Paid traffic conversion funnel guide

Conversion Funnel: Diagnose Friction From Visit to Value

By FroggyAds, the self-serve advertising platform. Updated: 2026-10-23.

A conversion funnel follows a defined customer route from eligible visit to accepted value. Use it to find where eligible visitors stop, check whether that loss is real, and choose the next repair before buying more traffic with FroggyAds.

conversion funnel

What should you check when traffic does not convert?

Quick answer: Check the first meaningful transition that fails, then verify the underlying events and customer records. A missing checkout event needs a different response from a rejected payment or an unqualified enquiry. With FroggyAds, you can control the traffic entering that test through supported targeting, bids, budgets and source settings. Keep the destination and outcome definition clear enough to explain the result.

Funnel configuration reference: Google Analytics: Funnel Exploration explains ordered steps and open versus closed entry. Those settings affect which visitors your analysis includes.

Five checks before increasing traffic

Define the confirmed customer outcome and the eligible entry group before calculating a funnel rate.
Check that each stage represents a real action. A page view or button click does not prove the next action succeeded.
Keep people, sessions, forms and orders distinct. Compare the same counting unit at successive steps.
Investigate missing events, delays and alternate routes before calling every absent step an abandonment.
Choose the next campaign or destination change from confirmed outcomes and operating capacity, not the tallest bar on a chart.

What does a conversion funnel measure?

A conversion funnel measures how a defined group moves through selected steps toward a business outcome. Its rules determine which entries and completions count.

For media buyers running lead campaigns, label a click to start the form separately from the server's valid submission and the sales team's accepted enquiry. A loss between those stages can belong to the website, qualification rules or response team. Keep campaign and destination context with the records so the next FroggyAds buying decision addresses the transition that actually changed.

Choose the entry rule deliberately. A closed funnel follows people who enter at the first step. An open analysis can include later entrants. Neither is universally better: the right choice depends on the buying question. For a paid landing-page test, a fixed entry group can make the result easier to interpret. Keep returning visitors and alternate routes visible rather than forcing every customer through one imagined path. In GA4 Funnel Exploration, a user is counted only once in the first qualifying sequence for the selected date range. Keep that user-based funnel separate from an order ledger in which the same customer can place several genuine orders.

Why diagnose the funnel before buying more traffic?

Buying more visits amplifies the route you already have. If a form rejects valid entries, additional traffic brings more people to that failure. If the route works but attracts ineligible buyers, the next change may belong in the offer or targeting. A funnel helps distinguish those situations before you spend on the wrong remedy.

Diagnose the funnel before buying more traffic because the largest percentage loss may not be the most valuable repair. An early stage can lose visitors who were never eligible, while a smaller later break can block customers your sales team would accept. Compare affected volume, commercial value and repair effort before asking FroggyAds to send the next controlled visitor group.

We recommend using the funnel to choose one concrete question for your next FroggyAds campaign. That might be whether a repaired mobile route now produces confirmed orders, or whether a clearer eligibility message reduces rejected applications. Keep the question tied to a business action so an interesting chart does not become a substitute for a decision.

Eight parts of a funnel you can trust

#	Component	What to establish
1	Entry group	Define who enters the analysis and when. For a campaign test, retain the eligible landing visitors associated with the chosen offer and period. Separate test traffic and internal checks from customer evidence.
2	Customer question	State what the next transition tells you. An application start suggests interest; a sales-accepted application establishes a different commercial milestone. Use stage names that preserve that distinction.
3	Message and destination	Record the advertisement and page version that set the expectation. If the campaign promises a quote, the page should explain the quote process rather than quietly demand an unrelated purchase.
4	Event confirmation	Define the successful action behind each event. Distinguish clicking submit, server acceptance and a confirmation page. Test duplicates, reloads and rejected attempts before trusting the stage total.
5	Handoff owner	Name the person or system responsible after a lead or order arrives. A completed form can still fail commercially when nobody follows up or fulfillment cannot serve the customer.
6	Window and counting unit	Choose people, sessions, applications or orders consistently. Give each entry enough time to progress, and state how repeated attempts or later visits are handled.
7	Recovery route	Describe what a customer can do after an error or a pause. Check whether the route permits correction, a return visit or an appropriate support option without counting that recovery as a new acquisition.
8	Outcome quality	Confirm the result in the business system that owns it. Remove cancelled or rejected outcomes according to the declared rule, and keep pending cases visible until they are resolved.

Build the eight parts of a funnel you can trust as working definitions, not decorative labels. Analytics, sales and operations should agree on what each stage means and which evidence confirms the final outcome. Resolve that agreement before comparing FroggyAds sources, so campaign optimisation rewards a result your business actually values.

Traffic for a clear route

Send your next visitors into a route you can explain

Choose FroggyAds for controlled visitor acquisition while you improve the customer route. Set a clear market and format, cap the test spend and keep source identifiers available for review. We give you the traffic controls; your stage definitions and confirmed outcomes show which visitors reached the result you intended.

Create My Free Account

Illustration of audience targeting controls for a conversion funnel test

Build and diagnose the paid visitor route

1. Define the confirmed outcome

Start at the business end of the route. For a retailer, that might be a confirmed order under a stated cancellation rule. For a service, it might be an enquiry accepted by sales. Describe the condition precisely enough that the team can distinguish an accepted result from a recorded event.

Define the confirmed outcome with an authoritative evidence source. Your order system or sales record may contain acceptance information that a browser event cannot supply, so keep pending and rejected records visible until their status is known. Freeze the outcome rule during a FroggyAds traffic comparison instead of changing it when one source looks stronger under a new definition.

2. Trace the necessary transitions

Trace the necessary transitions backward from the confirmed result to the eligible landing visit. Include only actions required to complete the route and keep optional browsing outside the sequence. If customers can purchase directly from a product page, do not require a homepage visit merely to make the funnel chart look orderly.

3. Choose the entry group and clock

Record the entry period and the allowed time to complete the route. Review older and newer arrivals separately when their opportunity to convert differs. A customer who arrived yesterday should not be treated as a confirmed loss when the normal sales review has not happened yet.

4. Check the promise at every handoff

Read the advertisement, destination and final action together. Verify that the price, eligibility and next step remain consistent. A misleading promise may produce clicks while sending visitors into a process they did not expect. Fix that mismatch before changing the bid or blaming the source.

5. Test event success and failure

Complete a controlled test through the route, including an invalid entry and a repeated confirmation where practical. Confirm which actions create events and which create business records. Record the test identifiers and exclude them from customer totals. A tracking repair should have its own change date.

6. Read counts before comparing rates

Inspect how many entries reach each step and how many remain unresolved. Check the denominator beside every percentage. A high rate from a handful of people deserves a different decision from a stable pattern across a useful test. Keep the actual counts available to the media buyer.

7. Investigate one suspected break

Investigate one suspected break by selecting a transition with a credible problem and a useful commercial consequence. Reproduce that route in the affected device or browser context, then compare its events with completed business records. Write one repair hypothesis before redesigning the page, so the next FroggyAds test measures a defined correction.

8. Compare after the outcome matures

Review the repaired route after the agreed follow-up period. Match event totals with accepted business outcomes and identify differences that remain unexplained. Keep pre-repair and post-repair arrivals separate. A larger event count is not enough if rejected applications or cancellations rose at the same time.

9. Repair the operating handoff

If the loss happens after a valid form or order, involve the team that owns the next action. Confirm response capacity, eligibility checks and any promised service. More media will not repair an unattended enquiry queue. Once the handoff works, resume the appropriate acquisition test.

10. Decide what the next spend should test

Increase traffic when the verified route can handle it and the resulting outcomes support the cost. Keep the next change small enough to interpret. If the evidence is still uncertain, narrow the question or extend observation within the agreed budget rather than treating uncertainty as success.

Read the stage count before the percentage

Use confirmed outcomes from eligible entries as the commercial anchor. Step completion helps locate the issue; it does not replace that final outcome. Keep the same cohort, counting unit and completion window across the calculation so each percentage describes a real transition.

Measure	Calculation and interpretation	When it helps
Overall completion	Confirmed completers divided by eligible entrants in the same cohort. State the unit and window. This describes the whole measured route, not the causal value of the advertisement.	Comparing complete routes after follow-up
Step completion	People who reach the next required step divided by those who reached the preceding step, under the same sequence rule. A later entrant in an open funnel may require a different interpretation.	Locating a weak transition
Step drop-off	For a closed, consistent group, subtract next-step completers from prior-step entrants and divide by prior-step entrants. Keep pending cases separate when they still have time to complete.	After the step window has closed
Time to progress	Elapsed time between the defined steps for records that complete them. Report what happens to incomplete cases separately; a fast average among completers can hide many people who never finish.	Finding delays and choosing review windows
Failed attempts	Recorded unsuccessful attempts divided by relevant attempts under a stated error rule. Inspect the failure itself before calling it customer disinterest. Repeated attempts may belong to one person.	Checking forms, payments and recovery
Acquisition cost	Attributed media spend divided by confirmed customer outcomes. Add fulfillment or service costs separately when judging profitability. A zero-outcome test has spend and no accepted acquisitions, not a zero acquisition cost.	Choosing the next media allocation

Read the stage count before the percentage and reconcile differences in clock, counting unit, attribution rule or completion window. Two reports can disagree while answering different questions, so document those rules before declaring the conversion funnel broken. Your next traffic decision needs a consistent interpretation of the counts, not cosmetic agreement between dashboards.

The next customer step

Match the first visit to the next customer step

We offer self-serve Push, Native, Display and Pop campaigns. Choose a format whose message and destination fit the next action. A short invitation to learn about an offer can lead to an explanatory page; a specific purchase promise needs matching product details. Keep formats separate when their entry context differs, so a blended funnel does not hide a weak route.

Create My Free Account

Illustration comparing advertising formats for conversion funnel execution

Three customer routes with different stopping points

Lead generation: submitted is not accepted

A lead funnel can run from landing visit to form start, valid submission, sales review and accepted opportunity. If forms increase but sales acceptance falls, inspect the rejected reasons and the promise in the ad. You may need clearer eligibility or better follow-up. Reducing form fields without that diagnosis could create more work and fewer useful enquiries.

Ecommerce: checkout is not a confirmed order

Track the difference between viewing a product, starting checkout and receiving an accepted order. Check payment failure, unavailable delivery and inventory restrictions where they apply. A visitor can return later or use another route. Keep that behavior visible before assuming a missing same-session purchase means the customer rejected the offer.

Subscription: signup is not activation

A subscription funnel may include registration, successful setup, first useful activity and paid status. Choose the milestones that actually matter to your product. If new accounts never complete setup, investigate onboarding before buying more signups. FroggyAds can supply the acquisition traffic; your product and billing records establish activation and paid outcomes.

Five reasons a funnel can point you the wrong way

Mixing counting units

Avoid mixing counting units inside a conversion rate. One person can create several visits, form attempts or orders, so dividing unique customers by an event total does not produce a customer completion rate. Choose a consistent unit for the route and label any deliberate exception before using the result to change FroggyAds spend.

Treating missing measurement as lost interest

A failed event request, blocked browser measurement or changed event name can remove a stage from the chart. Compare the underlying records and run a controlled check. If real orders exist without the expected preceding event, repair the measurement and mark the affected period before evaluating the source.

Requiring a path customers do not need

Define the funnel as a selected sequence that helps answer a question about the customer journey. People may skip an optional page, revisit an offer or return through another channel. Use path analysis or your available route evidence to understand those alternatives. Keep necessary business steps separate from the sequence you originally expected.

Optimizing the easiest event

Optimizing the easiest event can make a campaign report look stronger while sales receives weaker enquiries. Keep button clicks and form starts as diagnostic stages, then judge traffic against the confirmed downstream outcome. We recommend increasing a FroggyAds budget only when the full route shows that the additional visitors support the business result you selected.

Changing too much after a small fluctuation

A short-lived drop can reflect a small group, delayed decisions or a changed source mix. Check those explanations before rebuilding the destination. When you do test a repair, record the exact change and preserve a comparison. Several simultaneous changes make a positive result harder to explain and repeat.

Resist changing too much after a small fluctuation. Use the funnel to direct investigation, and say plainly when the evidence shows only a missing stage rather than its cause. That discipline lets the destination team inspect the route while your media buyer protects the FroggyAds budget and preserves a clean comparison.

How should you budget for a funnel repair?

Budget for the question you need to answer. A suspected form defect may first need implementation time and a controlled test, rather than a new media allocation. An unclear offer may need revised creative and a destination comparison. Name that work before assigning a traffic budget, and keep its cost visible alongside the media spend.

Set a limit on the spend you can afford while learning. Then choose a follow-up period suited to the customer action. There is no universal budget that makes every funnel conclusive. Frequent low-value purchases and infrequent qualified enquiries produce different evidence. If the available spend cannot answer the question, reduce the number of variants or choose a narrower transition.

Check the capacity after the measured action. More valid enquiries can still overwhelm the response team. More orders can expose fulfillment constraints. Build those limits into the launch decision and agree who can pause the campaign. FroggyAds gives you adjustable campaign spending; a useful budget also reflects what your business can handle after the visit.

Use FroggyAds to control the traffic entering the test

Our self-serve campaign controls let you define supported targeting, bids and budgets while you investigate the customer route. Keep the campaign context available in your measurement so you can compare relevant sources and destinations. A clearly defined traffic test is easier to interpret than a mixture of changing offers and unrelated audiences.

Control the traffic entering the test by keeping FroggyAds source IDs with each accepted outcome. After a fair review window, use source whitelists or blacklists when the evidence supports a focused change. Confirm the conversion route first and keep rejected or pending outcomes separate; our acquisition controls shape delivery, while your business records establish value beyond the click.

Before traffic enters the test, verify that the FroggyAds campaign and source identifiers survive the intended route and that a controlled conversion reaches the correct record. Use the tracking method supported by your setup. Do not assume an analytics tool, CRM or order system sends every funnel stage automatically; validate every connection you intend to use for a budget decision.

A focused acquisition test

Turn the diagnosis into a focused acquisition test

Choose the traffic context that can answer the repair question. A mobile issue needs a working mobile destination and a relevant device test. A source-quality concern needs comparable offers and acceptance rules. Keep spending under your control while the evidence matures, then use the confirmed result to choose what to retain or change.

Create My Free Account

Illustration of a campaign launch checklist for conversion funnel

Move from an unreliable chart to a useful decision

First: make the route observable

Agree on entry, required steps and the final accepted action. Test those events against actual records. Fix missing or duplicate measurement before comparing performance. Preserve an example of a successful route and a failed attempt so another colleague can check what each event means.

Next: test the suspected break

Test the suspected break next by making one justified correction to the affected transition. Keep the previous destination version and exact release time, then buy a controlled amount of relevant FroggyAds traffic when paid visitors are needed for the comparison. Review stage counts, accepted outcomes and unresolved cases before interpreting the new rates.

Then: establish the operating decision

Establish the operating decision after the follow-up window closes. Record whether the route supports more acquisition, which confirmed outcome supports that choice and what uncertainty remains. If the repair helped under the measured conditions, increase one major FroggyAds input at a time; otherwise preserve the evidence and investigate the next credible explanation.

Questions for your analytics or implementation partner

Can we state the funnel entry rule, sequence and counting unit without relying on an unexplained dashboard default?
Which records expose missing, duplicate and rejected events alongside accepted records?
Does the report distinguish an immature entry group from a completed observation period?
Can we separate page versions and traffic contexts when a repair changes the route?
How will the relevant business team confirm the final outcome using its own records?
Would another analyst be able to reproduce the numbers from the documented definitions?

Choose a tool that answers the diagnosis you need. A funnel report helps evaluate a defined sequence; a path report helps discover routes you did not specify. Neither replaces a working destination or a useful offer. Verify data access and implementation requirements before buying a tool for a result that simpler records could already explain.

What should the visitor understand at each step?

Each step should answer the next customer question. A product page explains the offer; a form explains what information is required; a confirmation explains what happens next. Keep the language consistent with the ad. Visitors should not have to infer an unexpected fee or an unmentioned obligation halfway through the route.

At each funnel step, the visitor should understand the promise, eligibility requirement, action and confirmation that follows. Make destination errors specific and recoverable, preserving appropriate progress where possible, then test the actual route with the people who own it. Clear guidance supports a better paid-traffic experience, but changing one label cannot guarantee more conversions.

Review the route after a material offer or service change. A funnel that still records events can become commercially misleading if the next step no longer matches the promise. Keep customer-facing terms, measurement definitions and operating follow-up aligned so the media buyer can trust what the final event represents.

Frequently asked questions

What is a conversion funnel?

A conversion funnel measures progress through a defined sequence toward an accepted outcome. It helps locate where to investigate. Start by naming the eligible entry group, the necessary steps and the confirmation source for the final result. A chart alone does not explain why someone left.

How many steps should a conversion funnel contain?

Use the steps needed to answer the buying question. A landing visit, valid application and sales acceptance may be enough for one service. A retailer may need product, checkout and order confirmation stages. Avoid making optional browsing a required step merely to fill out the diagram.

Which event should be the final conversion?

Select a confirmed action that the business values, such as an accepted order or a sales-approved enquiry. Early events can diagnose progress but should not replace that final standard. Record the evidence source and how pending, rejected or cancelled outcomes are treated.

How should funnel results be compared across audiences?

Keep each comparison tied to a useful question, such as whether the mobile route fails where the desktop route works. Use consistent entry and outcome rules. Do not split every possible device, source and creative combination if the resulting groups are too small to support a decision.

How do you investigate a weak transition?

Check the offer and destination before interpreting the loss. Compare the promise, eligibility and next action at the suspected transition. Then test the route and underlying events. A departure may reflect a real obstacle, a different customer route or a measurement gap; investigate before choosing a repair.

Does funnel completion prove advertising caused the sale?

A funnel describes progression under its rules; attribution assigns credit under a separate model. A completed route does not prove the ad caused the outcome. Keep campaign context for buying decisions and use a suitable experiment before making an incremental or causal claim.

Why can high event volume be misleading?

Large event totals can include duplicate actions, test traffic, returning visitors or weak enquiries. Inspect the counting unit and confirmed outcomes before increasing spend. If there are no accepted customers, report that fact alongside cost rather than treating a busy chart as commercial success.

How should sales qualification appear in a lead funnel?

Define sales acceptance with the team handling enquiries. Keep the reason a form was accepted, rejected or left pending. If the weak transition happens after submission, investigate eligibility and response capacity before changing the landing page or buying additional visits.

How do you evaluate a funnel repair?

Test the correction, record its release time and separate later entrants from the earlier group. Give both groups a comparable opportunity to complete. Review confirmed outcomes and customer quality alongside the repaired event. Treat remaining differences in traffic mix or offer conditions as limits on the conclusion.

When should you increase the traffic budget?

Increase the next test when the route is working, accepted outcomes support the cost and the business can handle additional customers. Keep source mix and destination versions visible. Scale one important input at a time so a new problem can be traced and spending can be adjusted.

Selected funnel and implementation references

The Google references explain their own analytics and advertising systems. Use them to understand measurement concepts and the settings you actually use. They do not establish an automatic integration with FroggyAds. The remaining references support responsible advertising, helpful content and accessible implementation; this guide does not claim a compliance certification.

Google Analytics: Funnel Exploration
Discover unplanned routes through Google Analytics: Path Exploration
Review lead-stage reporting with Google Analytics: Lead Generation Funnel Reporting
Check configured action definitions in Google Ads: About Conversion Measurement
Prepare the relevant website implementation with Google Ads: Set Up Web Conversions
Federal Trade Commission: Advertising and Marketing Basics
Google Search Central: Helpful, Reliable, People-First Content
W3C: Web Content Accessibility Guidelines 2.2

Write a repair brief for one weak transition

Use this sequence when the funnel suggests a problem. It turns a suspicious stage into a bounded investigation and a specific next campaign decision. Complete the brief below for that transition, keeping the actual evidence separate from the proposed explanation.

Verify the suspected loss with actual event and business records; separate missing measurement from an unsuccessful customer action.
Identify the affected entry group and destination version, then reproduce the route in that context.
Write one repair hypothesis and confirm who owns the destination or operating handoff.
Test the correction and compare later entrants after a consistent follow-up period, keeping the spending boundary in place.
Use the confirmed outcomes to choose whether to retain the repair, investigate another cause or increase the next traffic test.

Which transition needs investigation?

Name the transition that needs investigation in plain language. Distinguish an application reaching the server from an application being accepted by sales, then state what evidence is missing and how you know. Describe the affected step instead of calling the entire conversion funnel broken, so the responsible team can test a precise repair.

Keep a few appropriate test records and the relevant event definitions with the brief. Another team member should be able to inspect the same route and reach the same description of the problem. Use authorized records without copying unnecessary customer information into a general campaign document.

Which visitors are affected?

Identify which visitors are affected by recording the FroggyAds campaign, entry period, destination version and relevant device or browser context. Separate unknown information from confirmed exclusions. When the same destination problem appears across several sources, investigate that shared route before blocking each source individually.

What expectation reaches the page?

Document what expectation reaches the page by pairing the advertisement's promise with the action required on the next screen. A mismatch can explain why paid traffic arrives but does not progress. Keep offer terms complete and honest; removing a material qualification may lift an intermediate rate while creating weaker leads or cancelled orders later in the funnel.

How do you confirm the step?

Confirm the step with the exact successful action, its event and the authoritative business record. Include what happens after a failed attempt, duplicate submission or reload. If an event fires before customer acceptance, give it a diagnostic label so your FroggyAds buyer cannot mistake early activity for a confirmed commercial outcome.

Who owns the response?

Name the team that handles the next action and the actual response process. Check what happens outside staffed hours or when demand exceeds capacity. Use that operating reality in the campaign plan; do not promise a response time that the business cannot deliver.

When is the comparison ready?

The comparison is ready when the entry window and allowed follow-up time are complete for the groups being judged. List pending cases, compare cohorts at similar ages and state how unresolved outcomes affect the decision. Keep an early directional view separate from the mature commercial review before changing FroggyAds targeting or budget.

What can the customer do after failure?

Check whether the route explains the problem and offers an appropriate next action. A valid recovery might be correcting a field or returning to a saved application. Measure recovered outcomes consistently so the same customer is not counted as several successful acquisitions.

What will change in the account?

State what will change in the account as the smallest action supported by the funnel finding: hold a source, repair the destination, clarify eligibility or run another defined comparison. Add the reason and spending limit. When practical, keep other FroggyAds inputs stable long enough to interpret the repair while protecting the business from a known broken route.

Buy the next visitors with a clear purpose

Create your FroggyAds account when you are ready to test a defined customer route. Select the supported format and targeting, set your bid and budget, and verify the conversion path before expanding. We give you practical acquisition control so the next spend can answer a useful business question.

Create My Free Account

From diagnosis to spend

What should the next traffic budget depend on?

Use our Top Of Funnel Marketing guide when your main job is building qualified awareness. This guide addresses the measured route after an eligible entry: which transition needs investigation and whether the resulting customers justify more traffic. The distinction keeps a broad reach objective from being judged by a checkout-only chart.

Let the next traffic budget depend on the complete route, accepted outcomes and operational capacity. A strong early response can still lead to poor commercial results, while a smaller source may remain valuable when eligible customers complete the funnel. Use FroggyAds source and campaign evidence to fund the next specific question instead of rewarding the largest intermediate count.

Checkpoint	Campaign action	Evidence to keep
Fit	Confirm the route is relevant and the accepted event is real.	Entry rule, outcome definition and successful test record.
Test	Buy enough relevant traffic to investigate one defined transition.	Campaign context, destination version, source identifiers and spending boundary.
Decision	Choose the next repair or allocation from confirmed outcomes.	Accepted and pending counts, acquisition cost and the reason for the change.

Hypothetical closed-cohort example: Suppose 1,000 eligible visitors include 200 checkout starters and 40 confirmed buyers. Assume each buyer completed checkout within the same review window. Entry-to-checkout is 20%, checkout completion is 20%, and overall completion is 4%. Checkout drop-off is (200 - 40) / 200 = 80%. At USD 600 media spend, cost per confirmed buyer is USD 15. These are invented teaching inputs, not FroggyAds results.

Choose FroggyAds to put a clear funnel decision into action. Buy relevant traffic with campaign-level spending and targeting controls, inspect source evidence and connect the result to your confirmed customer outcome. Start with the route you can explain and improve. Create your free FroggyAds account

Why the final acceptance step changes the cost

Hypothetical lead example: USD 240 buys traffic that produces 12 valid forms, of which sales accepts 6. Media cost is USD 20 per form, but USD 40 per accepted lead. Use the second figure for a sales-acceptance buying decision and retain the first as a diagnostic. The counts share one review period; no customer result, benchmark or guaranteed return is implied.

Before your next campaign

What should you do before buying more visits?

Before buying more visits, verify the route, confirm the business outcome and identify the transition worth improving. Then use FroggyAds to acquire the next controlled visitor group for that exact question, review the mature result and make one accountable campaign decision.