Ilyes Ouhadj

Founder @ Zedkira
Engineering micro-SaaS growth
5-day SaaS Audit: your top fixes as tickets. Refund guarantee.
Building Zedkira Inventory - @zedkira_

[Ilyes Ouhadj @Zed_kira]

How much of your churn is not a decision?

In Stripe, look at last month's canceled subscriptions. cancellation_details.reason = payment_failed means the payment failed, nobody chose to leave.

Churnkey's Stripe data: involuntary churn is 22% of SaaS churn. What's yours?

[Ilyes Ouhadj @Zed_kira]

Veracode tested AI models on 80 coding tasks with no security prompting. SQL injection: 82% passed. Cross-site scripting: 15%.

So AI writes safe queries and unsafe HTML. Search your code for innerHTML and dangerouslySetInnerHTML. Read each hit by hand.

[Ilyes Ouhadj @Zed_kira]

Prompt caching can fail with no error.

Check the usage field in your API responses. If cache_creation_input_tokens and cache_read_input_tokens are both 0, you're paying full price.

Usual causes: prompt under the model's minimum length, or a timestamp above the cached part.

[Ilyes Ouhadj @Zed_kira]

Claude Sonnet 4.5 shuts off on the API Nov 30 (Anthropic's docs). Search your code for claude-sonnet-4-5-20250929.

Sonnet 5.5 looks like $3 to $2 per million input tokens, but the newer tokenizer makes ~30% more tokens, so the real saving is closer to 13%. Measure yours.

[Ilyes Ouhadj @Zed_kira]

Your average AI cost per user hides your real problem.

A 10-minute test:
1. Export 30 days of AI usage by customer.
2. Sort by cost.
3. Compare your top 10 users' cost to what each pays you.

If any cost more than they pay, a flat plan is hiding a loss.

[Ilyes Ouhadj @Zed_kira]

@bradynord @weavdotcom We'd take one. We run a small retail app with paying customers, so we can try it. Happy to send some honest notes.

[Ilyes Ouhadj @Zed_kira]

@marclou The 11% decline is the median. Total revenue actually grew in 3 of the 4 months.

So SaaS isn’t dying: revenue is concentrating in a few winners while the typical startup shrinks.

Matches ChartMogul: B2B median NRR 82% vs 49% for B2C. B2C leaks first.

[Ilyes Ouhadj @Zed_kira]

@FilipPanoski The real flex is $1k → $15k on the same two channels. Which funnel fix moved it most?

[Ilyes Ouhadj @Zed_kira]

Is your AI SaaS a product or a thin wrapper? Blind test:

1. Pick 10 real customer inputs.
2. Write scoring criteria first.
3. Run each through your product and plain Claude.
4. Hide labels. Someone who didn't build it scores both.

Can't tell which is yours? That's your answer.

[leo @leojrr]

Don't validate you SaaS ideas with friends or family.

Their opinions are biased and may sugarcoat the truth to spare your feelings.

Instead:

- Analyze competitors
- Interview ICP
- Build a wait list
- Launch MVP

Only the market will give you honest feedback.

#buildinpublic

[leo @leojrr]

You don't need to build a SaaS unicorn.

Find a niche in a validated market and build something.

The faster you build, the better.

#buildinpublic

[Dan Kulkov @DanKulkov]

Let's say $5000 is your profitability:

• Low-ticket SaaS: $9 x 555 customers
• High-ticket SaaS: $49 x 102 customers
• Digital Products: $99 x 50 customers
• Low-ticket Service: $299 x 16 customers
• High-ticket Service: $999 x 5 customers

Choose wisely.

[Kevin Yun @kevinyun]

I'm building a new SaaS business from scratch and this is how I'm starting my outreach, cold on LinkedIn.

First day results:

- 115 connection requests sent
- 24 accepted invite (20%)
- 24 conversations initiated -- see screenshot of my message asking for feedback
- 13 replied back -- no negatives, so counting that as a win (54%)
- Other engagements: 1 booked call, 3 really in-depth convos or insightful feedback, 4 website signups

Total 13 engagements of 115 connections sent = 11% "conversion rate".

Some takeaways:
- Keeping it human is best. Compared to other outreach I've done in the past, this is best reception so far.
- When I over-personalize, it sounds like pandering, which sounds like spam. Still figuring out which is better
- This is not a sales pitch, so the "conversion rate" isn't a benchmark to take inspiration from by any means
- Just sharing what seems like a good way to reach ICP (or figure out if they're not ICP). More ideas in the pipeline.

[John Rush @johnrush]

I think I've found a gold mine.
There's a big, untouched source of traffic for Dev Tools.

The SEO potential for Dev Tools is huge - 100x bigger than other non-dev SaaS.

Here's the proof, I'll try to replicate it with a few other dev tools ASAP.
I activated https://t.co/ht7ue4R04S for https://t.co/mPEjz3yUzr on Oct 19th.

Two weeks later, we were showing up on Google.
A few weeks after that we had 8k views and 417 clicks from articles made since Oct 19th.

But here's the twist.
Despite trying hard with my other "non dev" products such as indexgoogle, seobot itself, and other "marketing" products, Google gave us little traffic. Non zero, but at least 10x less than what we've got for dev tools, such as FloatUI.

So what does this mean?
It's simple: SEO for dev tools is an untapped resource. Developers are just ignoring it.

Next week I'm trying a new feature of SEObot: programmatic SEO + directories.
I believe it may just explode and bring even more traffic from google. If this works, it'll change the whole marketing game for dev tools out there.

[leo @leojrr]

If you're looking for SaaS ideas, I think there's a ton of opportunity in FB groups.

It's still a great place to discover niche communities and find pain points.

#buildinpublic

[leo @leojrr]

Churn is the silent killer of SaaS.

Regularly analyze exit feedback, enhance user experience, and keep innovating. Remember, it's cheaper to retain a customer than to acquire a new one.

#buildinpublic

[sophie @netcapgirl]

napoleon founding a b2b saas company

[William Salim @wilthewriter]

I'm happy.

I helped a SaaS founder get:

🔹19,700 followers
🔹Millions of impressions

In 4 months.

No paid ads. No paid retweets.

How?

🔹Share insights
🔹Share valuable posts
🔹Share accomplishments

Value leads to followers.

[leo @leojrr]

Too many SaaS founders focus on flashy features rather than solving actual customer problems.

Build solutions that matter, not just what looks good on a feature list. Your SaaS is only as good as the problems it solves.

#buildinpublic

[Kyle Anthony Miller @kyleanthony]

Recent pitch decks for b2b SaaS startups.