Hi everyone, I’m finally at the contract stage for Anomali after a lengthy trial with my team. 🎉 We’re a small operations group and this is our first "big" security tool procurement, so I’m trying to be super careful.
I’ve read the horror stories about SaaS bills ballooning after the first year, or fees for things you assumed were included. With Anomali, I’m particularly unsure about what scales and what doesn't. My main worries are around data ingestion and user seats.
Could you help me understand what specific cost drivers I should lock down in writing? For example:
- Is there a hard cap on data volume per month, and what happens if we exceed it?
- Are threat intelligence feeds priced separately, or are there tiers?
- What about support costs? Are they fixed for the contract term?
Basically, I want to walk into negotiations knowing which knobs they can turn later to increase the price. Any lessons from your own contracts would be so appreciated. Feeling a bit out of my depth here!
✌️ annie
Your list is a good start. Nail down the exact definitions.
What counts as a "user"? Is it a named account or a concurrent session? For data, get the specific unit (GB, events per second) and the overage rate in writing. That's where they get you.
Support costs often auto-renew at a higher tier. Get the renewal price locked in for at least two years.
Don't forget implementation and training. Those are usually separate line items that blow up initial cost.
Your list hits the main points. The data overage clause is where you'll bleed money if you're not careful. Don't settle for vague language.
Specifically, get the overage rate for data and users in the contract itself, not an appendix. Make sure it's the same rate for the entire term, not just year one. Support renewal pricing should be explicitly stated, with a cap on the annual increase, like CPI + 2%.
Also, define what "data ingestion" actually means. Is it raw logs, parsed data, or something else? A 10% variance in their definition can double your bill.
cost optimization, not cost cutting
Great questions. Data ingestion is the big one. Ask them to define the "metered unit" in the contract, like GB of raw logs vs. GB after parsing. That difference can change your volume by 30-40% without you doing anything differently.
Also, on support, get the specific service level agreement (SLA) attached. "24/7 support" can mean email-only after hours unless you have a guaranteed response time for phone or chat.
For threat intel feeds, see if they have a "core" feed bundle vs. premium ones. Sometimes they'll give you a baseline and charge extra for specialized verticals (financial, healthcare). That's a common add-on cost later.
Good luck! It's smart to ask this now.
Pipeline Pilot
You've raised an excellent and often overlooked nuance regarding the metered unit. That definitional ambiguity is a primary source of friction in later billing disputes.
A related clause to scrutinize is the process for any "unit definition change." Some vendors reserve the right to unilaterally change what constitutes a GB or an event, which can functionally alter your pricing without a contract amendment. It's prudent to require that any such change be mutually agreed upon, with a corresponding adjustment to your unit rate to keep your effective cost constant. This prevents a scenario where your data processing remains flat, but their new parsing logic suddenly counts 40% more "billable units."
Your point on support SLAs is equally critical. Beyond response times, the contract should define the conditions for escalating a case and the remedies available if those response times are consistently missed. Many agreements state the SLA but omit any concrete consequence for the provider when they fail to meet it.
Let's keep it constructive
The unit definition change point is critical. I've seen this exact scenario play out in a performance monitoring contract. The vendor switched from counting raw log bytes to counting "enriched event units" after a platform update. Our data hadn't grown, but the billable unit count jumped 35% overnight. The contract language was vague enough that they called it a "clarification," not a change.
Your remedy suggestion is sound. We now insist on a clause that ties the unit definition to a specific, versioned schema or data processing pipeline documented in an appendix. Any change to that pipeline constituting a new "unit" triggers a good faith pricing review to maintain cost neutrality. It turns an opaque operational shift into a contractual negotiation.
-- bb42
Yep, that "clarification" tactic is a classic one. It's a big reason we push for the pricing model to be tied to a specific, documented system state. In your performance monitoring example, they essentially changed the product's output without adjusting the input price.
One extra layer to add is a definition for "material change." We specify that any processing logic update that alters the billable unit count by more than 5-10% for a standardized sample dataset constitutes a material change, automatically triggering the pricing review. It removes the ambiguity about what a "change" is and preempts the argument that a 35% shift is just a minor clarification.
Integrate or die
Everyone's hit the big points on data and units, so I'll add one more from the Jira/Asana world that applies here: watch out for feature-based pricing creep.
You said you're worried about > which knobs they can turn later to increase the price. A classic one is when a feature you use daily gets moved from a "Standard" to an "Enterprise" tier in a future release. Your contract should list the specific modules or features included in your tier, and state that their functionality won't be materially reduced or moved to a higher pricing tier during your contract term.
Also, for a small team, clarify what "user seat" means for on-call rotations. If someone only needs read-only access for a weekly report, that shouldn't cost the same as a full analyst license. Good luck!
Great advice here already. Locking down the unit definition and material change clause is non-negotiable.
On threat intel feeds, get the list of included feed *names and providers* in an appendix. I've seen "threat intelligence module" go from generic OSINT to requiring a paid add-on for any commercial feed a year later.
For support, don't just ask if costs are fixed. Specify that the support *level* (e.g., 24/7 phone, 1-hour response) is also fixed for the term. They can't downgrade you to email-only while keeping the same price.
—cp
Your list covers the main pain points. Beyond just locking in the overage rate, push to have the first X% of overages included as a buffer without extra charge, like the first 10% of data or 2 extra seats. It's a small concession that covers natural growth without triggering a billing conversation every month.
On the threat intel feeds, get them to specify not just the feeds, but the *update frequency*. A feed included in your tier might drop from hourly to daily updates, which degrades its value while your cost stays the same.
For support costs, a fixed price isn't enough. The clause should explicitly state that the support scope, including response times and channels (phone, portal), cannot be diminished during the contract term. Otherwise, they can keep the price fixed but move you to a lower tier of service.
terraform and chill
Great advice so far, but let me zero in on something practical. Everyone's caught the "data unit" definition, but you specifically asked about **knobs they can turn later**. The biggest one nobody's mentioned yet is the API call quota.
Anomali's UI is probably just a wrapper on their APIs. Your contract likely has a line about "API access." If it doesn't define a specific, generous quota for automated access (like for your SIEM integration or custom dashboards), they can start metering and charging for that later as "excessive use." Suddenly your bill grows because your automation is working too well.
Get them to document the current, reasonable limits for API calls (requests per minute/hour) in an appendix, and lock in that anything under, say, 200% of that limit is considered fair use at no extra cost for the contract term. Otherwise, you're buying a car without knowing the price of gas.
Oh man, that API quota point is so real. It's the silent bill-killer. I'd take it a step further and specify that the documented quota applies to the *aggregate* of all your integration methods, not just a single API key. Otherwise, they might count your SIEM pull, your custom dashboard, and your backup script as three separate "users" each hitting a limit.
Also, watch for the "reasonable" loophole. If the appendix just says "subject to reasonable use," that's a blank check. Get them to state the exact numerical limits per endpoint in writing.
That's a strong operational definition for "material change." We've had success building on it by specifying the source of the "standardized sample dataset." It needs to be a mutually agreed, static data set from our production environment, archived at contract signing. Using a vendor-provided sample can skew the results, as they might optimize their new logic for it.
We also extend the clause beyond just billable unit count to include any degradation of core service levels. For instance, if a processing change increases latency by 15% for that same sample data, it also triggers a review, as the effective value of the service has changed.
RTFM — then ask for the audit
Good point on the threat intel feeds, but you have to lock down the refresh method too. If the appendix just lists "Feed X from Provider Y," they can switch the delivery from a real-time API to a daily CSV dump. The value plummets but the line item stays.
The support level clause is smart. I'd add that you need to define the escalation path and maximum resolution times in that same clause, not just response times. "1-hour response" could just mean an auto-reply ticket number.
Keep it simple
Annie, congrats on reaching the contract stage! That's a big step for a small team. 🎉
From my own stacks with monitoring tools, one knob that's often overlooked is the cost of data retention. Even if you lock in ingestion rates, check if there are tiered prices for how long data is stored. We got hit with a surprise bill when our retention period auto-increased after a year.
Also, for user seats, define what triggers a 'seat' beyond login. If your tool has API-based automation, ensure those service accounts don't count as full users. I've seen contracts where every integration token was billed as a seat, which can blow up your costs fast.
K8s enthusiast