Improve your Google ranking without an agency

I sell Google Ads. I don’t make a cent off your organic ranking, and I don’t have an SEO package to offer you at the end of this piece. That’s why this list is short: it only contains what I’ve actually seen work, on my own site and on the sites of clients whose Ads accounts I manage.

The honest starting point: a small or mid-sized business sitting on page three today can, in most cases, move up without an agency. Not because SEO work is worthless, but because the first big steps don’t need specialist tools. They need an afternoon to understand and a few months to stick with it. What an agency is genuinely better at starts after that: fiercely contested terms, thousands of product pages, technical SEO on a shop that’s grown organically over years.

First: is ranking even your problem?

Before you optimize anything, one question can save you a lot of work: how fast do you need the result?

Organic visibility is an investment with a delay. Even if everything below is done right immediately, you’re talking months before the position on a moderately competitive term settles down. If your order book is empty in four weeks, that’s not an optimization question, it’s a timeline problem, and ranking is the wrong lever for it.

Which channel fits which timeframe, I’ve written up separately: SEA vs. SEO: when Google Ads is worth it. This piece assumes you’ve made that call already and answers the other question: you want to move up organically. What do you actually do?

1. Does the page match the search intent?

This is the step that moves the needle most, and the only one I’d mark “do this first.” Everything else is polish on a page that either fits or doesn’t.

The logic is simple: Google doesn’t rank the page with the most keywords, it ranks the page that answers the question that was actually asked. If someone searches “patio roof prices,” they want to see prices. A beautiful product page without a single number doesn’t answer that search, no matter how often “patio roof” appears on it.

The honest test, no tool required: open a private browser window, search your term, and read the first five results. Don’t skim, read. Then answer three questions:

• What format are they? Guide article, price overview, product page, comparison table, location page? If four out of five results are guides and you’re offering a sales page, your problem isn’t optimization, it’s the format.
• What do all five have in common? If all five name a price range, describe the process, or show photos of finished projects, that’s not a coincidence. That’s the expectation searchers bring with them.
• What’s missing from all five? That’s your opening. If none of the five says roughly what it costs or how long it takes, and you do, you’ve just given someone a reason to prefer your page.

Run that comparison honestly, and afterwards you’ll either know your page fits, or you’ll know exactly what it’s missing. Either way, that’s more than any automated audit can tell you.

2. Title and description are your click argument

Position and clicks are two different things. You can rank fourth and still get fewer visitors than the result in seventh place, if seventh makes the better promise in the results list.

The title is the promise. And it gets cut off: Google truncates it based on pixel width, not a character count, wide letters count for more than narrow ones, which is why there’s no exact limit you could look up. I work with roughly 60 characters as a working limit and put the most important thing first.

This isn’t theory I read somewhere else. I measured the titles on my German site and found several that were getting cut off in the results, and what got cut was always the end, where the actual argument sat, while my name sat comfortably up front. The fix wasn’t optimization, it was order: benefit first, brand second, and if the brand no longer fits, it goes.

The same applies to the description, in a softer form: it’s not a ranking factor, but it’s the second sentence someone reads before choosing a result. Write it as an answer, not a table of contents. “What a patio roof costs, what the price depends on, and how long installation takes” is an answer. “On this page you’ll find everything about patio roofs” is not.

For the title, description, main heading and mobile-friendliness of your own page checked in one pass: my free landing page check (in German) enters a URL and shows the result immediately.

3. Internal links: your most important pages are often the loneliest

This is the step almost nobody takes, because it’s unglamorous and nobody sells it.

The mechanism: a page that barely gets linked from your own site is hard for a crawler to reach and practically invisible to a reader. Someone arrives through a guide article, reads it to the end, and finds no way at all to the page where you actually make money.

On my own site it was exactly this way round: the articles happily linked to other articles, and the pages meant to trigger an inquiry carried the fewest internal links of all. That’s not an exotic mistake, it’s the default, because articles get written and linking is its own, boring task.

How to clean that up in an hour:

1. Write down your three to five money pages. The pages where someone inquires, books or buys. There are almost never more than five.
2. Go through your guide content and ask, for each piece: which of those pages does it belong to? An article about pricing belongs with the pricing page, an article about the process belongs with the service page.
3. Add a link from the body text to it, with anchor text that says what to expect. “Click here” says nothing, “what a patio roof costs” says everything.
4. And don’t forget the other direction: link from the money page back to the article that answers the detail question. Someone hesitating reads on instead of bouncing back.

A link is only worth something if a human would plausibly click it. A block of twenty links at the bottom of the page isn’t internal linking, it’s decoration.

4. Load time: the overrated step

Here I’m being the awkward one, because the answer isn’t the one most checklists give.

I’ve never seen a page jump from position 14 to position 3 purely from a faster load time. Load time is one factor among many, and it only decides a race once two pages are equally strong on content. For most businesses, they’re not, the difference sits in the content, not in half a second.

What you should still fix is the obvious stuff. Not for ranking, but because it costs you visitors:

• Images uploaded straight off a camera. A 6-megabyte photo in the header is by far the most common find. Resize to the actual display size, use a modern format, done.
• Pages that noticeably hang on mobile networks. Don’t test on office Wi-Fi, test on a train. If you get impatient yourself, it’s too slow.
• Scripts nobody needs anymore. Chat widgets, tracking leftovers from tools you switched off, fonts loaded in eight weights.

And then stop. Chasing the last few points in a speed test is a project that eats weeks and delivers a ranking return close to zero. The return on conversion is real, but that’s a different calculation, and it doesn’t justify an SEO budget.

5. Evidence on the page, not claims

This point will probably be the most important one in five years, and it costs you nothing but honesty.

The mechanism: a page that states a number and says where it came from is citable. A page that states the same number with no source isn’t. Citable means others link to it when they write about the topic, and links from other sites remain one of the few factors nobody disputes the effect of.

There’s a new layer on top of that: more and more people get their answer inside an AI answer or a summary at the top of the search results, without ever clicking a result. A machine assembling an answer can only pull sentences that are verifiable. “Your competent partner for patio roofs” can’t be lifted. “Installation takes two to three days, based on the last forty projects” can.

What actually belongs on the page:

• Your own numbers, with context. Not “many satisfied customers,” but what you actually counted, over what period, out of how many cases.
• The method next to it. How did you measure? What’s excluded? Sounds like effort, is really one sentence.
• And the limit alongside it. If your number comes from forty projects, write forty. A small, honestly labeled sample is stronger than a big, unsupported superlative, and it’s the one piece of evidence nobody can pick apart.

If a number on your page is missing three of these, it’s a rough estimate, not evidence. I played out exactly how fast that falls apart in the real world using German click-price figures as the example: Google Ads cost figures, source by source.

6. Patience, and reading Search Console correctly

The last step is the one most people give up on, because they’re watching the wrong number.

Google Search Console is free, belongs to Google, and shows you what your pages actually get shown for. If you set up only one tool, make it this one. But read it in the right order:

1. Impressions first. An impression means Google considered your page for a search at all. If a new page has zero impressions after four weeks, it isn’t ranking badly, it isn’t in the race at all, and that’s a different problem: indexing, format, topic.
2. Then position. It usually moves before clicks do. A page moving from position 40 to 18 hasn’t brought a single visitor yet, and it’s still the first real signal that something is working.
3. Then click-through rate. Only once a page is within visible range is it worth polishing the title and description. Before that, you’re optimizing the click argument for an ad nobody sees.

And the point that causes the most confusion: the position shown in Search Console is an average. It averages over every query, device and location that page was shown for. “Position 8” can mean rank 3 for the one term that matters and rank 30 for twenty terms that don’t concern you. That’s why a page’s overall number is nearly useless, pick out the three to five queries that matter to your business and watch only those.

The timeframe I plan around: after four weeks, impressions tell you whether Google has picked up the topic at all. After three months, you know whether the position is actually moving. Decide after two weeks and you’re deciding based on noise.

What doesn’t work

The uncomfortable part. Four things that get recommended regularly, and that I wouldn’t put on a checklist.

Keyword stuffing. Writing the term forty times on the page hasn’t worked in years, because search engines match meaning now, not strings of characters. What’s left is a page that reads badly. That costs you readers, and readers who bounce straight back out are the opposite of what you were trying to achieve.

Bought links. I’m not saying it never works short-term, I can’t know that. I’m saying it’s a risk you can’t undo, it violates the guidelines of the very search engine whose results you want, and for a local business the potential upside is nowhere near worth it. One genuine link from a supplier, an association or the local paper is worth more, and costs only a phone call.

Mass AI text with no substance of its own. The problem isn’t that a text was written with the help of a model, I work that way myself. The problem is a text that says nothing that isn’t already on twenty other pages. If your article has no number of your own, no experience of your own, no opinion of your own, there’s no reason to prefer it, not for a reader, not for a search engine, and certainly not for a machine assembling an answer.

Staring at rankings daily. Results differ by location, device, history and day. Check daily and you’re measuring that noise and nothing else, and then making decisions on a movement that never actually happened. Once a month is enough. If you want to look more often, look at impressions, they fluctuate more honestly.

If you need inquiries this week

This is where the honest part of this article doesn’t stop, it starts.

Everything above takes months. If your problem is “my calendar is empty in four weeks,” that’s not a ranking problem, it’s a visibility problem with a deadline, and that’s exactly what paid search is built for. You buy the exact searches you’re not yet visible for organically, today.

That’s not an argument against SEO. It’s the division of labor I see with almost every client: ads carry the urgent, purchase-ready searches immediately, and the organic pages grow in parallel behind them until they can take over part of that load. What that costs, by industry, is under Google Ads costs (in German); what ongoing management looks like is under Google Ads management (in German).

And if your ranking problem can be solved with the six steps above: then you don’t need me for that. That’s exactly the point.

FAQ

How long does it take for a Google ranking to improve?

Plan for months, not weeks. After roughly four weeks, impressions in Search Console tell you whether Google picked up the topic at all. Whether the position actually moves can only be judged reliably after about three months. Fiercely contested terms take longer, local niche terms sometimes noticeably less.

Do I need an SEO agency to rank better?

Usually not for the first steps. Search intent, titles, internal linking and real evidence on the page can be implemented by any business itself, and that’s exactly where the biggest lever sits for small and mid-sized sites. An agency pays off once you’re dealing with fiercely contested terms, very large site structures, or technical SEO on a shop that’s grown over years.

Is load time really that important for ranking?

Less than most checklists claim. Load time is one factor among many and mostly decides when two pages are otherwise equally strong on content. Fix the obvious stuff, huge images and unnecessary scripts, and put the rest of your time into content instead. Speed still pays off for conversion, though.

What exactly does the position in Search Console mean?

It’s an average across every query, device and location a page was shown for. A reported position 8 can mean rank 3 for the important term and rank 30 for many unimportant ones. The number only becomes meaningful once you break it down to individual queries and watch only the ones that matter to your business.