▵Measuring success

Three. Not twenty.

The reason is practical: a figure that does not trigger a decision is just busywork. With twenty figures, none triggers a decision any more, because there is always one that happens to look good.

How to find your three: for every figure, ask: what would I do differently if it halved? If nothing comes to mind, you do not need it.

A typical good three for a small business:

1. Enquiries per month.
2. Quotes that turned into orders.
3. Revenue per order.

What these three contain: how many come, how well we convert, what it is worth. Almost every question can be answered with them.

And if you want more: collect them, but look at them once a quarter, not weekly.

Enlarge

The ones that rise without anything changing. Four classics:

Website visitor numbers. They rise with every mention, even if nobody enquires. They only become interesting in relation to enquiries.

Email open rates. Measured via a tracking pixel that many block. The figure is a lower limit at an unknown distance from the truth.

Number of contacts in the system. Grows automatically. Says nothing about quality — rather the opposite, if it is not maintained.

Number of emails sent. Measures effort, not return.

What these four have in common: they are easy to improve without anything getting better. That is exactly how you recognise them.

The counter-test: a good figure cannot be improved without something actually happening.

Long enough for chance to drop out — and for small businesses that is longer than you think.

The rule of thumb: you need enough cases, not enough time. With five enquiries a month, comparing two months tells you nothing. With fifty, it does.

What that means in practice:

• Few cases: think in quarters or half-years.
• Many cases: monthly is enough.
• Always: look at at least three periods before you talk about a trend.

What you have to take into account:

• Season. Compare with the same period of the previous year, not with the previous month.
• One-off effects. One large order distorts everything.

And the most common error in reasoning: reading a direction from two data points. Two points always make a line — that means nothing.

With yourself — not with industry benchmarks from the internet.

Why industry benchmarks are of little use:

• They usually come from providers who want to sell something.
• They are based on other business sizes and other markets.
• They often measure something different from what you measure.
• They are rarely substantiated.

What it makes sense to compare with:

• Your own period from the previous year.
• Two approaches against each other — campaign A against B, with the same audience.
• Your own goal, which you set beforehand.

The last point is the most valuable and the one least often done: write down beforehand what you expect. Afterwards you know whether it went better or worse — regardless of how the result feels.

And the honesty rule: the goal is set beforehand, not adjusted to the result afterwards.

So that afterwards you can say beyond doubt whether it has been achieved.

A usable goal has four parts:

1. What — the quantity concerned.
2. How much — a number.
3. By when.
4. Measured by what — the source of the figure.

Example: “15 enquiries via the website by the end of the quarter, counted from the form messages received.”

What is not a goal: “More visibility.” “Look more professional.” “Improve the website.” None of the three can be achieved or missed.

How high to set it? About 20 to 30 percent above what would happen without any change. Too low is no effort; too high is frustration.

And the most important part: write it down, with a date. A goal not written down becomes, in hindsight, whatever has been achieved anyway.

By enquiries that would not have come without it — and you only recognise those if you ask.

The chain: visibility → visits → enquiries → orders.

What you should measure:

• Enquiries via the form. Countable, unambiguous.
• Calls in which the website is mentioned. Ask: “How did you find us?” and note the answer.
• How many of them turn into orders.

What you can read from this:

• Many visits, few enquiries: the page does not answer the question, or the wrong people are coming.
• Few visits, many enquiries: the page is good, it just is not being found.
• Few of both: start with visibility.

And the simplest evaluation there is: a tally sheet next to the phone, for three months. It is more accurate than any analytics tool, because it captures the calls.

By asking — consistently, every single one, and writing down the answer.

Why asking is better than measuring: the paths are winding. Someone sees a vehicle, later searches for the name, finds the website, asks an acquaintance and then calls. An analytics tool sees a fraction of that.

How to ask: “May I ask how you came across us?” Ideally at the start of the conversation, before it gets down to business.

How to record it: a field on the contact, with a few fixed answer options. Free text cannot be evaluated.

What you know after three months: which channel actually brings customers, and which is only effort.

And the typical surprise: recommendation is almost always the strongest channel, and almost never the one that is invested in.

By the cost per customer won — and your time counts too.

The calculation:

Effort for the channel (money plus working time) divided by the number of customers won through it.

What comes out of this usually surprises people:

• A channel that costs nothing but eats up a lot of time is often the most expensive.
• An advertisement that seems expensive can be cheaper than three days of cold calling.

What you should add: not every customer is worth the same. A channel that brings cheap one-off customers is worse than one with few but lasting ones.

So calculate twice: cost per customer and value per customer. Only the two together give a picture.

And the time trap: your own effort is almost always underestimated. Estimate honestly, or better: measure it for a week.

With a simple comparison, and the discipline to write it down beforehand.

Beforehand:

1. What does it cost, one-off and ongoing?
2. What is it supposed to bring, in a figure?
3. By when?
4. What is my exit point if it does not work?

Afterwards:

Compare. Without changing the figures from beforehand after the fact.

What point 4 is worth: it prevents the most expensive pattern of all — continuing to invest because so much has already been invested. What has been spent has been spent; it is no argument for the next expense.

What you should include in the calculation: your own time, the effort of learning it and ongoing operation.

And what you should not forget: some things pay off without being measurable. A website that stops someone thinking you are not serious has no figure attached, and still has an effect.

With four rules, all of them uncomfortable:

1. Write it down beforehand. Goal, figure, deadline — with a date. A goal that only comes into being after the result is always achieved.

2. Always look at the same figure. Whoever changes the yardstick of success always finds one that fits at the moment.

3. Look at what did not work too. The reflex is to look at the good figures. The bad ones contain the information.

4. Ask someone who is allowed to disagree. Ideally someone who was not involved.

The most common self-deception: changing the reference figure. “We have 30 percent more opens” sounds good — until someone asks how many enquiries came of it.

And the second most common: attributing successes to your own approach and failures to circumstances. With every result, check the other explanation too.

Yes — with one important distinction: figures about the work, not about individual people.

What works well:

• Shared figures that concern everyone.
• Regularly, in short form.
• With an explanation of what they mean and what they do not.
• With the question: what do we conclude from this?

What works badly:

• Rankings of individual people. They create behaviour that improves the figure, not the result.
• Figures without context.
• Figures that are only shown when results are bad.

What to bear in mind legally: evaluations that allow conclusions about the performance of individual employees can be subject to co-determination. Clarify this before you introduce anything like that.

And the best effect: when the team knows the figures, suggestions for improvement come from where the work is done.

Then do not count; look at each case individually. With small numbers that is even better.

What you do instead:

• Review every lost enquiry. Why did it not come to anything? With five a quarter, that is an hour’s work and worth more than any evaluation.
• Do the same for every order won. What tipped the balance?
• Look for patterns, not percentages. If four out of five refusals mentioned the price, you know something — without statistics.

Better not: percentages with small numbers. “Our conversion rate has risen by 50 percent” means, with four cases: one more.

And the advantage of small numbers: you can know every case. Large businesses cannot — that is why they need statistics.

Little — and if it becomes more, you are measuring the wrong thing.

A good guideline: half an hour a month for the ongoing figures, half a day a year for the big picture.

What breaks this framework:

• Figures compiled by hand from several sources.
• Evaluations nobody reads.
• Tools whose maintenance costs more time than the insight is worth.

How you notice that it is too much: when you postpone measuring because it is too laborious. Then it is too laborious, and a simpler procedure that happens is better than a good one that does not.

What is almost always enough:

• Three figures, noted monthly.
• One question with every enquiry: where did it come from?
• Once a year, an honest look back.

And what costs nothing: writing down what you set out to do.