|Web analytics pricing and contract negotiation|
Looking for feedback to help with negotiation
| 10:30 am on Jul 14, 2004 (gmt 0)|
We are just starting a negotation with a new supplier for a new ASP-based web analytics package. Any feedback or suggestions you may have from your own experiences would be greatly appreciated. Questions/concerns I have are as follows:
-the vendor is proposing to charge on a page view traffic basis, which is not ideal for us given our high traffic volumes. Is this normal? Are there other ASP pricing models that you have seen?
-they are specifying a minimum 12-month contract period with an autorenewal after 11 months. We would like a break clause at 3 months but they are pushing back on this. What is the norm here for get-out clauses if we are not happy with the service?
-is it typical to charge a per user access fee on top of the monthly service charge?
-they wish to charge us for each domain separately. this seems unreasonable as all of the domains are part of our main website. Again, is this typical or should I push back on this?
-they are asking for payment in advance for the monthly service charge. Curious to know if advance or arrears-based pricing is the norm here.
What other pricing or contract traps have you seen that I should be aware of during this negotiation?
Many thanks in advance...
| 2:04 pm on Jul 14, 2004 (gmt 0)|
Seems like they have invoked every single pricing model layered together all in one package.
| 2:52 pm on Jul 14, 2004 (gmt 0)|
Yes, it is a rather restrictive pricing structure, hence my curiousity for other options or experiences. As with most technology purchases these days, the buyer is in a strong negotiating position, particularly in a market like web analytics where heavy competition exists.
| 2:54 pm on Jul 14, 2004 (gmt 0)|
|-the vendor is proposing to charge on a page view traffic basis, which is not ideal for us given our high traffic volumes. Is this normal? |
This type of pricing is common. Unit price is negotiable, especially in high traffic cases. Negotiate. It is quite important to agree on future pricing now (What might look like a good price now may get very expensive when the traffic doubles...)
|We would like a break clause |
Break clause is a perfectly valid request. Insist. It may save you a lot of money, especially if the service quality is bad...
|they wish to charge us for each domain separately |
This makes very little sense. They probably have a poorly designed accounting system (1 domain = 1 account, instead of 1 publisher = 1 account) and want to go the easy way. Imagine you had 50 domains, would they really want to send you 50 invoices every month?
|they are asking for payment in advance for the monthly service charge. Curious to know if advance or arrears-based pricing is the norm here. |
It's not a norm, but it's "defendable". This is the point where you could let them have what they want to balance the negotiation.
| 4:59 am on Jul 15, 2004 (gmt 0)|
* In technology, it is normal to trial products that cost more than $5000 or so, especially products that aren't from large companies. A good rule of thumb - if there is a salesman involved, you should demand a free trial.
* The purpose of the trial is to prove a) that the product works as advertised, b) that it will work in your environment, and c) that it generates an ROI sufficient to justify the cost.
* If the vendor balks at the trial, it generally means that they don't believe the ROI goals can be met. Best to find another solution in this case.
* With a trial, the break clause isn't as important, but you'll almost certainly be able to get one if you want it. Always remember: the vendors want to make the sale more than you want to make the purchase. That's DOUBLY true if there is a direct field sales force involved.
* Read over the for-cause cancellation clauses closely. Generally, vendors are going to try to make it impossible to get out of a contract. Make sure there are performance levels guarantees in there. You want wording that allows you to get out if utilizes their ASP solution "degrades performance" of your system, etc.
* Pricing is as much your responsibility as the vendors. You need to decide how much the information is worth to you and start the negotiations from there. If the information is worth $1000 a month, then you need to make sure you are paying less than that - regardless of the model chosen. Typically, the simpler the model, the happier everyone involved.
* Get the autorenewal clause out of there. You don't want to be the one who forgets to cancel. Trust me, the vendor will track you down and request the renewal. All that an auto-renew does is allow the vendor to get away with not re-opening negotiations.
* Let your finance types (e.g. controller, CFO, etc) look over the contract. He or she will almost always pick up things that you missed, and they'll be grateful for the opportunity to give input before a mess is made.
* When negotiating, play the intermediary, not the primary negotiator. The primary negotiator is always "your boss" or the "web analytics committee." You do not have the power to say yea or nay to anything - you can only make recommendations and see what the real decision maker has to say about it. This holds doubly true if you *are* the final decision maker. Never let the vendor know that you effectively have the power to say yes to the deal.
If they want to negotiate with your boss, just inform them that he's delegated responsibility for working through the negotiating points to you. He'll sign off when he's happy with the deal.
* Ask for references. All good vendors have happy customers that will be willing to talk to you. If a vendor pushes back here, walk away. If they can't find even three customers that would be good references, they aren't worth working with. Follow up with those references. You want references from companies that you've heard of. You want to speak with the references, not read case studies, and not just hear their names dropped.
Of course, if this is a $100/month deal, nevermind. ;)
| 3:11 pm on Jul 15, 2004 (gmt 0)|
That's an excellent list.
I've seen companies abandon, or want to abandon, analytics programs in which they invested hundreds of thousands of dollars just for the setup, let alone the ongoing fees. It would have been avoided if they had had hands-on experience with their own data AND of course a plan regarding what they wanted to see in reports. A consultant probably would have helped a lot too.
| 3:45 pm on Jul 15, 2004 (gmt 0)|
That sounds like a bad deal. Unfortunaltely a lot of these firms operate this way.
As for the cost per pageview thing, it's typical of trackers, as their costs will also go up when each of your pageview sends a request to their servers, so perhaps it's justified somewhat. Otoh, all the other stuff they want to charge for is BS - especially the part about charging for separate users. You should be able to let as many users log in as you'd like; it's your stats not theirs.
>> They probably have a poorly designed accounting system
I think it's their technology that works on a "per domain" basis. If so, that's not very good.
| 8:16 am on Jul 16, 2004 (gmt 0)|
I think the per user fee has to do with the database licensing model they have for their data warehouse platform, but I agree it's not fair for us to cover this cost. The per domain pricing does appear to be *flexible*, and my guess is that it's mean to protect them from companies that have multiple distinct businesses within the group that are effectively separate implementations with very different reporting requirements (like GE for example). That being said, why should we not benefit from the aggregate traffic across our business - wether we have one site or several?
In terms of pricing, we have the option to either agree on a fixed amount per month (which helps us from a budgeting perspective) with quarterly or semi-annual adjustments based on actual traffic, or simply get invoiced monthly in arrears based on actual traffic. I suppose either is reasonable. Any other pricing or payment models or options that you've seen?
Another aspect which they have proposed is 2-3 days of consulting per month for data analysis and interpretation. This is not mandatory but they have sold this as a value-added service. I'd be interested to know if anyone uses their vendor for these services.
| 7:06 pm on Jul 16, 2004 (gmt 0)|
wow. sweet deal (for them). I don't know what vendor this is, but don't the reports offer enough interpretability as it is? (i'm just feeling grouchy on a friday afternoon)
| 1:15 am on Jul 17, 2004 (gmt 0)|
It depends on how integral this system will be to overall operations as to whether or not consulting would be a good add-on. With some systems there are high-value custom reports that the vendor can add-on.
There is at lest on awesome (IMHO) high end analytics vendor out there who I think built everything inhouse and has no seat licences and integrates email & website tracking.
So, if they don't offer you want you want on your terms its not the end of the world. Also, if the contract is for a year, why not shoot for a month to month renewal after a year-long contract is over. If its a good system and all your data is stored in there, contrat or no, its tough to switch vendors.
| 3:16 pm on Jul 18, 2004 (gmt 0)|
Month-to-month after 12 months is a good suggestion. Certainly better than auto-renewal. We are also negotiating an SLA which if not met will give us the ability to terminate the contract with 30 days written notice.
I think the consulting piece is important part of the overall service. Their system is flexible and allows for custom reporting, email/campaign/website tracking, path reporting, funnel analysis and everything else, but the real issue for us is finding the time every day to sift through the various reports and get a summary of the main KPIs/information we need. We'd like the vendor to know our website, understand our business KPIs, and be able to advise us on a weekly or monthly basis on anything interesting or valuable in the data. Some of the mid-low end vendors are simply not able to provide this as they have very junior, inexperienced account managers with too many accounts to manage. The higher-end vendors will tend to have more senior, business-focused account managers or consultants who can provide this as a credible service. It obviously costs more though.
| 3:43 pm on Jul 18, 2004 (gmt 0)|
A few things to consider on this subject before you discount the current company:
(1) Pricing per hit is similar to cell phone companies pricing per minute. The more traffic the stats company has to log, the higher their costs are due to server capacity (most are running MS SQL 2000 or Oracle) and bandwidth. You might not think a stats company uses much bandwidth, but they do - each page view probably uses about 100 bytes, plus the size of the tracking image. Multiply that by a few million hits a day and you're talking more than 1 GB a day in bandwidth. So obviously their higher-traffic customers are going to pay more than the low ones since they cost them the most.
As for server load, the biggest toll on the server is getting the data for each hit *IN* the database, not looking at it later. Once it's in there, properly indexed, etc. running SELECT queries against the data to give you your stats don't cost (by "cost" I mean CPU / RAM not money) squat incomparison with inserting 100 bytes of data several times per second into a database (most likely with clustered indexes which would slow down an INSERT even further).
It's probably not a single server that handles the INSERTS, so you're probably looking at two or more clustered servers just for that. Then they have to be replcating the data over to one or more additional servers for people to be able to view their stats because there is no way you would be able to run SELECT queries against data that has new records being inserted at a rate of several dozen per second with any decent speed.
(2) Contracts are typical among companies that want to lock you in, so I would be a little worried about this part. I can understand contracts with a cell phone company since they only have a certain amount of phone numbers in the pool and they can't re-use one for so many months after another person stops using it. A web analytics company, however, has no such restrictions. If you leave their service, that FREES UP resources for the next customer so there is no reason they should require a contract unless they're giving you a personal account rep. If they ARE giving you a personal account rep, tell them you don't want one and they may skip the contract. Afterall ... what the heck do you need an account rep for for friggin web stats? You need to get it set up, and you need to be able to call or e-mail tech support if it doesn't work. That's pretty much it.
Before making a decision on this ask for a copy of the contract and read through it if you don't have one already. Look specifically for exit clauses, and if there aren't any tell them the contract is unacceptable. There should also be a bankruptcy and/or acquisition clause in there - if there isn't, demand one. For bankruptcy it basically says that if their company files for chapter 11 or chapter 7 you have the option to walk away because you have no guarantee that they're going to be around (no matter what they say). The acquisition clause basically says that if they are acquired by another company you have the option of terminating the contract because hey, maybe you don't like the other company.
One more thing - tell them you want a 15 or 30 day reversal option. This means that you can cancel within the first 15 or 30 days if you don't like the service and aren't bound to the contract.
If they are unwilling to put those clauses in there, as well as a clause for service availability, then walk.
(3) I don't really know if this is common or not, but I have used 3 different hosted stats companies before and none of them charged an extra fee for additional users.
(4) This is definitely not typical - it doesn't matter how many domains point to a single website. It's one thing where you have 5 domains that are 5 separate sites and you want to track them separately, I HAVE seen that priced higher (although I've also seen "unlimited domains" options for that too), but for the same website there should be no additional charges. I think maybe they misunderstood you and think you are referring to separate websites. If they didn't misunderstand your original inquiry, run :)
(5) I've seen this both ways too, where you pay your base monthly fee in advance and then any overage is billed with the next month's base fee. I've also seen full rear billing where you pay after you receive the service.
Let me know if there is anything else you can think of.
| 5:28 pm on Jul 18, 2004 (gmt 0)|
I see the point about the consulting for a complex site. As time goes on, if you go with this outfit and their consulting, if you could tell us about some examples ... it could be valuable to some of us. One hopes that the consulting will add value. At the same time, I have to wonder if the consulting makes up for them not investing in making their regular reports valuable and usable.
I really should NOT blindly criticize consulting add-ons since pushing the envelope on reporting and interpreting is how I make my living.
The thing that bothers me about the contract that has been described here is that it seems to combine every pricing model there is for this service, and the customer is going to be squeezed by whichever aspect gives them the worst deal. Well, not every model - I've seen pricing that takes into account the number of servers that hold the website, and worse yet, the number of CPUs (2 CPUs on a server equals double).
I can accept per-page-view pricing if it allows unlimited domains. I can even accept a reasonable one-time admin charge for setting up each additional domain or seat. But an ongoing monthly charge for it? That bothers me. And not having a detailed trial, at minimum a partial one that you pay for and that is subtracted from your fees if you sign up, is asking a lot.
If your site is a big name that they'll be happy to have showing up on their client list, they might need you more than you need them and I'll bet you have some room to dicker, on the domain charges at the very least.
Is this vendor financially stable? If the vendor is a division of a company that's principally known for something else, is the division financially stable and going to be around for awhile? (again, the voice of unhappy experience talking here) I don't think we're anywhere near finished with the metrics shakeout that's been going on.
| 8:23 pm on Jul 20, 2004 (gmt 0)|
Having demo'ed maybe 5 different solutions in the last year, this one sounds like its on the pricey end of the stick and its cost structure is really pushing things.
I would look out for hidden costs including the cost of running non standard reports. Another thing to consider is do you own the rights to your data and is it exportable should you want to move to another system?
Definitely do a trial of the system. There is no way to know if it works with your sites unless you try it. If its not compatible then you are looking at a lot of money in consulting setup fees to fit a square peg in a round hole.
Do you have to go with an ASP model or is this just a preference for you? While setting up something in house may seem expensive, in the long run its a lot cheaper and your data is a lot more accessible.
| 8:52 pm on Jul 20, 2004 (gmt 0)|
|Do you have to go with an ASP model or is this just a preference for you? While setting up something in house may seem expensive, in the long run its a lot cheaper and your data is a lot more accessible. |
Not to be contradictory, but I've actually found the opposite to be true. Buying the software required an up-front investment, plus you have to keep paying for upgrades. Sure you'll get so many "free" upgrades for a period of time, but as new versions come out that are after your upgrade period you have to start paying again. Add in the fact that you have to dedicate a system to it, and the costs of doing it yourself are often greater.
Without naming any names, if a good stats package costs $1,500 with a year of updates and a system to run it from costs $300 then you've got an up-front cost of $1,800. Even if "year two upgrades" come with a 10% discount you're still looking at ongoing costs of about $1,350 a year.
So if your hosted solution is $100 a month or less, why bother with having to buy, install, configure, and upgrade software? You'll have the "latest version" every time you log in with a hosted solution. Also remember that doing it yourself only has about a 70% accuracy rating where hosted solutions are much higher. If an ISP caches your page, as many large ISP's do, your log files which an in-house solution would be based on will never see a hit from a visitor. With a hosted solution the hit will be logged even if the actual page they were looking at was cached. Of course neither is 100% accurate - if the user is blocking referring documents you're not going to see everything you wanted to about that visitor anyway, but the accuracy rate of hosted stats solutions is much higher than local.
I've run both simultaneously and have always gotten a much higher accuracy rate with a hosted solution. Even if it was more expensive, I would prefer it over maintaining my own.
| 10:47 pm on Jul 20, 2004 (gmt 0)|
They could do the hosted solution in-house too, just direct the tags to their own server.
| 11:15 pm on Jul 20, 2004 (gmt 0)|
DigitalV: I think the scale of your site will determine which is cheaper, and from reading through all the costs associated with the vendor this person was looking at, I was assuming that their sites were fairly high volume. If you are handling over 10 million hits a month and only paying $100 please sticky me your vendor.
To be honest we run a home built logging system for all of our important transactional data, and just use a product for counting pages. Since we just use the purchased product for counting, we have not upgraded in 4 years and don't have a support contract either.
| 6:36 am on Jul 30, 2004 (gmt 0)|
Just got back from a business trip folks, so I'll continue with a few more comments. Thanks to everyone who has contributed to this, this has been a very useful and insightful thread.
Just to clarify, we are site with tens of millions of page view a month. Our site has thousands of pages. there are dozens of domains that each serve content in different (and in some cases complicated) ways. Several disparate content management systems are serving the pages from various locations. So, we made the decision to look at vendors in the high-end of the market only, of which in the UK we found there are really only 3 or 4 such companies. 3 of the ones we looked at are ASP model suppliers which are our preferred method of delivery for a number of reasons, not the least of which is the ability to bypass the IT 'powermongers' who made it extremely difficult for us to use and get value from our previous server-side system.
A few more questions on the pricing models. We are happy to pay the monthly fee based on page views, however the vendor's pricing matrix is quite linear, where I think it should flatten out at the higher volumes increase. I suppose more hardware is involved as traffic increases to maintain the SLA we have asked for, but this really seems to be a tough area for us to understand. In other words, should an increase from 5 to 10 million page views cost the same as the increase from 11 to 20 million, or should there be substantial economies of scale?
I have pushed hard on the trial, but the vendor we are negotiating with claims that they simply cannot do this as an ASP model. It's too expensive for them. They have offered a 3-month break clause, but this means we still need to pay the implementation charge which is quite high in it's own right. They work with two other large competitors in our space so they do have credibility and a proven track record, which is why we had shortlisted them in the first place.
They have also provided a quote for data archiving beyond 90 days. Is this normal from what you all have seen? They will restore from backup free or charge in the case of a failure, however archiving base data is a different issue. Do any of you have similar arrangements, and if to how long do you tend to archive the base data? Just to be clear, this vendor will keep aggregates for two years as part of the monthly charge, but the base (raw) data is a separate issue altogether and takes up a HUGE amount of disk space.
On the implementation fee, they have attributed a portion of the cost to 'hardware and software provision', which is fine as they are hosting the service. What I would like to clarify is, do they buy hardware and software for every new client, or would they have a massively scalable system at their data center that can easily turn on new clients a little or no additional cost? My view is that the latter may be true for 'cookie cutter' sites and the lower end vendors, but that the larger, more bespoke clients may need to be on their own hardware/software.
As for the need for consulting, the reason we wanted this was what our e-business team are simply too busy to go through the reports on a regular basis ourself. We are happy to have a consultant summarize, interpret and report a weekly/monthly summary in the format we like. It also gives the vendor an opportunity to get close to our business, which we see as a good thing in the long run.
| 10:37 am on Jul 30, 2004 (gmt 0)|
It sounds like you are on the right track with the decisions and directions you've set so far. If they work with other big players in your space, and if those players seem satisfied, then a trial installation seems much less critical.
I'm accustomed to substantial (really substantial) discounts with volume jumps as big as that. Are you saying that the cost-per-page-view for 10 million hits per month is (for example) 60% of the cost-per-page-view for 5 million hits per month? And the cost-per-page-view for 20 million hits per month is 60% of the cost-per-page-view for 11 (or 10) million hits per month? It seems fair to have the same discount every time you double the volume (10/5, 20/10). But maybe you're describing something else?
| 1:51 pm on Jul 30, 2004 (gmt 0)|
That's exactly what I mean. For example, the first million page views might be £1/1000 PV, but at 20 million, it might cost 15p/1000 PV. The price-per-page-view really should reduce by a factor of X for each million page view increase then eventually flatten out at Xp/1000 PV. I suppose the curve depends on how the vendor system is set-up and how their cost structure changes with higher volumes. Most importantly, it will depend on what the vendor can get away with!