Most people who come to me have already spent the money. They have run the ads, paid the agency, watched the budget go out and very little come back. By the time we speak they are not looking for a strategy. They are looking for results, and they would like them by the end of the month.

I understand that completely. I am still going to tell you no.

Not because the work is slow, and not because I am padding a contract. It is because three separate things have to happen before scaling makes any sense, and each of them takes real time that nobody can compress. You have to analyse. You have to test organic. And you have to let the machines learn.

Here is what each one actually costs you in weeks, using Google’s own published guidance rather than an agency’s optimism.

Stage one: you do not have trustworthy data yet

Before anything gets fixed, you need to know what is true. That is harder than it sounds, because the numbers in front of you are usually wrong in ways nobody has noticed.

Tracking fires twice, so every conversion is counted as two. Or it does not fire at all on mobile. Or the feed has been rejecting half the catalogue for months and nobody opened the notification. I see all three regularly, and each one makes every report downstream meaningless.

Getting to clean data takes longer than the audit itself. Google Analytics needs a day or two just to finish processing what it has collected. Beyond that, conversion reporting is not complete until the conversion window has closed, so a report ending yesterday is still filling in behind you.

The practical floor is around 30 to 90 days of clean, verified data before a baseline means anything. Not 30 to 90 days of data. Thirty to ninety days of data collected after the tracking was fixed. Everything before that is a record of the fault, not a measurement of the business.

This is the stage people most want to skip, and it is the one that makes every later stage possible.

Stage two: fixes take time to be recognised

You fix the thing. Google does not immediately notice.

When you ask Google to recrawl a page, its own documentation says plainly: “Crawling can take anywhere from a few days to a few weeks.” It adds, just as plainly, that “requesting a crawl does not guarantee that inclusion in search results will happen instantly or even at all.”

That is Google, in writing, refusing to give a timeline. Anyone promising you one is making it up.

Merchant Centre is the same. A product review after a fix runs to several business days, and the ranges Google publishes for different review types do not entirely agree with each other. Once approved, changes can take up to a further day to show in Shopping. Structured data fixes need a validation period that Google puts at two weeks or more.

So a fix made on the first of the month may not be visible in the results until well into the second. During that gap the numbers will look unchanged, and it will feel like nothing is working. Nothing is working yet. That is different.

Stage three: organic has to be tested, and it is slow by design

Organic is where expectations and reality diverge most.

Google’s own guidance talks in terms of a few hours to several months depending on what changed, and advises waiting weeks before assessing anything. Technical fixes can register quickly. Genuine improvement in content and authority takes, in Google’s own words, several months to confirm.

There is also a measurement problem. Google’s core updates roll out over one to two weeks or more, in irregular stages. If your test window is shorter than the rollout, you are not measuring your change. You are measuring the weather.

And the bar has risen. Ahrefs’ 2025 study of a million URLs found that only 1.74% of new pages reach the top ten within a year, and that the average page ranking first is now around five years old. Organic is a compounding asset. It is not a first-quarter deliverable, and anyone selling it as one is selling you something else.

Stage four: the machines have to learn, and you cannot rush them

This is the part clients find hardest to accept, because it feels like the ads are just sitting there doing nothing.

Google Ads’ documentation on the learning period states: “It can take up to 3 weeks or 1-2 conversion cycles for the bid strategy to calibrate to the new objective.” That clock starts when a strategy is created, when a setting changes, and when campaigns, ad groups or keywords are added or removed.

Read that again, because it has a consequence people miss. Every time you get impatient and change something significant, you restart the clock. Three weeks of learning, interrupted in week two by a well-meaning tweak, is not three weeks of learning. It is two weeks, twice.

Performance Max is more explicit still. Google’s own optimisation guidance says: “Run new campaigns for at least 6 weeks. This allows the machine learning algorithm to ramp up and gather sufficient data to optimize performance.”

Six weeks before you are entitled to an opinion.

Then there is volume. Google recommends measuring performance over periods containing at least 30 conversions, and at least 50 for Target ROAS. If your account produces twelve conversions a month, you do not have a performance problem in month one. You have a sample size problem, and no amount of optimising will fix that. You need more time, or more volume, before the numbers mean anything at all.

Why this cannot all run in parallel

The obvious objection is: fine, but run them at the same time.

Some of it you can. Auditing analytics, the feed, and the site structure are separate systems with no dependency between them, so those happen together in week one. Fixing tracking while fixing feed errors is sensible.

But the sequence has hard joins in it, and they are joins of logic, not of scheduling.

  • You cannot let a bidding algorithm learn from conversion data that is wrong. It will learn the wrong thing confidently, and then you will scale spend behind a machine that has been trained on a fault.
  • You cannot judge an organic test inside a re-index lag that has not finished. The change is not live in Google’s view of your site yet.
  • You cannot trust a feed fix inside its own review window. It has not been reviewed.

Each stage’s clock only starts when the one before it is genuinely finished, and “finished” means evidenced, not asserted.

What going too fast actually costs

It is not that scaling early merely wastes the increase. It corrupts the thing you were trying to build.

Google’s own troubleshooting documentation describes the mechanism: when conversion tracking drops out, automated bidding is optimising towards a signal that has disappeared, and serving and spend drop with it. The system is not broken. It is faithfully chasing something that is no longer there.

Scale spend on top of broken tracking and you get the worst outcome available. You spend more, you learn nothing, and you end the quarter with a pile of data you cannot use to make the next decision. That is how businesses arrive at my door having spent a year’s budget with nothing to show for it, which is very nearly always the actual story.

What a realistic first six months looks like

  • Month one. Diagnose everything in parallel. Fix tracking, technical SEO and the feed. No scaling. Nothing here is judged, because nothing here is measurable yet.
  • Month two. Fixes propagate. Re-crawls work through, Merchant Centre reviews clear, Search Console catches up. Ads can run at conservative budgets, but they are inside the learning window and are not to be judged.
  • Month three. First real reads, still provisional. If conversion volume has cleared Google’s own thresholds, and if nothing significant has been changed since, some of the numbers start to mean something. Organic still cannot be called.
  • Months four to six. Enough sustained conversion cycles to trust the bidding data. Careful scale-up begins, on evidence. Technical organic gains should be fully reflected by now. Content-led gains are still early.
  • Months six to twelve. Organic starts compounding. This is where the work done in month one finally shows up as an asset rather than a cost.

Six months is not a padded contract. It is roughly the shortest honest route from broken to scalable, and most of the waiting is imposed by Google, not by me.

What to ask whoever is doing this work

If you take one thing from this, take these questions. Ask them of me, or of anyone else quoting for the work.

  1. How will you prove the tracking is right before we spend anything?
  2. What are you not going to change during the learning period, and for how long?
  3. When exactly will we be entitled to judge this, and what is that date based on?
  4. How many conversions a month do we produce, and is that enough for the strategy you are proposing?

An honest answer to those four is worth more than a fast promise. Anyone who tells you they can scale you in week two either does not know how the systems work, or is counting on you not finding out until the budget has gone.

Fix what is broken. Build what is missing. Then scale. In that order, or you are just spending faster.

Sources

All checked 14 August 2026.