30 days of Voice AI blogging: what I learned
30 days of Voice AI blogging: what I learned
Thirty days. Twenty posts. One AdSense rejection, one reapplication, 52 Search Console impressions that became several hundred. This is the honest account of what building Voice AI Insider from zero actually looked like - what I got right, what I got completely wrong, what surprised me, and what I would do differently if I were starting today.
I did not start this blog expecting it to be easy. I have been a PM long enough to know that most things that look straightforward from the outside have at least one hidden layer of complexity that only becomes visible once you are inside them. Blogging about a technical niche, building an audience from zero, and monetising a site that did not exist a month ago - I expected that to be hard.
Some of it was harder than I expected. Some of it was easier. A few things were completely different from what I had anticipated. This post is the honest thirty-day retrospective - written the same way I write project retrospectives at work: what went well, what did not, what I would change, and what the data actually shows.
If you are considering starting a niche technical blog - in Voice AI, in any adjacent technology space, or in enterprise software - this post is for you. Not the version where everything works perfectly. The real version.
What I got right - the things worth doing exactly this way
Writing from real experience, not research. Every post on this blog is written from deployments I have actually managed, problems I have actually debugged, and conversations I have actually had with enterprise clients. The 58% cost reduction case study, the 200ms latency story, the six-week SIP-WebRTC gateway rebuild - these are all real. I did not have to research anything I wrote about because I lived it.
This turned out to be the single biggest advantage the blog has. In a niche as technical as Voice AI, generic content written from second-hand research is immediately obvious to the practitioners who are the target audience. The specific numbers - 44% carrier cost saving, 34% AI save rate, 0.3% call completion rate difference - are what make people stop and read rather than scroll past. You cannot make those numbers up convincingly. The only way to have them is to have done the work.
Choosing a genuinely underserved niche. Voice AI has enormous vendor marketing - every platform publishes content about their own features. It has almost no practitioner content - no one writing honestly about what production deployments actually look like, what goes wrong, what the real ROI numbers are. That gap is exactly where this blog lives. Three weeks in, every Quora answer I posted ranked on the first page of Google results for its query because there was almost no competition from genuine practitioner content.
Publishing consistently and frequently at the start. Twenty posts in thirty days is unsustainable long-term. But it was the right strategy for month one. Google ranks blogs partly on publishing consistency, and a new blog that publishes 20 posts in 30 days establishes a trajectory that helps with early indexing. It also means that by the time the first readers arrive, there is enough content depth to make staying worthwhile.
What I got wrong - the mistakes I would not repeat
Applying for AdSense too early. I applied after four posts. I was rejected for "low value content" and "insufficient content." This was predictable in hindsight - Google's AdSense threshold is roughly 15 high-quality posts before they will consider a new domain. The early rejection was not harmful in itself, but it cost a review cycle that took three weeks. If I were starting again I would publish 15 posts before touching the AdSense application.
Not adding meta descriptions to posts from day one. I published the first eight posts without proper search descriptions. This meant that even as Google indexed them, the click-through rate was lower than it should have been because Google was generating its own meta descriptions from the post content - which are almost always less compelling than a purpose-written description. Adding proper meta descriptions to all existing posts took about fifteen minutes once I knew to do it. I should have known to do it from post one.
Underestimating the Twilio vs Vonage vs Plivo post. This post generated 41 Search Console impressions in its first two weeks - more than any other post - and had a 0% CTR. The page was ranking at position 6 for competitive queries and getting no clicks because the title and meta description were not compelling enough. A post ranking at position 6 with 0% CTR is a missed opportunity, not a success. Updating the meta description immediately after discovering this should have been the first thing I did that morning. I let it sit for three days before fixing it.
Not setting up both www and non-www properties in Search Console from day one. Having only one property set up for the first two weeks meant Google was seeing inconsistent canonical signals when it crawled some pages. The redirect errors that appeared in week three were partly a consequence of this. Setting up both properties and the www redirect toggle in Blogger settings should be one of the first five things you do when launching on a custom domain - not something you discover needs fixing when you see redirect errors in Search Console.
The thing that surprised me most
The thing I did not expect was how quickly technical content from a genuine practitioner perspective cuts through on LinkedIn. I have a moderately sized LinkedIn network - not enormous - but posts from this blog consistently outperformed everything else I have posted in the past two years.
The demo failures post - "Why Voice AI demos fail in client meetings" - had more saves and shares in its first 48 hours than any post I have published in five years of using LinkedIn. Not because it was the most technically impressive piece of content I have written. Because it named a specific pain - that four-second silence when a demo breaks down - that every PM and pre-sales engineer in my network had experienced and had never seen written down honestly.
What this taught me: The most shareable content is not the most technically sophisticated content. It is the content that names a specific experience that your audience has had but never seen articulated. The SIP trunk setup post will get more Google traffic. The demo failures post will get more LinkedIn reach. Both matter - they just matter for different reasons and require different distribution strategies.
The numbers at day 30
I am sharing these because the Voice AI blogging content on the internet is full of aspirational numbers from successful blogs and almost nothing from the first thirty days of a genuinely new blog. Here is the honest picture:
| Posts published | 20 |
| Pages indexed by Google | 6 (of 20+ submitted) |
| Total Search Console impressions | 52 (week 3) → growing |
| Average Google position | 6.1 (top of page 1 for some queries) |
| CTR | 1.9% (target: 4-6% after meta desc update) |
| AdSense status | Reapplied - under review |
| Affiliate links placed | 5 products across 15+ posts |
| LinkedIn engagement | Consistently above baseline - demo failures post was highest |
These numbers are not spectacular. They are not supposed to be at day 30. A new blog with a custom domain, twenty posts, and six indexed pages after thirty days is actually performing ahead of what most SEO guides would predict for this stage. The average position of 6.1 - on page 1 - for a brand new domain is significantly better than typical. The indexing lag is normal and will resolve as more posts accumulate and as backlinks from Dev.to, Hashnode, and Quora start flowing in.
What I would do differently if I started today
Before publishing a single post. Before writing anything. Five minutes that saves weeks of redirect error debugging and indexing delays.
Not as an afterthought. As part of the publishing checklist. A good meta description takes three minutes to write and directly affects CTR from the moment the post is indexed. Retroactively writing them for twenty posts is tedious. Writing them per post as you go is not.
Every technical post published to the blog should go to Dev.to the same day with a canonical URL pointing back. Dev.to articles get indexed by Google within 24 hours and rank independently. This doubles your indexed content without writing a single additional word.
Not 4. Not 10. Fifteen. The review cycle is three weeks minimum and an early rejection extends that by another cycle. The opportunity cost of applying too early - in time and in the risk of a second rejection - is much larger than the few weeks saved by applying sooner.
Standalone insights that do not require clicking through to a blog post. The audience you build on LinkedIn converts to blog readers over time - but only if you are posting consistently enough for the algorithm to distribute your content. I started this at week two. I should have started it at day one.
The question I get asked most often
Since sharing that I had started this blog on LinkedIn, the question I have been asked most often - in messages, in comments, in conversations at industry events - is some version of: "Is it worth it? Should I start a niche technical blog?"
My honest answer after thirty days: it depends entirely on whether you have genuine practitioner knowledge that is not already being written about. If you do - if you are working in a technical field where the content available online is either vendor marketing or generic tutorials written by people who have not actually done the work - then yes. The barrier to standing out is low because the competition for genuine practitioner content is almost non-existent.
If you do not have that - if you would need to research your posts rather than draw from direct experience - then I would not recommend it. Generic content in a niche with established players is extremely hard to rank and even harder to monetise. The advantage this blog has is entirely a function of the experience behind it. Without that, the platform and the SEO optimisation are insufficient by themselves.
"The most shareable technical content is not the most technically sophisticated. It is the content that names a specific experience your audience has had but never seen written down honestly."
- The biggest thing thirty days of blogging taught me about what actually resonatesWhat happens in month two
The strategy for month two is deliberately different from month one. Month one was about establishing depth - publishing enough posts across enough topics to give Google and readers a clear picture of what this blog covers. Month two is about establishing authority - getting more posts indexed, building backlinks through Dev.to and Quora, and letting the SEO compounding that starts around the 60-day mark begin to work.
Publishing frequency will drop from twenty posts per month to eight to ten - higher quality, more deeply researched, more likely to earn the backlinks and shares that drive lasting traffic. The LinkedIn strategy shifts from promotion to standalone content - posting insights, not links, four times per week and building the kind of practitioner presence that drives organic followers rather than paid reach.
And the AdSense review is pending. If it approves, the monetisation layer is in place. If it rejects again, I will read the new reasons carefully, fix what needs fixing, and resubmit. The blog exists whether AdSense approves it or not. The content is what matters - the monetisation is a consequence of the content, not a reason to write it.
Thirty days in, Voice AI Insider is exactly what I wanted it to be when I started: the resource I wish had existed when I began working in this space. Whether it becomes more than that - a genuine income stream, a platform for industry conversations, a reference library for the next generation of Voice AI PMs - depends on the next thirty days, and the thirty after that.
Following along from the start?
Every post on this blog is free to read and always will be. If you have found it useful - share it with one person who works in Voice AI, product management, or enterprise tech. That is the only thing that helps this blog grow.
Follow with your Google account and get new posts in your Blogger reading list automatically.
Comments
Post a Comment