Tag Archives: performance

Reflections on startup life: Week 39

Leave a reply
We haven't shipped the new release, however it has passed a few significant milestones entering into integration and now "live" and able to be tested.

Unfortunately once we started wrapping everything back up together (integrating the engine into the website), the speed performances we thought we had gained have tailed off. We need some more research to properly understand where we are at and what we can do.  The good news is that regardless it is still much more scalable – we should be able to throw more users at this engine without losing performance.  So we might not of reached our "sub second" goal, we are able to do some of the critical deliverables regardless.

From a market development point of view, things are going well.  I invested time in creating a Social Network Analysis of the Australian Election commentary.  It took around a day and a half to complete properly, but was very educational and turned into our most successful blog post to date, generating a lot of comment on Twitter.  Even better, I was called up by a journalist from The Weekend Australian to answer some questions about the post and these quotes made it into an article in the weekend paper.  It's always nice to get our name out there broader than just online.  It's strange how our psyche is still wired to feel that seeing something in print provide more "validity" – or at least it does for me!

We are continuing to build our presence and I think slowly migrating from a "purist" tools company to a strategy and tools offering.

I was also invited to speak at the Australasian Language Technology Association (ALTA) to make a short presentation about how we use text analysis and natural language processing technology tools.  This was a pleasant surprise as well and I'm looking forward to sharing with people researching in this area how we use this for business purposes.  This happens in December.

Alex has taken four days off – it's been a while since we've really had a proper break and something we've come back to again and again is that this journey is a marathon, not a sprint – stopping to refresh and recharge is essential.  The positive news is I didn't break the servers (yet) while he was away so that's a good thing.

Overall?  I continue to feel that the pressure of interest is building.  It hasn't broken yet, but the dice will start to fall our way – there is plenty of validation that we are doing "the right things", if just keep at it, things will come together over the next month or two.

Highlights

  • Article in the Australian
  • Generating some interest with our analysis
  • Invite to the ALTA conference
  • Meeting a market researcher looking for a tool – we have a meeting planned next week to see how we can coordinate and work together.
Lessons Learnt

  • Keep at it.  I can never pick which "thing" we do will generate the interest, but I'm convinced that it's not a "one off".  Consistency leads to recognition, not the individual post itself.
Goal this week – Customers

  • Knuckle down and refine our messaging – the BinaryPlex website is not really doing our brand any good, it needs to be better aligned with the Tribalytic one now.
  • Launch new engine to customers
  • Follow ups
Goal this week – Engineering

  • Ship the new release

Reflections on start-up life: Week 37

Leave a reply

Typically we know how to do whatever it is that we need to do (we may not always be the best at it, but we know the mechanics on how to go about it), but we sometimes don't do it because of blind spots in our thinking.

The realisation I spoke about last week, that we are in a direct sales model, really opened up my eyes to a range of things we should be doing that we weren't.  Monday I was "out of the blocks" in a rush and tried hard to focus on getting back in touch with lots of different people and contacts that I hadn't been chasing for a variety of reasons.

While it hasn't generated more sales (no, I didn't meet my target of 4, or even 1), it has generated a lot more meetings which I hope lead to sales or at the very least, more knowledge about why we are struggling to sell.

It's beginning to get very frustrating this "pause" in sales.  The first few went by in such a rush it was misleading in many ways.  It's turned out to be a very lean month.

On the engineering side, Alex is kicking goals – the performance release is coming on very well.  The new HyperTable implementation has already improved speed substantially and now he's ironing out other bottlenecks which mean we should meet our performance goal (of sub second searches).  He's also found other places we could improve even further which would enable us in the future to improve performance by 10x or more.  This is important for us in so many ways, although we can live without the future performance improvements for now.  I'll return to performance in a moment.

As well as the "usual" activities (coding, blogging, meeting, emailing, accounting etc.) that go on, we had a hard think about product value.  Sooner or later (in our case 3 – 4 weeks since the last sale) you have to ask yourself if the product is worth it.  I keep meeting people that love what we are doing (they love it, the get it, they think it's a great idea) but that won't pay for it, or at least won't pay for it yet.

So what does this mean?  I think it just means we aren't a must have product yet.  We don't yet have our MVP and need to continue to pivot.  It's a bit like a game of "hot or cold" – we are getting good warmer signals, but we aren't hot yet.

Here's where we now think we are at.
  • We charge ~$100 per month for access to Tribalytic.
  • Let's assume that (because some people have bought us and returned) we deliver more than this in value.  The question is, how much more? More importantly, for who?
  • I think realistically, the value we generate is currently at best a borderline amount. Why?  All our existing customers have other benefits – relationships, early adopters, a really specific business problem, that the more general customers don't.
This is OK, because it shapes our thinking on what the "next thing" we should do is.  The short answer is "something that generates more unique value and that customers want".

It has to be both of these.  For example, reports are desired, but don't generate unique value (of course something like reports may have retention value which is also worth considering however it's a different problem).  Although we may not necessarily change the price of the tool, another way of answering this question is "if we did X, could we charge more for what we do?"

We think we have it.  Something that takes advantage of what we think is our Unique Selling Proposition and delivers significant added value that will support the way customers are using Tribalytic and turn it from something that's "a convenience – we do it better than you could do it yourself but you still could if you had to" into something where we deliver something you couldn't do.  For perhaps obvious reasons, we plan to sit on this idea for a bit until we work it through, but I think this was the week were we saw how to pivot into something that isn't just useful, but that certain companies and marketing agencies will really want to pay for.

Highlights

  • Vision – excitement despite the challenges.
  • Finding a good niche where we can really help (but can we make money there… more on this in the future).
Lessons Learned
  • Matching your activity to the mode you're actually in really works.  Direct sales time.
  • Continually challenge and critique why you're doing what you're doing BUT stay focussed too.
Goal this week – Customers
  • More meetings and setting up more for next week.
  • Explore the idea we've come up with and see what the appetite for it really is.
Goal this week – Engineering
  • Fingers crossed, an early test release of the new improved engine ready for launch next week.

Reflections on start-up life: Week 33

Leave a reply
I'll say up front that this last week has been one of the toughest.  Some of the best highs and the lowest lows to date.  It's perhaps the most challenging post to write as well – I've always wanted to keep this an open dialog yet I find some topics this week a bit sensitive.  Here it goes anyway. 

Ultimately this week has been about performance.  Our performance with clients, the performance of the server and even the performance of the DJ.

For the first time we've hit significant performance issues.  It's compounded by a range of factors, but ultimately in trying to improve the speed with which we gather data it caused our site to fail.  Multiple times.  Every time we thought we'd fixed it, the problems would appear in a different area – it's like forcing more water down a pipe, ultimately the whole pipe has to be bigger, not just the opening.

This was annoying at first and then over time, more and more stressful.  We decided to significantly change the approach which finally looks like it's fixed things for now.  I suspect that as we grow and scale, this will be a recurring theme, but I think we've bought ourselves a couple of months breathing space.

In doing this, we dropped all the other items on the plate – this became issue number one.

As far as clients go, we've signed more clients last week than before, but we also lost one as well.  The lesson we take away is our renewal processes need a lot of improvement.  I don't believe the issue was related to the product per se, but the way we approached the license renewal was not good enough.  I'll leave that one alone – it's like a scab I keep coming back and picking – what if I'd done this instead of this? What's done is done, we can only do it better next time.

The DJ's performance on the other hand was awesome!  We sponsored our first event – International Social Media day.  Kate Kendall approached us to ask if we'd like to sponsor the DJ and we leapt at the chance.  I'm increasingly finding when I speak to people that they feel they've heard of us before, regardless of if they have or not, having a presence is on the whole a good thing.

The whole social media meetup night was great fun, good networking, great people, great entertainment, but my overall favourite moment was when someone asked me about Product Camp (which was on Saturday and I'd said I was attending).  They asked if I was going to speak, or if I'd be sending one of my staff.  I don't know if we ever will have staff, but it's nice people think we might!

Product Camp was really great.  I need to get out and do more events like this one – there is just so much to learn and listening to, talking with and presenting to people who do product management as a profession was so useful.  There is a big difference between a start-up and a large business of course, but that made it all the more interesting too.  Imagine taking 3 months to survey customers and analyse opinions to decide what features to schedule for the next 6 months development! Inconceivable!

Highlights

  • Sponsoring our first event.
  • Signing new customers and great indications for a couple of others.
  • People "getting it"
  • Vision for the future – Alex's email about a new reporting approach was just what I needed to lift my head out of the trenches and look at what might be in a month or two.  Exciting.
Lessons Learned

  • So much more to do – there are two ends to a funnel and you have to work on both.
  • I really don't like performance issues.

Goal this week – customer

  • Met some great people and have some interesting conversations to follow up on.  Lots of leads.

Goal this week – engineering

  • We are close to resolving all the performance issues – pipeline is back up and running, but a few niggly little things left to do to make sure we don't need to look at this again for a while.

Reflections on start-up life: Week 18

Leave a reply
I woke up Tuesday morning last week to see that Twendly had suddenly gone “nuts”.  Since our launch of Twendly we’ve been studiously ignoring it, not wanting to get distracted from the main goal – it taught us a lot but we didn’t want to waste more time on it in the short-term.

It’s no surprise that while it’s actually held some traffic, it’s generally been declining week on week in search numbers.  People still use it, but without any changes to the site or promotion, they drift off.  So it was a massive surprise to see that suddenly 80 new users had signed up and we had some 200 – 250 mentions on Twitter in one night.  For the most part, they were all in French so I spent a bit of time in Google Translate to understand what was going on.

I learnt a LOT of new french phrases for interesting sexual practices.  Someone mentioned Chat Roulette and Twendly in the same Tweet and all of a sudden we picked up all this cross-traffic.  Our number one search on Twendly last week (by a massive amount) was for L’ejaque Faciele.  I’ll let you run that through Google Translate yourself.

On the one hand I’m really pleased that something we developed is getting use and proving popular, on the other hand our contribution seems to (inadvertently) help the world of adult porn.  It’s not exactly solving cancer or anything is it?  As Fenn said to me the other day, ‘The Internet, it all trends towards cocks’.

On a more serious note, we again made significant steps forward last week, to the point where we are getting our separate code streams integrated back together (some minor hiccups with that but now resolved) and this week should see us actually producing the prototype and integrating with the UI.

I think it’s important for both of us to get out of the technical tunnels that we’ve been delving in, pull it all together and just see where we are at.  It’s too easy to become concerned about things that we know will be issues eventually, but at the same time, we don’t know if they are critical right now.  They certainly won’t stop us showing people the experience we are building.

Highlights?

  • A burst of usage in Twendly
  • Processing pipeline framework in place and seeing the stages working (handing off to the next process in the pipeline).
  • Generally an “up” week with the feeling that we are continuing to emerge from the rabbit warrens we’ve been in.

Lowlights?

  • Never get as much done as we’d like.
  • Concern about performance which is annoying both of us and will take discipline to put it aside for now to focus on showing the capability – speed can come later if people like what we can do.  If they don’t like what we can do with the data, the speed at which we can do it will be irrelevant.

Goal this week?

  • Integration – a fairly ambitious stretch goal, but to get the elements of Tribalytic plugged together and working so we can let people experience an early (non-performant) beta.