10 Best Self-Service Analytics Tools for Marketers and Business Teams in 2026

Marketing teams run on data, but getting a straight answer out of that data is another story. Reports sit in a queue, dashboards need an analyst to build, and by the time the numbers arrive, the campaign has already moved on.

The cost shows up fast. 91% of marketers take at least several days to react to a campaign that’s clearly underperforming, and one in three often wait weeks. If you’re shopping for a self-service analytics tool, that delay is probably why.

Self-service is supposed to fix it: ask a question, get an answer, skip the ticket to the data team. But many tools still need SQL or an analyst, and long reporting times remain marketers’ biggest barrier to acting on insights (33%).

So we compared 10 platforms on what buyers actually need: answers without SQL, numbers you can trust, connections to your marketing data, and the ability to act on what you find. Our top pick is aiRA by Capillary, which handles analysis and campaign execution in one conversational workflow, so there’s no handoff between the question and a live campaign.

aiRA isn’t the right fit for everyone. If you have an analyst-led BI team or mainly need dashboards, a tool further down the list may suit you better. Here’s who each one is for.

TL;DR

• Best overall and best for marketers: aiRA by Capillary. Combines analysis and campaign execution in one workflow, so insights turn into action without a separate tool.
• Best for non-technical teams: ThoughtSpot. Search-style questions in plain English on well-modeled data.
• Best for enterprise: Looker. Governed metrics and a semantic layer for large teams that need one version of the truth.
• Best budget pick: Metabase. An open-source option that works well for startups and smaller teams.

What Is a Self-Service Analytics Tool?

A self-service analytics tool is software that lets non-technical people query, visualize, and share data on their own, without waiting on a data engineer or analyst. You ask a question, you get an answer. Self-service analytics software covers the same ground: it’s the layer that sits between your data and the people who need to make decisions with it.

How it differs from traditional BI

In a traditional BI setup, analysts build the reports and everyone else requests them. That works until the queue gets long, which is usually when you need an answer most. Self-service flips the model: the marketer, the ops lead, or the growth manager explores the data directly, and the data team focuses on keeping the foundation clean instead of answering one-off questions.

It’s also different from embedded analytics. Embedded analytics puts dashboards inside a product for that product’s end users, like the usage reports a SaaS app shows its customers. Self-service analytics is for your own team, working on your own data.

What’s under the hood

Most self-service analytics tools are built from the same five pieces:

• Data connectors: pull data from your warehouse, CRM, ad platforms, and other sources
• Semantic layer: defines metrics once so everyone gets the same number for “revenue” or “churn”
• No-code or natural language querying: lets you ask questions without writing SQL
• Dashboards: visualize and share what you find
• Governance: controls who can see and change what, so access doesn’t turn into chaos

The quality of each piece varies a lot between platforms, which is why two tools that both say “self-service” can feel completely different in practice. One more thing to watch: most tools stop at the dashboard. They tell you what happened, and acting on it is still a separate job. We’ll come back to that gap throughout this guide, because it’s what separates a tool that informs your decisions from one that helps you execute them.

Why Teams Are Switching to Self-Service Analytics

If you’re reading this, you’ve probably felt the pain already: you need a number, you ask for it, and then you wait.

The global self-service analytics market was valued at USD 5.70 billion in 2025 and is projected to reach USD 21.93 billion by 2034, a 16.14% CAGR. That growth reflects a clear shift: organizations are investing in these tools because delays in reporting now carry a real cost.

And the problem usually isn’t a lack of data, 53% of marketers say they don’t consistently turn data insights into campaign action. The numbers are there. They just show up late, sit in the wrong tool, or need someone else to explain them first.

Here’s what that looks like day to day:

• A weak campaign keeps running because the report is still in the queue
• Budget keeps flowing to a channel that stopped performing days ago
• Everyone builds their own spreadsheet, so no two people report the same number

None of these fix themselves, and they get pricier the longer they go on. The faster your team can get its own answers, the faster they stop.

How We Evaluated These Platforms

Most comparison guides in this space rank tools by feature count: more integrations, more dashboards, more AI badges. That tells you what a tool can technically do. It doesn’t tell you whether your team will get answers faster, and that’s the only thing that should drive a purchase.

So we scored each platform on four questions a marketing or business buyer actually cares about.

1. Can a marketer get an answer without SQL or an analyst?
   This covers natural language querying, no-code exploration, and how steep the learning curve is. If every question still needs someone technical to translate it, it’s not really self-service.
2. Can you trust the number?
   Speed means nothing if two people get two different answers. We looked at semantic layers, governed metrics, security, and permissions, because trust in the data is what makes people actually use the tool.
3. Does it connect to where your marketing data lives?
   That means your CRM, ad platforms, warehouse, and loyalty or commerce data. This matters more than it sounds. 92% of marketers are already experimenting with AI in their reporting workflows, yet slow reporting is still their biggest operational headache. Smarter features don’t help much if the tool can’t reach your data in the first place.
4. Can the platform act on the insight, or does a person still have to?
   This is the question most comparison guides skip. We checked whether each tool closes the gap between finding an insight and doing something about it, or leaves that step for your team to handle manually in another tool.

Every platform was scored across all four, not ranked on one standout feature. A tool with deep reporting but no way to act serves a different buyer than one with lighter reporting that can launch a campaign from a prompt. Both have a place. The right pick depends on which part of the process is slowing your team down.

Best Self-Service Analytics Tools Comparison Table

Here’s the quick view before we get into each tool. The last column is the one to watch, since it shows which platforms stop at the answer and which can help you act on it.

10 Best Self-Service Analytics Tools in 2026

1. aiRA by Capillary

aiRA by Capillary is Capillary Technologies’ agentic AI marketing and loyalty co-pilot. It brings segmentation, reward optimization, campaign creation, and performance analytics into one conversational interface, so your team isn’t hopping between tools to get from a question to a live campaign.

Here’s what sets it apart from the rest of this list. With most self-service tools, you ask a question, get a chart, and then someone has to go build the campaign somewhere else. With aiRA, asking and acting happen in the same workflow, with human approval at each step, so your team stays in control of what goes live.

The results back that up:

• Fashion retailer in Saudi Arabia: during a Ramadan campaign, aiRA outperformed a dedicated analytics agency on 8 of 9 measured KPIs, including a 29.4% increase in new customer acquisition.
• Consumer superapp: promo creation ran 50% faster with zero configuration errors.
• Indian footwear brand: aiRA helped build a six-persona framework that supported 85.3% year-on-year retention.

Key features

• Conversational analysis and campaign launch in a single workflow
• Segmentation and propensity modeling
• Reward and loyalty optimization
• Omnichannel campaign coordination
• Analytics reported back in the same system you launch from

Honest limitations

• Delivers the most value for retail and loyalty-led marketing teams
• Best suited to teams ready to run campaigns, not just view dashboards
• Implementation is typically weeks and starts with a pilot

Pricing: Custom, quote-based.

Who it’s for: aiRA by Capillary is a strong fit for marketers who want analysis and execution in one place and are tired of the handoff between insight and action. If your team mainly needs generic BI dashboards across finance and operations, one of the tools further down this list will likely suit you better.

Want to see how this works with your own data? Book an aiRA demo and walk through a live campaign build.

2. ThoughtSpot

ThoughtSpot is built around a simple idea: you should be able to ask your data a question the way you’d ask a colleague. Its search-based interface and Spotter AI assistant let business users type a plain-English question and get an answer, without writing SQL or waiting on a report.

The catch is in the setup. ThoughtSpot works best when your data has already been modeled, so a data team usually needs to do the groundwork before business users can start asking questions. Once that’s in place, it’s one of the more approachable options for non-technical teams.

Key features

• Search-style questions in plain English
• Spotter, the AI assistant, for conversational analysis
• Shareable Liveboards for dashboards and reporting
• Connects to cloud data warehouses
• Embedded analytics option for putting insights inside your own product

Honest limitations

• Needs data-team setup and modeling before business users can self-serve
• Reporting only: once you have the answer, acting on it happens in another tool
• AI features and data limits depend on your tier, so a plan that looks affordable can force an upgrade conversation later
• Implementation typically runs weeks to months

Pricing: ThoughtSpot publishes per-user tiers. Essentials starts at $25 per user per month and Pro at $50, billed annually, with Enterprise on custom quotes. Watch the limits as you scale: Essentials tops out at 25 million rows and 50 users, and Pro at 250 million rows. Before you commit, check which tier includes the AI agent and how much usage it allows.

Who it’s for: ThoughtSpot is a strong fit for non-technical teams who want to ask plain-English questions on well-modeled data, and have a data team to set that up. If you want your marketers to go from a question to a live campaign without leaving the tool, or you don’t have data resources to spare for modeling, it’s less of a match.

3. Looker

Looker is Google Cloud’s enterprise BI platform, and its big selling point is consistency. Metrics are defined once in a central semantic layer called LookML, so when the CFO and the head of marketing both pull “revenue,” they get the same number. If your teams have ever lost a meeting to the question of whose spreadsheet is right, you’ll see the appeal.

Business users can then explore that governed data on their own, including through Gemini, Looker’s AI assistant. The tradeoff is that someone has to build and maintain the model first, which makes Looker more of a data-team project than a pick-it-up-and-go tool.

Key features

• LookML semantic layer for governed, consistent metrics
• Gemini-powered, AI-assisted natural language exploration
• Dashboards and reporting that business users can explore without writing SQL
• Data actions that can trigger limited follow-up steps from a report
• Strong fit with Google Cloud and BigQuery environments

Honest limitations

• LookML needs developers, so the setup falls to your data team
• Implementation typically takes months
• Reporting first: data actions help, but launching a campaign still happens elsewhere
• Not a lightweight tool for a marketing team that wants answers this week

Pricing: Custom, quote-based.

Who it’s for: Looker is a strong fit for enterprises that need governed, consistent metrics across teams and have the data resources to model them. If your marketers want to move quickly without waiting on a data team, or you want to act on an insight in the same place you found it, it’s likely more tool than you need.

4. Power BI

Power BI is Microsoft’s BI tool, and its biggest advantage is that it fits neatly into the Microsoft world. If your team already lives in Excel, Teams, and Azure, Power BI feels familiar from day one, and getting data in and reports out takes less effort than with most alternatives.

It’s also a capable tool well beyond marketing. Finance, operations, and sales can all report from the same place, which is a real plus for organizations that want one BI platform across the business. In most setups, though, analysts build the reports and business users mostly view them.

Key features

• Deep integration with Excel, Teams, Azure, and the wider Microsoft stack
• Natural language questions through Copilot and Q&A
• Wide range of data connectors and flexible data modeling
• Shareable dashboards and reports across departments
• Activation through Power Automate and Power Apps, set up separately

Honest limitations

• Analysts typically build the reports, and business users mostly view them
• Reporting first: acting on an insight means wiring up Power Automate or Power Apps
• AI features can depend on your license or capacity tier, so check what your plan actually includes
• Implementation typically takes weeks to months
• Less of an advantage if your stack isn’t Microsoft-based

Pricing: Per-user licensing keeps entry costs low. Pro is $14 per user per month and Premium Per User is $24, with Fabric capacity pricing for larger rollouts. Costs can climb quickly once you factor in capacity and AI features, so it’s worth mapping out your license mix before you commit.

Who it’s for: Power BI is a strong fit for teams already in the Microsoft ecosystem that want a cost-effective BI tool across departments and have analysts to build the reports. If your marketers want to go from question to live campaign without a separate build step, or your stack lives outside Microsoft, you’ll probably be happier with another option.

5. Tableau

Tableau is the tool a lot of analysts reach for when the visualization itself matters. It’s known for flexible, polished dashboards and a level of control over how data is explored and presented that’s hard to match. Now part of Salesforce, it also connects well with the Salesforce ecosystem.

That depth comes with a catch for self-service. Tableau is analyst-led: skilled people build the dashboards, and most business users interact with what’s already been built. Newer AI features like Pulse and Tableau Agent make it easier for non-technical users to ask questions, but the heavy lifting still sits with your analysts.

Key features

• Deep, highly customizable data visualization
• Tableau Pulse for proactive metric insights
• Tableau Agent for AI-assisted analysis in natural language
• Tableau Prep for cleaning and shaping data
• Activation through Salesforce Flow and Einstein, set up separately

Honest limitations

• Analyst-led, so business users mostly explore dashboards built for them
• Reporting first: acting on an insight means wiring up Salesforce Flow or Einstein
• AI features can sit in higher tiers or add-ons, which raises the real cost
• Every deployment needs at least one Creator license
• Implementation typically takes weeks to months

Pricing: Role-based, per user, billed annually. On the Standard edition, Creator is $75 per user per month, Explorer $42, and Viewer $15, with Enterprise tiers priced higher. Watch the add-ons, because Tableau Agent and premium Pulse features can push costs up.

Who it’s for: Tableau is a strong fit for analyst-led teams that want deep, flexible visualization and have the people to build and maintain it. If your marketers want to ask a question and get an answer without waiting on an analyst, or you want to launch a campaign from the same place you found the insight, it’s a better fit as a reporting layer than a complete solution.

6. Sigma Computing

Sigma takes a different angle on self-service. Instead of asking business users to learn a new BI interface, it gives them one they already know: a spreadsheet. Rows, columns, pivot tables, and formulas, but running live on your cloud data warehouse, so nobody is exporting CSVs or working from stale copies.

For teams that think in spreadsheets, that’s a big shortcut. It also goes a step beyond most BI tools with write-back, which lets users enter data, like forecasts or budget assumptions, and push it back to the warehouse.

Key features

• Spreadsheet-style interface on live cloud warehouse data
• Ask Sigma and AI assistance for natural language questions and formulas
• Input tables for write-back to the warehouse
• Collaborative workbooks with version history
• Embedded analytics for customer-facing use

Honest limitations

• Built around a cloud data warehouse, so you need one in place
• Write-back updates your data, but it doesn’t launch campaigns or act on insights in your marketing channels
• Works best for people who are comfortable with spreadsheet formulas and logic
• Costs can shift to your warehouse, since queries run there
• Cloud-only, with no self-hosted option

Pricing: Not published. Sigma is contact-sales only, with pricing based on user licenses across functionality tiers, so you’ll need a quote. A free trial is available.

Who it’s for: Sigma is a strong fit for spreadsheet-first teams, like finance, ops, and analytics-minded marketers, who want to work on live warehouse data without learning a traditional BI tool. If you need to go from insight to a live campaign in the same place, or you don’t have a cloud data warehouse yet, it’s not the right match.

7. Metabase

Metabase is the friendly, no-nonsense option. It’s open source, quick to get running, and built so that people who don’t write SQL can still ask questions of their data through a point-and-click interface. Analysts can drop into SQL when they need to, so one tool serves both groups.

That mix of low cost and low friction is why startups and smaller teams love it. The flip side is that it’s a general-purpose BI tool. It connects to your databases and warehouses, so marketing data from ad platforms and your CRM needs to land there first.

Key features

• Point-and-click question builder for non-technical users, with SQL for analysts
• Dashboards you can share across the team
• AI-assisted questions through Metabot
• Connects to 20+ databases
• Embedded analytics option
• Free open-source edition you can self-host

Honest limitations

• Reporting only: acting on an insight happens in another tool
• Marketing data has to be in your database or warehouse first
• Advanced permissions, single sign-on, and white-labeling sit on higher paid tiers
• Self-hosting the free edition needs engineering time
• Lighter on deep data modeling and governed metrics than enterprise tools like Looker

Pricing: Starts at zero, with the open-source edition free to self-host. On the cloud side, Starter is $100 per month and Pro is $575 per month, with per-user fees on top, and Enterprise starts around $20,000 per year. Optional extras like Metabot AI can add to the bill, so check what your plan includes.

Who it’s for: Metabase is a strong fit for startups and smaller teams that want fast, affordable self-service analytics and have someone who can set up the data connections. If you need governed metrics at enterprise scale, or you want your marketers to go from a question to a live campaign without leaving the tool, you’ll outgrow it quickly.

8. Hex

Hex is built for data teams first. It’s a collaborative workspace where analysts write SQL and Python in shared notebooks, with an AI agent that helps write and run the analysis alongside them. When the work is ready, they publish it as an interactive app or dashboard that the rest of the company can use.

Business users get in through Threads, Hex’s conversational interface, where they can ask questions in plain English against data the team has set up. So Hex does serve non-technical people, but the platform is shaped around the people who build things for them.

Key features

• Collaborative SQL and Python notebooks
• Notebook Agent for AI-assisted analysis
• Threads for conversational questions from business users
• Published data apps and dashboards
• Semantic layer tools so the AI works from trusted definitions
• Slack integration for asking data questions

Honest limitations

• Data-team-led, with business users getting access through the apps and Threads the team sets up
• Reporting only: acting on an insight happens elsewhere
• The conversational experience depends on your data team keeping the semantic context in good shape
• Editor seats are the main cost, and compute and AI credits can add to it
• Lower tiers cap published apps, so growing teams may hit limits

Pricing: Pricing is published and priced per editor seat. Community is free, Professional is $36 per editor per month, and Team is $75, with Enterprise on custom quotes. Keep in mind that editor seats, compute, and AI usage are the three cost layers that can push total spend above the base price.

Who it’s for: Hex is a strong fit for data teams that want notebooks and AI in one place, and want to give the rest of the company a friendlier way to explore their work. If your marketers need to run their own analysis and launch campaigns without leaning on a data team, it’s more of a data team tool than a marketer tool.

9. Amazon QuickSight

Amazon QuickSight is AWS’s BI service, now branded Quick Sight and part of Amazon Quick Suite. If your data already lives in AWS, it’s a natural fit: it connects directly to sources like Redshift, Athena, S3, and RDS, runs without servers to manage, and scales as your user base grows.

Natural language questions come through Amazon Q, where authors set up Q Topics on specific datasets so business users can ask questions in plain English. Higher tiers add generative AI extras, like executive dashboard summaries and data stories.

Key features

• Direct connections to AWS data sources like Redshift, Athena, S3, and RDS
• Amazon Q for natural language questions on topics your authors set up
• Executive summaries and generative data stories on Pro tiers
• Interactive dashboards with embedded analytics options
• Serverless setup that scales with your user base

Honest limitations

• Built for AWS, so it’s most useful if your data and team already live there
• Reporting only: acting on an insight happens in another tool
• AI features sit on Pro tiers, and authors need to set up Q Topics before business users can ask questions
• Pricing has several moving parts: author and reader licenses, Pro tiers, sessions, and storage
• Product names and bundles keep shifting, which can make evaluation confusing

Pricing: Per-user and usage-based. Readers start at around $3 per user per month and Authors at around $24, while Pro tiers with generative AI features cost more, such as Reader Pro at about $20. An account-level fee can also apply when Pro users have Q&A enabled, so confirm the current numbers on the AWS pricing page before you budget.

Who it’s for: QuickSight is a strong fit for AWS-native teams that want a cost-conscious, scalable BI layer next to the data they already have in AWS. If your stack lives elsewhere, or you want your marketers to go from a question to a live campaign without leaving the tool, you’ll likely want a different option.

10. Databricks

Databricks is less a BI tool and more a full data platform. Its lakehouse approach puts data engineering, analytics, and AI on one governed foundation, and BI sits on top of that. For companies that already run their data on Databricks, that’s a big draw, because analytics happens right where the data lives.

For self-service, Genie is the piece that matters. It lets business users ask questions of governed data in plain English, and Databricks One gives them a simpler front door to it. Behind that, though, data engineers and analysts do the heavy lifting of building and maintaining what business users see.

Key features

• Lakehouse platform that unifies data engineering, analytics, and AI
• Genie for natural language questions on governed data
• Databricks One as an entry point for business users
• Databricks SQL for BI queries and dashboards
• Unity Catalog for governance and access control

Honest limitations

• Needs in-house data engineers to build and maintain the foundation
• Implementation typically takes months
• Gives you analysis and natural language answers, but activation needs extra tools
• Usage-based pricing is harder to predict than a flat per-seat plan
• Overkill if you only need self-service reporting for a marketing team

Pricing: Consumption-based, billed in Databricks Units (DBUs) and quoted by sales. Genie is now metered too: Databricks began charging for Genie usage beyond a free monthly allowance on July 6, 2026, and the SQL warehouse compute that runs each query is billed separately. For a sense of the rate structure, third-party trackers list rates that vary by workload type.

Who it’s for: Databricks is a strong fit for enterprises with in-house data engineers that want one platform for data, analytics, and AI, with natural language access layered on top. If you’re a marketing team that wants answers and actions without a data engineering project behind them, it’s a heavier lift than you need.

Your Checklist Before the Demo Call

A vendor demo is built to show you what the platform does well. It won’t volunteer what it doesn’t do. These questions help you close that gap before you sign anything, so bring them along and ask them out loud.

• Integrations: Ask whether the platform connects to your CRM, ad platforms, warehouse, and loyalty or commerce data without extra engineering. If the answer involves a multi-month project before you see any value, factor that cost in now, not after the contract is signed.
• Real-time vs batch: Find out whether you’re working with data as it happens, or data that’s a day old by the time you act on it. For most marketing decisions, that lag is the difference between a timely move and a missed one.
• Usability for non-analysts: Bring someone from your marketing team to the demo, not just your data lead. If every question still needs an analyst to translate it into a query, that dependency won’t disappear after onboarding. It becomes part of how the tool gets used.
• Governance and trust: Ask how metrics are defined and who controls access. This matters more than it sounds, because Gartner has warned that self-service initiatives across decentralized teams often turn into siloed analytics that are rarely tied to measurable business outcomes. If nobody owns the definitions, you’ll get fast answers that nobody trusts.
• Insight to action: Look past the dashboard walkthrough and ask what happens after the answer shows up. Does the platform stop at telling your team what to do, or can it do something about it?
• Pricing transparency: Push for a real number, including what it costs at your team size and once you add AI features or heavier usage. If every answer routes back to “let’s get you on a call with sales,” pricing is probably negotiated case by case.

One question matters more than the rest, so make it the one you lead with: does this tool need a separate step to turn an insight into a live campaign, or can it launch one on its own? Most platforms on this list will answer the first five questions well. Very few can say yes to this one, and that single answer will tell you more about fit than anything else in the demo.

Where This Leaves You

Self-service analytics has moved past the question of whether you can ask a data question without writing SQL. Most tools on this list can do that now. The real question is what happens after you get the answer.

That’s where the list separates. Most platforms handle the early checklist items well: integrations, usability, governance, pricing. And those basics matter. Organizations that get real value from analytics and AI tend to invest in trustworthy data, governance, and security first. Get those wrong and no tool will save you.

But even with a solid foundation, there’s a handoff most teams have learned to live with. A tool tells you which segment to target or which campaign is slipping. Then someone builds the campaign, routes it for approval, and pushes it live in another system, on another timeline. That handoff is where speed dies, and it’s the step most comparison guides never ask about, because a feature list doesn’t capture it.

That’s the gap aiRA by Capillary was built to close. Instead of ending at the insight, it moves from a plain-language question to a live campaign in the same workflow, with your team approving each step. In a Ramadan campaign for a fashion retailer in Saudi Arabia, aiRA outperformed a dedicated analytics agency on 8 of 9 measured KPIs, a strong sign that closing the loop between insight and action beats the traditional handoff.

Every extra day your team spends waiting on a dashboard is another day campaigns run on stale data. If you’re choosing a self-service analytics tool this quarter, pick one that doesn’t stop at the answer.

See what aiRA can do for your marketing team. [Book a demo call]

FAQ

What is the best self-service analytics tool for marketers?
It depends on what slows your team down. If the bottleneck is going from an insight to a live campaign, aiRA by Capillary is our top pick, since it handles analysis and campaign execution in one workflow. If your team mainly needs dashboards and has analysts to build them, Tableau or Power BI are safer choices. For a smaller budget, Metabase is a solid place to start.

What is a self-service analytics tool?
It’s software that lets non-technical people query, visualize, and share data without waiting on a data engineer or analyst. You ask a question, you get an answer, and the data team stays focused on keeping the foundation clean instead of answering every request.

What is the difference between self-service analytics and traditional BI?
In traditional BI, analysts build the reports and everyone else requests them, which means a queue. In self-service analytics, the business user explores the data directly. The tradeoff is that self-service still needs good data modeling and governance behind it, or you end up with fast answers that nobody trusts.

Do you need SQL to use self-service analytics software?
Not for everyday questions. Most tools on this list let you ask in plain English or build charts with point-and-click controls. That said, some of them, like ThoughtSpot and Looker, need a data team to model the data first, so SQL skills may still be needed behind the scenes even if your marketers never touch them.

How much does a self-service analytics tool cost?
It ranges widely. Metabase’s open-source edition is free to self-host, Power BI Pro is $14 per user per month, and ThoughtSpot Essentials starts at $25. Enterprise-grade options like Looker, Sigma, and Databricks are quote-based, and aiRA by Capillary is custom-priced as well. Whatever you choose, ask what AI features, extra usage, and add-ons cost, since that’s where bills tend to grow.