27 Things to Do, and 19 of Them Name the Number to Move

They hand you a dashboard. The dashboard is beautiful. It has sparklines.

And you sit there thinking: yes, but what do I do on Monday?

So we made the tool answer that, and then we made the answer prove itself.

On one brand this week there are 27 recommended actions. Fine, lots of tools generate recommendations, usually by string-substituting your competitor's name into a template.

Here is the part I actually care about.

19 of those 27 say which number they are supposed to move, and when we will check.

Not "improve your content strategy". A specific metric, a direction, and a date. Which means in six weeks you can go back and see whether the advice was any good, including the advice that was not.

The other eight do not carry a number, and they say so, in writing, with the reason. Some things worth doing genuinely cannot be measured cleanly, and pretending otherwise by attaching a made-up KPI is how these tools lose your trust. An action that admits it has no measurable outcome is more honest than one that invents one.

There are 40 rules behind all of this, and they only fire when the underlying data actually supports them. No data, no card.

Then there is the thing that started as a small internal argument and ended up as enforced code.

At one point our advice mentioned two tools by name, four times, with no stated relationship. Four people testing the product read it as hidden advertising. There was no relationship. Did not matter. Once a reader suspects one line is paid for, they suspect all of them, including the numbers.

So now there are rules a check enforces before anything publishes. One tool per action, maximum. The free way to do it comes first, and the action has to contain real steps before a tool can be named at all. Disclosure sits inside the sentence, not in a footnote. And no tool is ever named next to a number in a report section.

The gap I will admit to: we can tell you which number an action should move, but we cannot yet verify that it moved. That measurement loop is half built, and until it is finished, this is a well-designed promise rather than a proven one.

One thing to do next: take the last recommendation any tool gave you.

Ask it which number that was supposed to move. If it cannot say, you have learned something about the tool.