Billing Model Update: A Story about Learning

For two reasons, which we’ll explore in more detail momentarily, we’ve decided recently to change our billing model.  We feel this is in the best interest of both our client accounts, and our team.

Shilo and I actually started working on this in November, at the indirect request of a couple of our existing clients.  Our all-or-nothing model was great in theory, they argued, but when so much of their seasonal income was riding on just a couple of months, they wouldn’t want us to ease up out of a fear of a $50 overage costing StatBid $5,000.

We were slow to respond to this, however, as we were really attached to the combination of simplicity and transparency it afforded.  It was easy to explain, easy to implement, and ensured we had plenty of skin in the game. “Besides,” we mentioned in a few of these conversations, “slowing down the spend to try to hit a monthly target would be like hitting the brakes in the middle of the corner--it’s already too late, and adding more healthy revenue to the mix is the best way to tilt the scales, not a dribble of especially cheap revenue.”  True enough, right?

And so we kept on trucking.  November turned out alright, but December was starting to give us trouble.  The Black Friday and Cyber Monday behavior across the accounts was dissimilar enough to the previous year that it was mucking with our models, and we took manual control of most of the accounts till mid-January.  But even with all hands on deck, we had a number of narrow misses, that added up to a big hit to StatBid’s internal performance.

Many of these misses were actually smaller than the discount we applied to the invoice, which sounds great when you’re explaining the model to a prospect (or the client receiving the discount), but stings a bit when multiple accounts trip the COS guarantee at the same time--something we hadn't really planned for.

So, we started rebuilding our platform, looking for ways to improve our ability to detect issues, and correct them, more quickly.  Throughout Q1, we’ve made numerous improvements throughout the business, and we’re healthier, and more responsive as a result. However, we started to realize… our clients had been right all along.

If we are running a bit over target at mid-month, for example, it’s true that we’d want to add as much revenue as possible to dilute that expensive revenue. However, imagine we did our jobs perfectly, and earned millions and millions in revenue at just under COS target.  Well, that would reduce the overspend as a percent, but would leave it malignant.  If you’re following a cake recipe that calls for a cup of sugar, but you start with a cup of salt, how many cups of sugar do you need to add before you undo that cup of salt?  You can’t--you’re having a salty cake (or, I suppose, a mountain of slightly salty sugar).

One our our newest team members, who ran paid search for evo for many years, pointed out that if we were internal team members, rather than an external agency  (which we do our best to emulate) narrow misses wouldn’t be considered misses at all, but rather strong performance.  A few dollars over isn’t much different than a few dollars under, after all.  She had a point.

Our unwavering devotion to our core values forced us to run the accounts in a very firm manner, even when giving ourselves a little fudge factor may have been otherwise beneficial to us.  We promised we wouldn’t brake in the corner, and so we didn’t. And while we’re still proud of the honest way we operate, perhaps we were being dedicated to a faulty premise?  Our attempt to avoid divergent motives between client and agency actually produced them!

There’s a lot of noise in this business.  Every metric we work with varies wildly from day to day, and so it becomes a game of averages.  However, in any given month, that noise can determine whether we’re just under target, or just over.  That ends up being, quite frankly, luck, and luck is not what an allegedly data-driven company should be relying upon.

We considered a series of increasingly complicated models, where each day was some kind of data-silo, or with convoluted Markov chains to determine billing.  But none of those could really be used in the real world. Could you imagine assigning Wikipedia articles to clients and prospects just so that they can understand the billing model?   

Our exercise in radical simplicity had failed, and so we went back to the drawing board.  How could we allow ourselves to be aggressive on our clients behalf, without reintroducing the “spend as much as you can” villany that we entered this market to combat?

The answer we came to was a simple set of tiers--a few if/then statements.  If we overspend by X, then Y happens. If we overspend by Z, then A happens.  And that’s what you see on our new pricing page.

For clients, this gives StatBid room to be more aggressive, try more things, rock the boat a little--but without taking our skin out of the game entirely.  We feel the new model does a better job of simulating what it would be like having us on your in-house team than before.

For StatBid, this obviously offers a bit more predictability from month to month.  Till now, we’d been able to grow despite the inherent risk, due to being small and extremely scrappy.  Our team has expanded, and we’ve added full-time staff that’s allowed us to continue to improve our performance and the performance of our accounts.  But to keep growing, and adding more talent to the team, a modicum of stability is necessary. Otherwise, we risk outgrowing our own capacity to provide the quality of service we’re dedicated to providing, and we’d become exactly what disgusts us most--a mediocre agency.  We won't let that happen, full stop.

You pay us for a reason, and our original model (despite its elegant exterior) wasn’t serving either of our needs adequately.  While we aren’t going to force any existing clients onto the new model, we feel that the new version is in the best interest of both our company, and yours.  If a current client disagrees, and would prefer to remain on the original model, we will respect that, as well.

If you have any questions or concerns, I encourage you reach out to myself (roy@statbid.com), or my co-founder, Shilo (shilo@statbid.com) at any time.

Up and to the right,

Roy

 

UPDATE (3/29/18): I think this video helps explain the benefit to the client account, as well: