Five Education

Where it started: February and March 2026

Five Education places SEND teaching assistants and support staff in schools across the South West. Matthew Webber founded it. It's work I care about: I have family who have benefited from great SEND staff, and I said so in my first email to Matthew on 3 February.

Matthew knew that useful writing would help schools and candidates find them. But when your team's day starts with filling cover before the school bell, the blog is the first thing to slide. Rob Saunders at Splendid Web already looked after the site, and he brought me in. We've done it together since March: I run the AI side of the research and writing, and Rob polishes and publishes.

Building an engine, April and May

We could have written a batch of articles and left it there. Instead we built something that runs every month. It looks at what schools and candidates are actually searching for, picks two topics on the evidence, writes a brief, drafts, checks every factual claim against a source, and only then hands the article to the Five Education team. Early on we also sent a short monthly questionnaire so Matthew could steer what we wrote.

Each month follows the same fixed steps, and each step has to pass before the next one starts. Research, topic choice, writing, the fact check and the monthly report run largely by themselves. I read and approve every article. The Five Education team then has two working days to object, and Rob uploads the finished pages to the website.

The early drafts had faults, and the team found them. The writing drifted North American in tone. Someone on their team pointed out that the dashes everywhere made it read like AI. Both fair, and both are now built into how we write: British English, the two industry bodies (REC and APSCo) always named together, and none of the tell-tale punctuation.

In May the team corrected how we'd described the government buying framework Five Education is approved on. We'd got it wrong. They had also tidied some pages directly on the live site, and our next pass would have quietly undone their changes. So every month now starts by reading what's actually live and working from that, not from our own copy.

June: the most useful feedback we got

The first articles went live in June. By the end of the month the checking was taking more of the team's time than they had, and it wasn't always obvious the work was paying off. They said so plainly, and asked whether one article a month would be enough.

It was the most useful feedback of the whole engagement. A service that asks a small team for homework every month is a service that small team will stop using, whatever the numbers say. Fixing that was my job.

July and August: taking the homework away

We took the homework away. The monthly questionnaire went in August. The report became one page in plain English that follows searches all the way through to enquiries. And we asked the team to flag what's wrong rather than edit pages directly, because a quick edit can quietly undo the search work underneath.

On 30 July the SEND teaching assistant pay guide went live. Someone on their team had gone through it line by line before it did. It became the site's most-read blog page in August.

Across the whole site, August against July: shown in 41,347 Google searches (up from 20,445), 698 clicks (up from 647), 2,431 visits (up from 2,011) and 36 school enquiries (up from 32). Not all of that is the articles. It's the whole site, but the direction is right.

September: the pay guide and 900 clicks

In mid-September Rob shared the Search Console view with the team: 791 clicks from Google search, up 29% on the previous period, with the site shown 57,600 times (up 127%). The pay guide alone brought 121 clicks, up 142%. On 22 September Google emailed to say the site had passed 900 clicks from search in 28 days.

The reactions in the team's group chat were short and happy. When Rob shared the September view, Matthew replied "This is great thank you both". When Rob posted Google's email ("This is ace!!"), Matthew wrote "Thank you Rob and Rian".

The searches that matter most are the ones where people don't already know the name. For those, September against August, clicks rose from 89 to 116 and the site was shown 4,903 times instead of 2,414. Its average position moved from about 12th to 9th.

AI assistants have started to notice too. Five Education came up in none of our 20 test questions in July, 3 in August and 4 in September. The pay guide is the one blog article Perplexity cites.

Now the honest weak spot. By September only 2 of the 10 articles we'd published earned clicks from search. The Bristol article had none after a full month. One strong article carried most of the gain, and that's normal for this kind of work, but it's worth knowing before you start.

The other weak spot is speed. The September articles took 15 days from research on 1 September to going live on 16 September. About a week of that was waiting for my approval, and about a week was waiting for the upload. Both are fixable, and both are on my list.

If you're thinking about something like this

Ask how much of your time it will need each month, and hold the supplier to it. If it's more than a quick look, it will slide the moment you're busy, and you'll be paying for something you can't keep up with.

Judge it on clicks from searches that don't use your name, and on enquiries. Those are the people who didn't know you before. Expect most articles to earn little and one or two to carry the rest, so look at the whole set over a few months rather than each piece on its own.

Where it is now

Six monthly cycles ran from April to September, and the work carries on. Next come fresh articles on the questions nobody in the sector has answered well yet, one piece each month written for schools rather than candidates, and a regular piece built on original data (the first, on the Department for Education's SEND staffing figures, is live) to earn more mentions from AI assistants. And fewer asks of the team's time.

In their words

“Rian has done a great job at improving traffic to our website through the new blog while also raising awareness in SEND and education supply.”

Matthew Webber, founder, Five Education

Questions people ask

How long did it take before the articles made a difference?

The first articles went live in June 2026. The clearest gain came in September, when clicks from Google search were up 29% on the previous period and Google reported 900 clicks in 28 days. Expect a few months, not a few weeks.

Did every article bring in traffic?

No. By September only 2 of the 10 articles published earned clicks from search. The SEND teaching assistant pay guide did most of the work, and the Bristol article had none after a full month.

How much of the client's time did it take?

Too much at first. A monthly questionnaire and article checks took more of the team's time than they had, so in August we dropped the questionnaire and cut the report to one plain-English page. The team now flags problems instead of checking everything.

Do AI assistants like ChatGPT pick up articles like these?

They started to. Five Education came up in none of our 20 test questions in July, 3 in August and 4 in September. The pay guide is the one blog article Perplexity cites.