if that flow didn't exist — that's an incrementality question, and answering it properly requires a controlled experiment (a holdout group that doesn't receive the flow, compared against one that does), not a reporting dashboard. Saying this plainly to a client is a stronger position than implying otherwise: it's the difference an agency that understands its own numbers, versus one hoping nobody asks the follow-up question. It's also why this isn't something Inteleve calculates — uplift and incrementality are a different, harder measurement problem than attributed performance, and conflating the two erodes trust in both.

The honest ROI framing

Retainer cost against Klaviyo-attributed revenue is a legitimate ratio to report — as long as the caveat sits next to the number, not buried in a footnote: "For every $1 of retainer, Klaviyo attributed $[X] in flow conversions this quarter. This reflects Klaviyo's attribution model, not a controlled measurement of incremental revenue." That sentence is longer than most agencies want to write. It's also the one that survives a CFO reading it twice.

Three artefacts that work

Three things build a stronger ROI case than a single blended ratio: the account's own revenue-per-recipient trend over time (its own baseline, not an industry one), one or two concrete flow results with period, metric and source stated plainly, and a visible record of the work actually done — tests run, flows built, segments refined — regardless of whether every test moved a number.

The sceptical client script

Four questions worth having a ready answer for: "How do you know this wasn't going to happen anyway?" (you don't, precisely — explain the attribution model honestly). "Why doesn't this match my own analytics?" (different attribution models, not an error — see the difference between Klaviyo and other platforms). "What if you stopped working on this account?" (point to the specific, ongoing work, not just the historical number). "Can I see the raw data?" (yes — a verifiable page beats a static deck every time).

Making it verifiable

The most convincing thing an agency can do isn't a bigger number — it's a page the client can open themselves, with the period, the metric and the source all visible on it, so the client isn't taking the number on the agency's word alone. A result that can be independently checked is worth more than one that has to be trusted.