Built an SMS lead service for three clients, lost $6,400 in four months

The three week gap is the expensive part, and it's the part I'd have done identically. Sent count is the number the platform puts in front of you and delivered takes an extra click. Do you know now what triggered the filtering, was it content or volume?

I've only ever been on the receiving end of these texts, and I had no idea a third of them don't arrive. What does the sender see when a carrier filters, nothing at all?

$1,200 to a consultant who set up infrastructure and didn't tell you to watch delivery rates. That's the line item I'd be annoyed about. Scope was probably "get it sending" and it did technically send.

Delivery collapse from 78 to 34 is a normal event, not a freak one, and the number rotation in month three is what most people do and it's the accelerant. Carriers read fast rotation across many numbers as exactly the behavior they're filtering for.

Where I'd disagree with your own diagnosis: the retainer wasn't the mistake. Per-delivered-lead pricing at your volume would have paid you almost nothing in month two, and you'd have shut down in month three with the same total loss and less dignity. The mistake was building a service on a channel whose economics you'd never watched through a bad month. If I were you I'd have run your own list for 60 days before selling anyone a retainer.

@cobalt's last line is the one. Sell after you've survived a bad month, not before.

I looked hard at starting something similar and this thread is why I didn't. The part that reads as the real cause to me is that beacon had no operational metric he trusted, so he was passing a vendor's dashboard through to clients as if it were his own reporting. Once you're an intermediary you have to independently verify the thing you're billing for. That's true whether the channel is texting or anything else.

What happens to the three clients' lists and the records you skip traced, do they get anything back for the $900?

Five years in and the version of this I lived through was a phone dialer, different channel, same shape. My delivery equivalent was connect rate and I watched dials for two months before I understood connects were falling.

The correction that stuck for me: I now send every client a monthly report with three numbers I'd be embarrassed to fake, and one of them is always the one that makes me look worst. It costs me a client a year and it means I've never had to write a $3,000 credit check. @plumb is right that you were reselling somebody else's dashboard. That's the fixable part and the fix is boring.