Project the ROI Before You Build

Happy Wednesday!

A brand came to me a few months ago and asked us to rebuild their retention flows. I looked at the account for twenty minutes and told them not to hire us.

Not because the work was wrong. Because nobody had run the numbers on what the work would make — including me, until I sat down and did it.

That's the actual mistake in this story, and it isn't a small-brand mistake. It's the one I see at every size. Somebody decides to build a thing, and the decision gets made on a gut feeling, or on what everyone else does, or on "our welcome flow is embarrassing." Nobody runs the numbers first.

Every build should have a number attached to it before a single email gets designed. If you can't figure out what it'll make you back, you're not making a decision. You're just guessing.

What the account actually looked like

Roughly 2,500 sessions a month. Campaigns making almost nothing. Flows quiet.

Not broken. Not set up wrong. Just quiet, because there was nobody moving through them.

So I ran the numbers on the rebuild before I gave them a price.

The math

Here's the thing people skip: a rebuild doesn't earn what the whole flow makes. It earns the difference between the new flow and the old one. That difference is the whole ballgame, and I'll come back to it.

Start with what the existing welcome flow can possibly do.

(Illustrative numbers, but they're the ones I'd defend. Move any of them a point in either direction and nothing about the answer changes.)

Now say we rebuild it and it's really good. Say we double what it makes. That's +$2,300 a year — against a build fee several times that. It doesn't pay for itself. It doesn't come close.

And here's the part that mattered more to me than the invoice: it doesn't change anything about where the brand is headed. No version of that flow changes their path.

Same brand at $50k a month, though — 25,000 sessions, same rates — is looking at ~$23,000 a year out of that one flow. Double that and the exact same work makes ten times as much. The work didn't get better. The traffic underneath it got bigger.

The part big brands get wrong

If you're doing $200k a month, you just read all that and thought sure, but that's not me. It might be.

Plenty of traffic protects you from one specific mistake — building a retention system on top of no traffic. It doesn't protect you from the more common one, which is building big things that nobody ran the numbers on.

Take a collection-based split on your welcome flow — a separate version for each product collection. Everyone wants one. Run the numbers on it first.

Say 750 people enter that welcome flow every month, and at these same rates the flow makes about $23k a year. A split might add 8% on top of that — about $1,800 more a year. And it's a build you maintain forever, because every future edit now happens in three places instead of one.

Then it gets worse. The split cuts those 750 people into three groups of 250. At a 4% order rate, each group makes about 10 orders a month. An 8% lift on 10 orders is less than one extra order — too small to spot against normal month-to-month swings. You will never be able to prove the split did anything. Not this quarter, not this year.

So you spent the hours, added the upkeep, and bought yourself a result you can't even measure. You'll just have a more complicated account.

That's the exact problem that first brand had. Different budget, same mistake: nobody figured out the difference before the build started.

My $30k–$50k rule of thumb

For what it's worth, my rule of thumb is that retention work starts earning its keep somewhere around $30k to $50k a month, steadily — a three-month average that holds, not one big month off a viral post.

But treat that as a shortcut, not a rule. It's a rough way to guess "will this build pay for itself," and rough guesses only work at the low end. Once you're above that cutoff, you still have to run the numbers on every build. The cutoff only tells you retention is worth doing. It doesn't tell you what to build.

The boring part

So I told that brand the truth: fix your traffic and your site first. Get people to the site, get them buying once. Then retention has something to multiply.

Those brands stuck with me. A lot of the ones I told "not yet" first came to me at $5k a month, went and did the boring work, and are well past $50k a month now, as paying clients. Turns out being honest that the math doesn't support the bill is a good way to earn the next five clients.

Grow your traffic first. Then we'll multiply it.

If you want, I'll run the numbers with you — your triggers, your rates, your order value — and we'll find out together whether your traffic is big enough to make a rebuild pay off.

Sometimes the answer is no. That's a useful answer, and it's free.

— Raymond