In 2013, Helen and I ran SP24, a 24-hour online SharePoint conference.
We counted roughly 4,500 registrations and did not buy advertising to get them.
I do not tell that story as a repeatable marketing formula. I tell it because, after 19 years in business, I am no longer confident I could reach the same audience in the same way.
That is an uncomfortable admission from someone who is supposed to know how to sell what he builds.
Helen and I were walking Hugo through Telford when the words came out:
I can't think of a single reliable way to reach people now. Can you?
She could not give me a neat answer either.
Building became easier while distribution became stranger
Collab365 used to be built around in-depth Microsoft 365 content, events and communities.
Then generative AI changed what people expected from information. We rebuilt large parts of the business around AI-assisted systems because our old model no longer felt sufficient.
The odd result is that Helen and I can now build more than we could have imagined in 2013.
Reaching the right person has not become correspondingly easier.
That statement is our experience, not proof that every small business has lost every channel. Different audiences and offers behave differently. But the old Collab365 playbook has become much less dependable for us.
Social reach belongs to the platform
In the SP24 days, we could share an event link and watch registrations arrive. Our community was already gathered around SharePoint, and social platforms gave those links enough visibility to travel.
Over time, our posts stopped travelling as far.
I cannot honestly reduce that change to one percentage or one algorithm update. We changed too. Our audience changed. The products, platforms and formats changed.
The conclusion I can defend is narrower: a following on somebody else's platform is access, not ownership.
The platform decides how much of that following we can reach on a particular day.
Search now answers more of the question itself
Google's own guidance for generative AI features says AI Overviews and AI Mode are rooted in its search systems and that existing SEO practices still matter.
Google also says these features can create opportunities for websites. That matters because it contradicts the easy, dramatic claim that AI search has simply killed all traffic.
Our concern is more practical.
When a search page can answer part of the question directly, publishing another generic explanation is a poor bet. The work needs original experience, useful evidence or a tool the answer itself cannot replace.
That is one reason this article keeps the awkward personal story instead of pretending I have discovered a new acronym for search optimisation.
Email is owned, but delivery is still conditional
We spent years building email lists because an address felt more durable than a social follower.
It is still valuable, but owning the list does not secure an inbox placement or a reader's attention.
Gmail's current sender requirements require authentication and other controls, with additional requirements for senders above its bulk threshold. Those are reasonable anti-abuse measures. They also show that an email programme now needs technical and reputation work, not just a list and a send button.
The honest job is to earn the subscription, authenticate the mail, make leaving easy and send something worth opening.
We cannot treat a historic list as a captive audience.
Paid reach is a test, not a rescue plan
Paid advertising can still work. Conferences can still work. YouTube can still work. Local outreach can still work.
I removed the old claims that ad costs were universally spiralling, a conference stand required a fixed minimum spend and human video would soon be replaced. Those claims sounded certain but did not describe every market, campaign or event.
Our problem is not that every route is dead.
It is that none deserves blind faith.
Every channel now needs a bounded test:
- Which specific person are we trying to reach?
- What problem are they already trying to solve?
- What would count as a qualified response?
- How much time and money will we risk before stopping?
- What do we keep if the platform disappears tomorrow?
That last question changes the work. A customer relationship, permission to email, useful first-party data and a page on our own domain survive more platform changes than an impression count.
We have not solved distribution
There is a temptation in founder writing to turn every uncomfortable moment into a seven-step framework by the final paragraph.
I do not have one here.
We are testing useful articles on our own site, clearer answers for AI systems, founder-led posts, smaller paid experiments and more direct conversations. We have also talked about hyper-local work in Telford because a pavement and a handshake cannot be algorithmically hidden from the person standing in front of you.
That is a set of experiments, not a proven growth engine.
The useful shift may be to stop asking for one channel that behaves like 2013 and build several small routes we can measure, learn from and leave without destroying the business.
Nineteen years in, I know more about building than I ever have.
I am less certain about distribution.
Perhaps admitting that is a better starting point than another confident marketing prediction.
If you run a one-person or two-person product business and distribution has become the harder half of the job, The 50x Founder Space is where we are working through those experiments without pretending one channel will always work.