I’ve had a number of people come to me looking for a W3Counter plan that can handle more than the “Pro” plan. More websites, more pageviews, or sometimes both. If people are knocking on the door asking to buy, I need to figure out a way to sell what they want. I decided against letting people download and run W3Counter themselves — it’s complicated to set up, with a complex database, a slew of cron jobs, and requires pretty recent versions of PHP and MySQL many hosts still don’t offer. It’s also unlikely to perform well on shared hosting, and I’m sure some would try to run it there anyway.

What I’m going to investigate now is how much performance I can expect from the average VPS hosting account. If I can track, say, half a million page views a day with good response times on a $39 hosting account, I know I have people that would easily pay $100 a month to get a W3Counter account with that kind of dedicated capacity. If I can get a million daily page views out of a $59 account with double the RAM, I know at least one person that is willing to pay $150 a month or more to use W3Counter on his network. With a VPS per account, I can also do private labeling, allowing a web designer to brand W3Counter as their own and sell or bundle web stats with their other services. They can even run it on their own domain by pointing a subdomain to the VPS’s IP.

I don’t trust most web hosts selling VPS accounts. A relatively small company is likely to push the limits of how many VPS’s a single physical server can handle, resulting in poor performance for each virtual server. If there are so many accounts running that the memory is being maxed out, then whatever amount of burstable RAM they claim isn’t going to be attainable. Instead I tried to find the biggest host with decent prices — even though they’ll still pack as many VPS’s as possible per server, they’ll have enough accounts already to know how much is too far, and they are less likely to have money pressures dictate they suddenly squeeze in more accounts on existing servers. 1&1 seemed like the right match for those criteria, and each account comes with root SSH access, Plesk (my current favorite control panel), and an SSL certificate for the secure tracker.

Once it’s set up and I have access, I’ll be installing a copy of W3Counter there, building the database, then throwing as many page views at it as I can. I’ll see how many it can handle tracking before response time starts dropping. That’s the limiting factor since a slow-responding tracker image can miss page views and slow down loading of the sites it’s tracking.

Think this’ll work?

I’m also strongly considering eliminating either the Plus or Pro plan in favor of a single plan in-between the two at the $15 price point, adding 2-3 VPS-based private instance plans, and making the PWidget premium only. I’m also tinkering with letting new accounts use the premium features for the first 14 days free so they know what’s available.

None of this takes that much more work to accomplish. I don’t need to do major re-architecting or build advanced new features to offer these new plans. That’s why I think it’s worth trying. I also need to see if there’s really a market for white labeling services like this before I build out the ad tracking program. And I need W3Counter subscriptions to sell well if I want to scale out to handle more users and, eventually, features that require more resources… like archiving data to provide date ranges on all reports, and outbound click tracking.