The question "what tools should we be running?" comes up in almost every consulting engagement. It's the wrong question, asked at the right time. The better question is: at this stage, what's the highest-leverage place to add tooling — and what are we running that we've outgrown?
Tools are a downstream decision. The upstream decisions are: what does your ICP look like, what channels are you investing in, and what data do you actually need to make decisions? The tools should answer those questions. If you're buying tools before those questions are answered, you'll end up with a $60K/year stack that tells you nothing useful.
That said: here's what the stack looks like at each stage, and where the real switching costs are.
$1M–$3M ARR: Stay Lean
At this stage, the constraint is not tooling sophistication — it's signal. You don't have enough closed-won data to run meaningful cohort analysis. You don't have enough traffic to A/B test with statistical significance. You don't have enough pipeline to need automated lead routing.
What to run:
CRM: HubSpot Starter or Pipedrive. Not HubSpot Pro. The reporting in Starter is enough to track pipeline and closed revenue. You don't need lead scoring or sequences at this stage — you need discipline in logging activity. A simple CRM used consistently beats a sophisticated CRM used inconsistently.
Email: Loops, Customer.io, or HubSpot's email tool. Pick one. The deliverability and feature differences are marginal at this volume. The switching cost later is higher than most teams anticipate.
Analytics: GA4 plus a product analytics tool (Posthog or Mixpanel Starter). GA4 for acquisition; product analytics for activation and retention. You need both views.
Attribution: UTM parameters tracked to the CRM source field. That's it. Any more sophisticated attribution model at this stage produces false precision — you don't have the data volume for it to be reliable.
What to skip: SEO platforms, ABM tools, sales intelligence databases, marketing data warehouses. These become useful later. At this stage they're expensive distractions.
$3M–$10M ARR: The Upgrade Window
This is where most of the tooling decisions that actually matter get made. You've validated at least one channel. You have 6-12 months of closed-won data. You're starting to see patterns worth instrumenting.
CRM: Upgrade to HubSpot Pro or migrate to Salesforce. This is the most consequential tooling decision at this stage. HubSpot Pro's reporting is legitimately good for marketing attribution; Salesforce's flexibility is better if you have a complex sales process or need custom objects. The cost of migrating between them later is significant — CRM data is sticky. Make the decision deliberately.
Attribution: Move beyond UTM-only to multi-touch attribution. HubSpot Pro has decent native attribution. If you're running significant paid alongside content, you need to understand the assist value of your content — last-click attribution will misrepresent it.
SEO: Ahrefs or Semrush. Whichever you choose, use it for keyword gap analysis and backlink monitoring, not as a content calendar generator. The AI-generated "keyword opportunity" lists these platforms produce are useful for identifying topics, not for deciding what's worth writing.
Email sequences: If you have a sales team, Outreach or Apollo for sequencing. If you're product-led, Customer.io or Loops for lifecycle email. Don't confuse the two — they're different tools solving different problems.
Paid: Google Ads + Meta Ads. Enough volume to feed the algorithms (minimum $5K/month per platform before you'll see stable optimisation). Below that threshold, the platforms don't have enough signal and performance is unpredictable.
$10M–$20M ARR: Operationalise the Stack
At this stage you're not choosing the stack — you're operationalising it. The tooling decisions are mostly made. The leverage is in making the existing tools compound on each other.
Revenue operations: This is where a RevOps function (person or tool) starts to justify its cost. The data your CRM produces is only useful if it's clean and the definitions are consistent. Revenue stage definitions, pipeline hygiene, attribution consistency — these are operational problems, not tooling problems. HubSpot Pro's reporting is sophisticated enough to surface them if the underlying data is clean.
Data warehouse: At $10M ARR with significant marketing spend across multiple channels, you probably have a reporting problem. GA4, CRM, paid platforms, email — data in four places that tells four different stories. A lightweight data warehouse (BigQuery or Redshift) with a BI tool (Metabase or Looker Studio) unifies the view. This is overkill at $5M. At $10M+, the cost of not having a unified view is missed attribution and misallocated spend.
ABM: Only if your ICP is well-defined and your sales cycle is 60+ days. ABM tools (6sense, Demandbase) are expensive ($30K–$80K/year) and return their cost only at meaningful deal sizes. Below $50K ACV, the math rarely works.
The Switching Costs That Actually Hurt
Most teams underestimate these:
Email platform migrations are expensive because your deliverability reputation doesn't transfer. Warm-up periods on a new domain/platform can take 4-6 weeks and require careful volume throttling.
CRM migrations are expensive because historical data is messy and your team's workflows are embedded in the current tool's UX. Budget 6-8 weeks for a proper migration with no data loss.
Analytics platform migrations lose historical data — GA4 data doesn't export cleanly to Mixpanel. If the comparative data matters (and it usually does), plan the migration carefully.
The Principle Underneath All of It
Tools compound when the data from one feeds decisions in another. Email open signals feed CRM lead scoring. CRM closed-won data feeds paid audience suppression. Paid channel data feeds content prioritisation. When that feedback loop is working, the stack is earning its cost.
When the tools are siloed — each reporting independently without informing the others — you're paying for dashboards instead of decisions.
The audit question for any tool you're running: what decision does this data make better, and is that decision actually being made better because of it? If the answer is unclear, the tool is probably overhead.