Buying revenue attribution software usually makes measurement more expensive, not more truthful.
The uncomfortable result of a revenue attribution software comparison is that most companies buying these platforms are replacing CRM discipline they do not have. The tool inherits the same broken fields and missing sales events at a higher monthly cost.
Why reported Google Ads ROAS can mislead covers the paid-media version of this problem. Dedicated attribution software becomes useful only when clean identities, commercial outcomes and adequate volume already exist.
Our position is blunt: no tool, native reporting, specialist software and a custom build are separate decisions, not maturity badges.
Revenue attribution software comparison: pass the readiness test
Four conditions must all pass before a product demo becomes a procurement project.
- Identity coverage: at least 90% of closed-won and closed-lost deals must join to a known company or contact. Anonymous sessions can provide context, but they cannot anchor revenue.
- Outcome completeness: at least 95% of closed deals must contain usable dates, status, value and stable identifiers. This tests data hygiene without redesigning the CRM.
- System reconciliation: compare the same named event across the advertising platform and CRM for 30 days. The difference must be 10% or less. In an illustrative account, Google Ads reporting 64 accepted opportunities versus 31 in the CRM is a tracking dispute, not an attribution opportunity.
- Decision-grade volume: require at least 60 closed outcomes, won plus lost, for each materially different sales motion over the previous 12 months. Below that threshold, simple source and cohort views are more defensible than elaborate fractional credit.
If one condition fails, pause procurement for 60 days and appoint an operational owner. Buying during that pause merely makes the inconsistency harder to inspect.
We predict that below 90% identity coverage, specialist software will not produce two defensible budget changes within 60 days; traceable deal-level evidence producing those changes would prove us wrong.
Clean inputs are the admission price for attribution software.
Revenue attribution software by company size: the four-tier map
These four tiers use illustrative UK annual planning ranges, not overseas list prices or vendor quotes. The ranges cover incremental attribution tooling and implementation, excluding existing CRM and media spend.
| Company revenue | Sensible category and examples to investigate | What the tier buys | Illustrative annual range |
|---|---|---|---|
| £1m–£3m | Native CRM reporting with GA4 and Google Ads; existing HubSpot or Salesforce views | One revenue ledger, basic source cohorts and campaign-to-opportunity comparisons | £0–£6,000 |
| £3m–£10m | Lead and call attribution connectors such as Ruler Analytics, WhatConverts or CallRail | Better joining of forms, calls, campaigns and subsequent deal outcomes | £6,000–£18,000 |
| £10m–£25m | B2B journey platforms such as Dreamdata, HockeyStack or Factors.ai | Cross-channel journey evidence, model comparisons and buying-group context | £18,000–£45,000 |
| £25m–£50m | A governed warehouse stack or enterprise journey suite | Custom commercial logic, controlled exports and reporting across several business units | £45,000–£100,000 |
The significant cliff sits between lead attribution and B2B journey software, usually around the second and third tiers. Cross it only when the company has three or more active acquisition motions, at least 60 closed outcomes per motion annually and an owner available one day each week.
Missing any one of those thresholds means staying with the lower tier. More dashboards cannot compensate for an absent operator.
The honest limit here is that revenue is confounded by commercial complexity. A £4m enterprise seller with an 18-month buying cycle may need stronger infrastructure than a £20m transactional firm.
Keep attribution procurement separate from paid-search strategy and account management. Attribution explains observed commercial paths; it does not create demand or improve an advert.
Operational complexity, not turnover alone, determines the correct tier.
Build versus buy: the cost is labour, not the licence
A custom model looks cheap when engineering time is treated as free.
Take an illustrative UK B2B software firm comparing a warehouse build with a specialist platform.
Build inputs: analytics engineer, 180 setup hours at £70 per hour; RevOps, 120 definition and QA hours at £55 per hour; infrastructure, £3,600 annually; maintenance, eight engineering hours monthly at £70 per hour.
Build year one = (180 × £70) + (120 × £55) + £3,600 + (8 × 12 × £70) = £29,520
Buy inputs: licence and standard connectors, £24,000 annually; implementation, 60 internal hours at £60 per hour; administration, six hours monthly at £55 per hour.
Buy year one = £24,000 + (60 × £60) + (6 × 12 × £55) = £31,560
The first-year comparison is therefore £29,520 to build versus £31,560 to buy.
Internal time = 396 build hours versus 132 buy hours
Assuming unchanged requirements and pricing:
Build over three years = £29,520 + (£10,320 × 2) = £50,160
Buy over three years = £31,560 + (£27,960 × 2) = £87,480
The illustrative build saves £37,320 over three years but consumes 264 more internal hours during year one. What this calculation cannot tell you is the cost of a connector breaking, an engineer leaving or management demanding a rewritten model.
Build only when all three conditions hold:
- A maintained warehouse and stable cross-system identifiers already exist.
- At least 400 internal hours are reserved for year one and 100 hours for each following year.
- Two or more essential commercial decisions cannot be answered by shortlisted products during a pilot.
If readiness passes but any build condition fails, buy the lowest viable tier. If readiness fails, choose neither.
Available technical ownership decides build versus buy.
If the path from ad click to signed revenue still has gaps, Actualyse will trace each hand-off with you — book a call
What buying revenue attribution software does not fix
A missing sales update remains missing after the dashboard turns purple.
Four specific failures survive implementation:
- Unrecorded commercial events: private calls, partner introductions and late deal updates cannot receive reliable credit when no usable event exists. The software can only redistribute credit among observed touches.
- Consent and identity gaps: Google Consent Mode changes how tags behave; it does not grant permission to reconstruct an individual journey. GDPR and PECR obligations remain regardless of the licence fee.
- Weak website architecture: attribution may expose different journeys but cannot design them. Lanteria needed routes for distinct stakeholders, AfriCap Hub needed catalogue-to-registration flow, Savgen needed a technical multi-industry structure, and Lake Erie needed separate stay and ownership paths. Those decisions are visible across our multi-audience website and brand projects, without fabricated performance claims.
- A poor optimisation target: feeding every form submission back to Google Ads encourages the system to find more form submissions, not necessarily profitable customers. Software cannot replace sound campaign choices.
One procurement signal deserves blunt wording: if the pricing or the methodology is vague, it is not vague in your favour. A vendor is selling complexity when it cannot explain credit assignment, connector costs and post-cancellation exports in plain English.
Attribution software exposes operational truth; it cannot manufacture it.
A five-test procurement process for attribution software
Bring one commercial decision to every demonstration. “Which campaigns should lose £10,000 next quarter, based on recognised gross profit?” is useful; “show us the dashboard” is not.
Set the commercial denominator before involving vendors through a contribution-aware ROAS calculation. Decide separately which B2B Google Ads conversions deserve optimisation status, because attribution and bidding should not inherit an accidental form event.
Apply these five procurement tests:
- Decision test: the vendor must answer one named budget question using your anonymised records, not prepared sample data.
- Coverage test: the pilot must display joined and unjoined deals separately. Coverage below 90% fires a data pause, not a discounted contract.
- Traceability test: select three closed deals and trace every credited touch back to a timestamped source record. Reject unexplained fractional credit.
- Operating test: an internal owner must reproduce one core view and export its supporting data within two hours. Dependency on vendor support for routine questions predicts slow adoption.
- Commercial test: require one sterling year-one figure covering implementation, connectors, usage and support. A quote increasing by more than 10% after technical discovery triggers a new comparison.
Run the surviving product beside the existing method for 60 days. If it does not change at least two budget decisions using evidence the old stack lacked, refuse the annual agreement.
A pilot must change a decision before a contract changes hands.
FAQ
These five questions resolve choices that feature matrices usually obscure.
Which attribution model should we use?
Use first-touch and opportunity-creation touch as separate views until one sales motion has 100 closed outcomes. Above 100, test one multi-touch model against both baselines and retain it only if it changes a named decision.
Can GA4 be the revenue system of record?
No. Choose CRM closed-won value as the commercial ledger and GA4 as behavioural evidence. If monthly revenue differs by more than 5% between them, label the systems separately and stop publishing a blended total.
How long should the attribution window be?
Choose the longer of 90 days or 1.5 times the median lead-to-close period, capped at 365 days. For an illustrative 120-day median, 1.5 × 120 = 180 days, making a 7-day view less suitable than a 180-day view.
Does real-time attribution matter?
Use daily refreshes only when monthly media spend exceeds £25,000 and a named operator can move budget within 24 hours. Otherwise choose weekly reporting; real-time ROAS tracking earns its cost only when someone can act at the same speed.
Should an agency own the attribution account?
The company should own the contract, raw export and administrator access, with at least two internal administrators. An agency may operate the platform, but vendor-only or agency-only export rights should disqualify it.
Attribution choices should follow commercial decisions, not fashionable models.
Summary
The five operating rules are:
- Fail one readiness condition → pause procurement for 60 days.
- Three acquisition motions plus 60 outcomes each → shortlist specialist software.
- Fewer than 400 available build hours → buy the lowest viable tier.
- Vague pricing or methodology → reject the vendor before technical discovery.
- Fewer than two changed decisions in 60 days → refuse the annual contract.
Discipline earns the software; software never creates the discipline.
Actualyse builds measurement and attribution setups that tie B2B ad spend to real revenue. Book a call to talk through where yours stands.
