BI dashboard build vs buy: what it costs a $6M brand

Most $6M DTC brands look at their analytics bill as a single line on the credit card statement. They see $499 a month and think that is the cost of being informed. It is not. The real question is what it costs, in dollars, over 24 months, to actually know your numbers: the software, the labor to wire it up, and the ongoing work to keep it from breaking every time Shopify changes an API or Meta reshuffles its attribution window. This is a build-versus-buy decision, and the cheapest software is rarely the cheapest stack.

We modeled four real paths for a brand doing $6M a year with a typical 5-source setup (Shopify, Meta Ads, Google Ads, Klaviyo, GA4): Glew, the managed DTC metric layer; Triple Whale, attribution-first; Looker Studio with paid connectors, the "free but not free" route; and an in-house Fivetran plus Metabase build. The finding is simple and a little uncomfortable. The operator decision is not really about features. It is about whether you pay the subscription or pay the engineer.

What "fully-loaded cost" actually means (and why sticker price lies)

Fully-loaded cost is three things added up over 24 months: the software subscription, the one-time labor to set it up, and the recurring labor to maintain it. Sticker price captures only the first. That is why a $499/month tool and a "free" tool can end up in a completely different order once you count the whole thing.

The industry data backs this up hard. In a standard BI total-cost-of-ownership breakdown, licensing is only 20-30% of the real number. Operational cost is roughly a third, and human resources, the people who build and run the thing, is the largest slice at around 45% (Kyubit's BI TCO framework). When I talk to founders running a brand this size, the mistake I see again and again is optimizing the 20% they can see on the invoice while ignoring the 45% that lives in someone's calendar.

Look at how the ranking flips. Glew Pro is genuinely the cheapest total path at $11,976 because its labor bars are near zero: the subscription bundles connector setup and maintenance. Triple Whale Advanced ($27,596) and the Fivetran + Metabase build ($32,385) land close together at the top, but for opposite reasons. Triple Whale is expensive because the software itself is pricey; the build is expensive because you are buying an engineer's time. Looker Studio sits in the middle at $16,350, which surprises people who thought "free dashboard" meant free.

Glew: the managed DTC metric layer

Glew is the closest thing to a done-for-you DTC dashboard. On its published revenue-band grid, Glew Pro runs $499/month for the $5M-$10M band, which is analytics-only. Over 24 months that is $11,976 with effectively no setup or maintenance labor, because Glew handles the connectors. There is also a heavier Glew Plus tier that bundles a managed data warehouse and runs closer to $2,800/month at $10M+ GMV, which is a different product for a different problem.

What you get is a pre-built metric layer: revenue, contribution margin, customer cohorts, product performance, all wired to Shopify and your ad platforms without you touching a connector. What you still owe is data hygiene and the acceptance that you do not own the stack. The limitations are the usual managed-tool ones: rigid contracts, real exit friction, and it is not a general-purpose BI tool you can bend to any question. Glew does not publish pricing on its own site, so the $499 figure comes from third-party snapshots and should be confirmed with a quote before you sign. It is the right pick for a brand that wants the metric layer to just exist and does not want to hire anyone to make that happen.

Triple Whale: attribution-native, not general BI

Triple Whale is the tool operators most often confuse for general BI, and the confusion is expensive. It is attribution-first. Its pricing is GMV-gated: roughly $549/month at $1.5M-$2.5M, about $1,129/month at the $5M-$7M Advanced tier, and around $1,849/month by $10M-$15M (Conjura's 2025 GMV-band grid). For our $6M brand that is $27,096 in software over 24 months, plus a bit of setup, landing at $27,596.

Here is the part the sticker price hides. Triple Whale does multi-touch attribution and creative analytics extremely well. It does not give you a P&L dashboard, an inventory tracker, or a clean cross-channel margin view. A founder who buys it gets attribution clarity and still needs a second tool (or a spreadsheet) for ops reporting. When I talk to founders about attribution, the honest line is that attribution is a bit of a black art no matter what you spend, so pay for the tool that sharpens the paid-media picture, but do not expect it to answer questions it was never built to answer.

The scaling matters too. Triple Whale steps up at every GMV band, so a brand growing from $6M to $12M should expect the line to climb, not hold. Glew Plus stays flat through the middle then jumps at scale when the managed warehouse kicks in. Neither is "cheaper" in the abstract; the answer depends on where your GMV lands over the next two years.

Looker Studio: free but not free

Looker Studio is the path that looks free until the founder has to call a developer. The dashboard platform is $0 (or $9/user/month for the Pro edition), but it does not connect to your DTC sources on its own. A 5-source connector package (Supermetrics Starter or similar) runs about $150/month. Then someone has to build it: 20-40 hours of data-engineering work at $65-$120/hour, and 4-8 hours a month of maintenance to keep it alive when a platform changes an API.

Put realistic numbers on it and the 24-month total lands around $16,350: roughly $3,600 in connectors, $2,550 in setup, and $10,200 in ongoing maintenance labor at $85/hour. Notice the shape. The software is the small part; the maintenance labor is the biggest single line. This is the path where "we saved money on tools" quietly becomes "our ops person spends a day a month babysitting connectors." When we've seen this go well, it is because the brand already had a data-savvy ops hire who owned it. When we've seen it go badly, it is a founder who priced the $150 connector and forgot the human.

Fivetran + Metabase: the in-house build

The build path buys you ownership. Fivetran handles the pipes (priced per connector at roughly $500 per million monthly active rows since the March 2025 billing change). For a low-to-moderate-volume 5-source DTC stack, that lands around $500-$580/month (roughly $116/connector if each connector carries close to its minimum MAR threshold). That is the model's baseline, and it is important to name the assumption: it holds only if your catalog syncs and order-update frequency are light. A busier feed, or a brand with high-frequency Meta pixel events, can push each connector's MAR well past the minimum, and Mammoth.io's 2026 guide puts a realistic 3-5 connector setup at $1,500-$3,000/month for brands with typical transaction volumes (their analysis cites a 40-70% cost increase for multi-connector setups). Use the $580/month figure as your best-case floor; build your downside case at $1,500/month and model the difference. Metabase Cloud Starter adds $100/month (the base tier includes 5 users). At the floor assumption, that is $13,920 in Fivetran and $2,400 in Metabase over 24 months. Then the engineer: 30-60 hours to build it and 5-8 hours a month to run it. On a freelance data engineer at $85/hour, the whole thing totals $32,385 over 24 months at the low-volume floor, or roughly $57,000-$67,000 at the Mammoth.io realistic range.

The benefit is real: you own the warehouse, the logic is extensible, and it can grow into an actual data platform rather than a rented dashboard. The risk is equally real: you own the maintenance burden, and it does not go away.

This is the chart that settles most build-vs-buy arguments at this size. The software is identical in both scenarios. The engineer is the entire swing. A freelance data engineer keeps the build near $32K over 24 months. The same stack with a full-time hire blows past $76K. That figure uses a 12-month-equivalent slice ($60K, which is 75% of an $80K loaded annual salary), not the full two-year allocation; if you charge the full 24 months of that 75% allocation the labor alone reaches $120K. Either way, for a $6M brand, BI is not a full-time job, so the freelance or fractional route is almost always the right economic call. The pattern we see again and again is founders hiring a full-time analyst to justify a build, then discovering the person is idle three weeks a month.

The decision matrix: which path for a $6M brand?

There is no single winner; there is a winner per problem. Before you price any of them, get clear on the metrics that actually run a DTC brand, because the right tool is the cheapest one that reports the numbers you have decided to own. If your problem is attribution, Triple Whale earns its price. If you want a general ops and margin view with zero engineering, Glew Pro is the cheapest complete answer. If you have a data-savvy person on staff, Looker Studio or a Metabase build gives you control at a labor cost you can absorb. The capability matrix below is the honest version of the sales-page comparison.

The tool is never the answer. All of these dashboards are just the guardrails on the road, and only a really smart person can make sense of a dashboard anyway. What separates the brands that get value is not which vendor they picked, it is whether one person owns the number and looks at it every week. Buy the cheapest complete answer to your actual problem, then spend your energy on the human who reads it.

Hidden costs nobody puts in the brochure

Three costs never make the pricing page. First, exit and lock-in. Cancel Triple Whale or Glew and you keep your raw Shopify and ad-platform data, but the modeled views, custom metrics, and attribution history do not export cleanly. That switching cost is a real tax on the buy decision. Second, overage mechanics: the Fivetran build-path model uses a low-volume floor of $500-$580/month, but Fivetran's per-connector MAR billing (since March 2025) means a catalog resync, a high-frequency order feed, or a busy Meta pixel can push each connector's bill materially higher. Mammoth.io's guide puts a realistic 3-5 connector setup at $1,500-$3,000/month. Run a MAR estimate for your actual catalog and order frequency before you commit to the build path. Triple Whale's GMV bands mean growth automatically raises your price. Third, the cost of bad data, the decisions you make on a dashboard that quietly broke when an API changed and nobody noticed for six weeks. On the managed paths (Glew, Triple Whale) the vendor absorbs that risk. On the DIY paths it is yours, and it is the single most common reason a "cheap" build turns expensive.

Related reading. For the dashboard finance and marketing should share, see the finance and marketing dashboard every DTC brand needs, and for the cash view that sits behind it, see how to build a 13-week cash flow forecast. For how we stand up reporting a board actually trusts, see our fractional CFO work.

Sources and methodology

All figures modeled for a $6M DTC brand with a 5-source stack. The comparison assumes Shopify, Meta Ads, Google Ads, Klaviyo, and GA4, with pricing taken at the $5M-$7M revenue band. Labor is modeled at $85/hour, the midpoint of the $65-$120 mid-level US freelance data-engineer range, applied to path-appropriate setup and maintenance hours. Managed tools (Glew, Triple Whale) carry near-zero labor because they bundle connector setup and upkeep.

Vendor pricing is from third-party snapshots, not scraped from live quote tools. Glew does not publish Pro/Plus pricing on its own site; the $499/month figure comes from a TrustRadius pricing snapshot, with the $2,800/month Glew Plus scale scenario from a Polar Analytics competitive comparison. Triple Whale's GMV-band pricing is from Conjura's 2025 breakdown; the vendor's own GMV slider is the authoritative source and should be checked before you commit.

Build-path pricing draws on current vendor guides. Fivetran's per-connector, monthly-active-rows model is from the Mammoth.io 2026 pricing guide and confirmed against Fivetran documentation; the post-2025 per-connector model is material, so older account-level pricing articles are outdated. The $500-$580/month Fivetran figure models a low-to-moderate-volume 5-source stack where each connector's MAR stays close to its minimum threshold. Mammoth.io puts a realistic 3-5 connector setup at $1,500-$3,000/month for brands with typical transaction volumes (citing a 40-70% multi-connector cost increase vs. single-connector estimates); treat $580/month as the best-case floor and model $1,500/month as the realistic midpoint for planning. Metabase Cloud Starter pricing ($100 base plus $6/user, 5 users included) is from the official Metabase pricing page.

Labor rates are from a crowdsourced benchmark. Freelance data-engineer rates come from contractrates.fyi, which reports an $83/hour average (as of March 2026), cross-referenced against published mid-level ranges of $65-$120/hour. The full-time DE scenario uses an $80K/year loaded salary allocated 75% to BI; the $60K labor figure in the chart is a 12-month-equivalent slice of that allocation, not a full 24-month charge.

The TCO ratios come from a published BI framework. The finding that licensing is only 20-30% of total BI cost is from the Kyubit BI total-cost-of-ownership analysis, which splits a representative BI TCO into roughly 20% acquisition, 35% operational, and 45% human resources. Directional support for mid-market build totals comes from ecommerce-analytics cost frameworks by Ask-Luca and Improvado.

Frequently asked questions