Could You Build This Yourself? An Honest Answer From Someone Who Sells the Alternative

Quick Takeaways

1. A good number of small operators can build most of their own AI visibility foundation. The ones who should are the ones with spare hours, patience for fiddly detail, and a tolerance for being publicly wrong for a few months.
2. One weekend gets you the basics: structured data on your key pages, a Google Business Profile done properly, and a first pass at making your site readable by AI assistants.
3. A few months gets you the harder things: campaign judgement, tracking you actually believe, and writing that sounds like a person typed it.
4. Hand over anything where being wrong costs more than paying someone to be right. Ad budget sits in that category, as does anything with a one way door.
5. Our own site moved from 54 to 91 out of 100 over three months, and this is what we do full time. Budget your own timeline accordingly.
6. The Billy Bot Blueprint is a free Skool community where I post the actual work, including the parts that went sideways.

The conflict of interest, stated up front

I have an obvious commercial interest in the answer to this question being no. I run an agency. When an owner asks whether they could build their own AI search visibility, the sensible reply for my bank balance is a sympathetic wince and a line about how complicated it all is.

For a while that was roughly what I said, and it was a reflex more than a considered view. I stopped because I kept being wrong. Twice now I have talked to an owner, priced the work, and then watched them go away and build a perfectly respectable version of it themselves over a couple of months. They got what they needed, they understand their own setup, and they still ring me when a job sits outside what they want to learn.

So here is the honest version. A decent slice of this is learnable by anyone who can read documentation and stay interested in something fiddly for longer than one afternoon. Whether you are one of those people has almost nothing to do with how technical you are, and almost everything to do with how you feel about detail that nobody thanks you for.

What we are actually talking about

AEO, answer engine optimisation, is the work of making your business legible and quotable to the tools people now ask when they once would have typed into a search bar. ChatGPT, Gemini, Perplexity, and the AI Overview sitting at the top of a Google results page. The machines read your site, decide what kind of business you are, decide whether they trust you, and then either name you in an answer or name somebody else.

That work sits in three layers and they behave completely differently. The first is plumbing, which is mechanical and has right answers. The second is judgement, which arrives after you have done it badly for a while. The third is consistency, and that is the one which quietly defeats most people, agency or otherwise. Sort out which layer you are asking about and the DIY question gets much easier to answer.

The weekend layer

This is the part I would encourage almost any owner to attempt. The return on one Saturday is genuinely good, and you will understand your own business better afterwards.

Structured data

Schema is a shared vocabulary that tells a machine what each thing on your page is. This block is a business name, this one an address, these are opening hours, this is a review, this is a question and its answer. Picture a building with every room labelled against a warehouse holding the same furniture in a pile. Same contents, and only one of them lets a stranger find the bathroom.

You can hand write this. The vocabulary is published free at schema.org, Google documents the types it reads, and a testing tool will tell you when you have made a mistake. An operator who has never touched code can get LocalBusiness, FAQ and Review schema onto their main pages in an afternoon, provided they are careful about commas.

A Google Business Profile done properly

Almost every small business has one, and almost every one is half filled in. Doing it properly means the correct primary category, the service area drawn to match where you actually travel, real opening hours including public holidays, photos taken this year, a description written by a human, and a genuine habit of asking for reviews and replying to every one.

None of that is clever. It is a grind, it takes about two hours to do well and fifteen minutes a month to keep, and it moves local AI answers more than most of the expensive things people buy.

A first pass at machine readability

Open your robots.txt file and check you are letting the AI crawlers through, because plenty of site builds block them by default and nobody notices for a year. Then look at your key pages and ask whether each heading is shaped like a question a customer would ask, and whether the first paragraph underneath answers it in plain language before you start selling.

That one habit, question in the heading and answer in the first forty words, is the highest leverage writing change available to a small business right now. Alpha Ski Tomamu went from 50 to 90 out of 100 in seven days, and the bulk of that move was plumbing of exactly this kind. Mechanical work produces fast numbers because machines are doing the reading.

The few months layer

Here is where the honest answer gets less comfortable, because the things in this layer cannot be read into existence.

Campaign judgement. You can learn the mechanics of Meta Ads Manager in a weekend, same as anyone. Judgement is a different animal. We test five images against five pieces of text, which makes twenty five ads, and the whole value sits in what you do on day four when three are middling and one is quietly working. Reading signal against noise, knowing when to leave something alone, treating a return of 8:1 as the floor of the acceptable range, all of it arrives through spending money imperfectly for a while. The tuition fee is real.

Tracking you actually believe. Most small business analytics are wrong in a boring and specific way. Conversions fire twice, the thank you page counts anyone who lands on it, a phone enquiry from a returning customer gets credited to last week's ad. Getting to a number you would bet on takes weeks of unpicking, and every decision you make afterwards rests on it being true.

Writing that sounds like you. The best organic result I have ever had was a Facebook post for an accountant with no ad spend behind it, no link in it and no call to action. It passed 750 likes and produced around 70 leads, and it worked because of details nobody could have researched, including me being wrong about doing my own tax. Writing that reads like a person usually costs you a few months of publishing things that read like a brochure.

Our own site went from 54 to 91 over three months. We built the scoring system and we work on this full time, and it still took three months of steady effort. Fitting the same work around a business that already fills your day, double it and be kind to yourself.

What to hand over

My rule is simple enough to apply in about ten seconds. Hand over anything where being wrong costs more than paying someone to be right.

Ad budget is the clearest example. Warner Pool Care in Brendale sat at 42 out of 100 when we started. We got them to 96 before a dollar went into Google Shopping, and then $30 a day returned 14:1. Push that same $30 a day at the site as it was at 42 and you are paying full freight to send people somewhere the platforms cannot read. My policy exists because of this: we won't run your ads without doing an audit and getting the score above 75.

The analogy I use is from a game I watch too much of. You would not hand the kicking tee to a middle forward when you have a fullback landing 85 percent. The ad budget is the shot at goal, the website is who takes the kick. Stack the odds, then kick.

The other category is anything with a one way door. Site migrations, redirect maps, changing your business name across forty directories, touching a booking system customers are mid way through using. Mistakes there are expensive to reverse and often invisible for weeks.

A way to decide, in five steps

1. Get a baseline score. Out of 100, by whatever method you choose, so you have a number from before you touched anything. Guessing at progress is how people spend two years feeling busy.
2. Give the plumbing one weekend. Schema on your top five pages, the Google Business Profile finished properly, robots.txt checked, question shaped headings with answers underneath.
3. Re-score on the Monday. If the number moved and you enjoyed the process, you have a builder's temperament and you should keep going. If the weekend felt like dental work, that is useful information and no reflection on you.
4. Pick one channel and one measurement, then leave everything else alone for 90 days. The most common DIY failure I see is four half built things where one finished one would have done.
5. Write down the cost of being wrong before you touch anything new. Above your own threshold, whatever it is, that job goes to someone who does it every week.

Where to learn it, free

I have started putting the actual work somewhere public. The Billy Bot Blueprint is a free Skool community at skool.com/the-billy-bot-blueprint-4837, covering AI bots, automation and marketing for small businesses, plus the people who build these systems for them.

What it is for is the layer sitting above documentation. Google will tell you what a valid FAQ schema looks like. It will not tell you which pages are worth putting it on when you have eleven hours a week, or what a decent result looks like in month two, or why the approach that worked for a pool shop in Brendale falls flat for a mobile mechanic. That gap is where the community lives, and it suits two sorts of people: owners who have decided to build this themselves and want to compare notes with someone doing it daily, and the builders and freelancers assembling these systems for clients who want to see how the work actually goes.

It is free, and it stays free. I post the real jobs into it, including the parts that did not go to plan, because a campaign that flatlined teaches more than a case study where everything worked.

The short version

Most of the plumbing is genuinely learnable in a weekend and worth doing yourself. Campaign judgement, tracking you trust and writing in your own voice take a few months plus a bit of public embarrassment. Anything where an error costs more than the fee belongs with someone who does it constantly. The owners who do best are the ones honest with themselves about which category each job falls into.

If you would rather hand the lot over, that is what we do at Infinity1, and a conversation costs nothing. If you would rather build it, come into the Blueprint and build it alongside people doing the same thing. Both doors are open, and I genuinely do not mind which one you walk through.

FAQ

How long does it realistically take to do this myself? A weekend for the structural basics, then roughly three months of steady weekly effort for the judgement layer. Our own site took three months to move from 54 to 91, and we work on this full time.

Do I need to know how to code? No. Schema is a formatting exercise, the Google Business Profile is point and click, and robots.txt is a plain text file. Patience with detail matters far more than programming ability.

What is the most common mistake owners make when they DIY this? Starting four things and finishing none of them. Pick one channel, one measurement and a 90 day window, then let the others wait their turn.

Is the Billy Bot Blueprint community actually free? Yes. It is public at skool.com/the-billy-bot-blueprint-4837, with no paid tier sitting behind it. I post the real work in there, including the jobs that did not land.

When should I stop doing it myself? The moment the cost of getting it wrong exceeds the cost of having someone do it properly. Ad budget, site migrations and anything customers rely on mid transaction all clear that bar for most businesses.