I get this question in almost identical words every week: I have five thousand pounds, where should I put it. The question is the problem, not the answer. Whoever asked has already decided they are buying advertising, before working out what they are advertising, to whom, and how they will know whether it worked.
This is not a philosophical objection. The sequence is written into the platforms' own documentation, and anyone can open it and read it. The platform tells you plainly that it cannot optimise your campaign until one specific thing exists first, and that thing is measurement.
The platform itself treats measurement as a precondition
Take Target CPA in Google Ads, one of the most widely used bid strategies. The official help page states it directly: "Before you can set up a Target CPA bid strategy, you'll need to set up conversion tracking". That is not advice, it is a gate. Without conversion tracking the strategy does not become available at all.
There is a second sentence on the same page that matters more in practice. To judge performance, Google says measure the last 30 days, including at least 30 conversions. Read that number again and think about what actually happens. Someone launches a campaign, waits three days, and decides advertising does not work for them. By the platform's own stated standard, three days is not data. It is noise.
The same page also says advertisers can start using Target CPA with no conversion history, which is true and worth knowing. But notice the distinction: you can start, you cannot yet judge. Starting is available. Judging requires data.
The growth engine order, and why the order is the whole point
I work in four stages, and each one answers the question the next stage cannot proceed without:
Analysis: who actually buys from you, why, and what stops the ones who do not. Goals: which single number means the business moved. Strategy: the route from where you are to that number. Tactics: ads, content, email, anything that gets executed.
Advertising is stage four. Starting there means asking the platform to answer the first three stages on your behalf, and it cannot. A platform can optimise delivery towards a person likely to perform an event you defined. It cannot tell you the event you defined is the wrong one.
The gap between a marketing goal and a business goal is the first casualty of that confusion, and I wrote it up separately in why marketing goals are not always business goals.
Analysis means a decision, not a report
The biggest misunderstanding about the word analysis is that it describes desk work. It does not. Analysis here means answering three questions with specific answers rather than pages:
Which customer brings the highest margin and the least trouble? What makes somebody buy from you specifically? And which objection comes up repeatedly before a purchase?
Those answers do not come from tools. They come from your own records and from conversations with real customers. If you already have customers, the data exists and you are not reading it. The method I use for defining a market from numbers rather than guesswork is in how to identify your target market from numbers not guesses.
On the technical side, what actually feeds ad delivery now that detailed interest targeting has lost most of its meaning is covered by Mostafa Faried in his piece on Meta targeting signals and what really drives delivery, which is written in Arabic and aimed at people running campaigns themselves.
Goals: the one number that earns its place
If you cannot name a single number that would make you happy if it moved, you are not ready to spend. That is protection rather than gatekeeping, because the platform has to be given an event to optimise towards. Give it the wrong event and it will reach that event efficiently, which is precisely the problem.
I see this constantly: an account optimising for messages or video views, and an owner puzzled that sales have not moved. The platform did exactly what it was told. The four numbers I think are defensible in front of anyone are in how to measure marketing results with four defensible numbers.
When to actually start spending
The rule I use: start spending when you can complete this sentence with no vagueness in it. I am paying money so that [a specific kind of customer] does [a specific event] at a cost below [a number], and I will measure it using [a specific method] within [a specific period].
If the sentence is complete, spend. If one slot is empty, that slot is next week's work. The fuller treatment of what has to be true before the first pound of advertising is in when to start spending on ads and what has to be true first.
The honest limit: analysis becomes procrastination too
There is a second half to this and it would be dishonest to leave it out. Analysis turns into avoidance. I have watched people spend three months on market research because they were afraid to launch. That is the same mistake viewed from the other side.
The practical boundary I use: initial analysis takes a week for a small business, not longer. After that week, any question still open gets answered by a small real spend rather than by more thinking. The market answers questions no amount of research can.
Telling the two apart is easy. If the question can be answered from your own records or from a customer conversation, that is analysis, so do it. If answering it requires people to see an advert and react, that is not analysis, it is an experiment, so run it at the smallest budget that produces a readable result.
So the point is not do not advertise. The point is that advertising is the fourth step, that the platform itself demands measurement as a condition of entry, and that it asks for 30 conversions across 30 days before you draw a conclusion. The order is not an opinion, it is in the manual. The primary source is Google's own page on Target CPA bidding.
You can read this as a decision on its own, but it is really part of a bigger question: where will you win? My book Growth Engine gives that question a whole part, with the six strategies, how they stack, and how to pick between them. It is free, in Arabic: Part 3, Strategies: where will you win.