A maternity ward keeps a log of the number of boys and girls born on a given day. Which of these is more likely to be a real record from the log?
Boy Girl Girl Girl Boy Girl Boy Boy Boy Girl
Boy Boy Boy Boy Boy Girl Girl Girl Girl Girl
When it comes to PPC account management, we’ve all had clients like it, and you can’t really blame them. They’re spending a lot of money on Google ads, and if it all seemingly goes wrong, their job could be on the line. So they call and send emails, with perfectly reasonable questions:
“We’ve had a really good week. What are you doing differently? Can you keep doing it?”
Or, more frequently it seems:
“We’ve hardly sold anything this week. Our return on investment is terrible. What’s going wrong?”
So, you frantically rack your brain for ideas (Brexit?, the election? Post-Black Friday lull?) and for quick fixes, desperately redistributing budgets to restore profitability.
Stop. Stop now. Accounts suddenly and inexplicably hit bad patches. Then the bad patches end. Sometimes there are reasons behind this, but often there simply aren’t.
Randomness, Cause and Effect and PPC
Humans are awful at randomness. We look for events to have causes. If you show people a clip of featureless circles, moving randomly around a screen, they will quickly assign motivations and personalities to them, describing them as hunters or bullies or layabouts. The ‘uncaused cause’ has for centuries been used as an argument for the existence of God (essentially ‘the universe had to be caused by something, so that something must be God’).
Lucia de Berk, a Dutch nurse, was unlucky enough to be followed around by a series of deaths on her ward, which would only occur naturally one time in millions. Which sounds damning until you realise that the NHS employs a third of a million nurses. So in the UK, clusters of deaths like this could come up every three years or so. The judge, however, was no great statistician and she was sentenced to life imprisonment.
How A Client’s Brain Works and Challenging Perceptions
Some things simply don’t fit into the cause and effect pattern. Some things are random. Whether the next child born is a boy or a girl is independent of the previous child’s gender. Both of the sequences of births listed above are equally likely.
While I can process that on an intellectual level, on a gut level randomness just isn’t satisfying. I’d be avoiding the ward where that poor, unlucky nurse was working. I can’t blame the client for wanting a better explanation – ‘it’s just a bad patch’ isn’t a very convincing story. So how can we deal with this flaw in our brains?
Managing Expectations With Marginal Gains and Patience
Managing expectations is important. Talk about the long term plan with clients, and focus on long term results. There’ll always be clients who call, insisting you explain why we haven’t sold any Size 7 widgets this week, when we sold five last week.
Be patient, ask them to look at the whole month’s data, and make them aware of past fluctuations. Put things in the context of previous bad weeks.
Tell them “In an average week, we have three good days and four slow days. This week we’ve had five slow days in a row.” Show them data day-by-day, as well as month-by-month – the impact can be quite different, as the revenue data below shows:
If a client hits a good patch and sales skyrocket, resist taking too much credit for yourself. Point out that there are many factors which determine sales, and what you do is only part of it.
Talk about marginal gains: slow, incremental changes which will improve the profitability of the account over time. And when you do, inevitably, have a period when return on ad spend is low, don’t panic and overreact. Making desperate, disproportionate changes to bids and budgets and modifiers and targeting isn’t going to make anything better in the long term.
Slow down. Don’t wait it out forever, of course, if the evidence starts to build that something really has gone wrong, you’ll need to do some digging into the details and work out why everything’s changed before you can confidently set about fixing the problem.
And it pays off in the end. It takes time and patience, but most clients will eventually come to accept day-to-day fluctuations in account performance as par for the course. Even that poor bloody nurse was released after just seven years. There’s always hope.